CAI4CAI

Establishing a Surgical AI Collaboration at OAUTHC, Nigeria

Sun, 02 Aug 2026 00:00:01 +0000

In recent weeks, members of the CAI4CAI team visited Obafemi Awolowo University Teaching Hospitals Complex (OAUTHC) in Ile-Ife, Nigeria, to strengthen our collaboration with local surgeons and lay the foundations for long-term surgical AI research.

The visit focused on building the infrastructure, relationships, and workflows needed to support the sustainable collection of high-quality laparoscopic surgical data for future research in computer-assisted interventions and artificial intelligence.

Building the foundations

One of the major milestones of the visit was the successful deployment of dedicated data infrastructure within the Department of Surgery.

A central network-attached storage (NAS) system, backup power supply, and local network were installed to enable surgeons to securely upload and manage surgical videos. The deployment required close collaboration with the local hospital team and involved setting up a dedicated workspace, networking equipment, and user accounts before training members of the surgical department on the new workflow.

The newly installed NAS system, backup power supply, and local networking equipment.

Working with the local team to install and configure the data infrastructure.

The infrastructure now provides a sustainable platform for ongoing data collection and future collaborative research.

Strengthening partnerships

Beyond the technical work, the visit provided valuable opportunities to strengthen relationships across OAUTHC.

We met with the Chief Medical Director of the Hospital, Professor John Okeniyi, and the Provost of the College of Health Sciences, Professor Bernice Adegbehingbe, who both expressed strong support for the collaboration and its long-term vision. The visit also included discussions with clinicians from multiple surgical specialties about opportunities for future research and clinical translation.

Meeting with OAUTHC Chief Medical Director Professor John Okeniyi.

Discussing the collaboration with Professor Bernice Adegbehingbe, Provost of the College of Health Sciences.

A presentation to the Department of Surgery introduced CAI4CAI’s research, the goals of the collaboration, and the potential of artificial intelligence to improve surgical care. The enthusiasm from surgeons and trainees highlighted the growing interest in developing locally relevant AI technologies through international collaboration.

Introducing CAI4CAI’s research and the goals of the collaboration to the Department of Surgery.

Building a new surgical dataset

Alongside establishing the infrastructure, we worked with the surgical team to organise an expanding collection of laparoscopic procedures.

The dataset currently contains dozens of carefully reviewed operations spanning multiple procedure types, including cholecystectomy, appendectomy, fundoplication, and colorectal surgery. Working closely with Professor Adewale Adisa and his team, we are also establishing clinical annotation protocols for surgical workflow, scene understanding, and disease severity grading. These annotations will create valuable resources for future AI development.

Reviewing laparoscopic surgical video with Professor Adewale Adisa.

Engaging medical students in surgical data and AI research.

To support patient privacy, a dedicated de-identification workflow has also been established, ensuring videos can be prepared for research while protecting patient confidentiality.

Looking ahead

With the infrastructure now in place, the collaboration enters an exciting new phase.

Future work will focus on expanding the dataset, developing high-quality surgical annotations, and supporting collaborative research between King’s College London and OAUTHC. By combining clinical expertise with advances in artificial intelligence, we hope to create broadly useful resources that contribute to the development of safer, more accessible, and more equitable surgical AI systems.

This visit represents an important milestone, but it is only the beginning. We are grateful to Professor Adewale Adisa, the surgical team at OAUTHC, and all our collaborators for their enthusiasm, hospitality, and commitment to building this partnership.

We look forward to the next stage of the collaboration and the research opportunities it will create.

Photo gallery

Members of CAI4CAI and the OAUTHC surgical team following the departmental presentation.

A teaching session with medical students at OAUTHC.

A collaborative session in the OAUTHC Surgical Skills Laboratory.

Oluwatosin Alabi answering questions following the departmental presentation.

The Robot Doctor Will See You Now | Live podcast at the Great Exhibition Road Festival | Saturday 06 June 2026

Sat, 09 May 2026 00:00:01 +0000

Robot Talk is returning to The Great Exhibition Road Festival in London this Saturday 06 June 2026 at 17:00 for another live recording of the podcast.

Claire Asher has been chatting to Dr. Antonia Tzemanaki, Prof. George Mylonas, and Prof. Tom Vercauteren about robotics and AI in medicine and healthcare.

Listen to the full podcast here or your preferred podcast platform.

Robot Talk is recording a live podcast entitled ‘The Robot Doctor Will See You Now’ in London, UK, on Saturday 06 June 2026 at 17:00.

Overview

Learn how robotics and artificial intelligence are transforming medicine, and what the future holds, at this live podcast recording.

Since the first robot-assisted surgery was performed, over 40 years ago, major advances in robotics, computer vision and artificial intelligence have fundamentally changed medicine and healthcare.

Innovative new technologies are already aiding skilled medical professionals in diagnosis, surgery, rehabilitation and beyond. But many questions remain: What ethical issues arise as medical tools become increasingly autonomous? How do we regulate technologies that can learn and change over time? And how can we ensure that cutting-edge medical devices are accessible to all?

Join experts in robotics and AI to explore these topics in a live recording of the Robot Talk podcast.

Now in its sixth season, Robot Talk is funded by the Hamlyn Centre and covers all aspects of robotics research and innovation. It aims to engage and inspire people to understand, interact with, and pursue careers in robotics, and to boost the profile of robotics research in the UK and globally.

This event is part of the Great Exhibition Road Festival on 6-7 June 2026.

Great Exhibition Road Festival

6-7 June 2026

South Kensington’s annual celebration of science and the arts returns this summer with a weekend of free events for all ages.

Enjoy hands-on workshops, fascinating talks, performances and installations from iconic museums, research and culture organisations in South Kensington, including Imperial College London, the Natural History Museum, Science Museum, V&A, the Royal Parks, the Royal Commission for the Exhibition of 1851 and many more!

Find out more about the Festival and see the full programme on the Festival website at www.greatexhibitionroadfestival.co.uk.

LM-SURG: Large Models Meet Surgical Data Science Workshop at ISBI 2026

Sun, 08 Mar 2026 00:00:01 +0000

We are excited to announce the LM-SURG workshop, titled “Large Models Meet Surgical Data Science,” which will be held at ISBI 2026 at Excel London organized by the CAI4CAI group.

Workshop Aims & Key Topics

We strive to create a forward-thinking environment focused on:

Methodologies for Training Large Models in Surgery and Interventional Science.
Deployment Challenges in Operating Rooms (OR) and Devices.
Ethical and Privacy Considerations.
Interdisciplinary Collaboration and Translational Research.
Emerging Applications and Future Directions.
Creation of Large-Scale Datasets for Surgical and Interventional Science Applications.

Invited Speakers

The workshop features keynote insights from leading experts:

Dr. Sharib Ali – University of Leeds
Dr. Patrick Godau – German Cancer Research Center (DKFZ)
Dr. Kyle Lam – Imperial College London
Prof. Miaojing Shi – Tongji University
Dr. Binod Bhattarai – University of Aberdeen

PhD opportunity [October 2026 start] on "Learning-based vision system for marker-free external ventricular drain (EVD) neuronavigation"

Sun, 08 Feb 2026 00:00:01 +0000

Applications are invited for a fully funded 1+3 years MRes+PhD or 4 years PhD EPSRC CDT PSI studentship (including tuition fees, annual stipend and consumables) starting in October 2026.

Award details:

Focus: Learning-based vision system for marker-free external ventricular drain (EVD) neuronavigation
First supervisor: Tom Vercauteren
Second supervisor: Jonathan Shapey
Funding type: 4-year fully-funded EPSRC CDT PSI studentship including a stipend, tuition fees, research training and support grant (RTSG), and a travel and conference allowance.
Start date: October 2026

A simulated navigated neurosurgery in a mock operating room. The project aims at replacing the complex marker-based neuronavigation system by a marker-free approach for external ventricular drain (EVD) placement.

Project description

This research aims to develop a marker-free, AI learning-based neuronavigation system for external ventricular drain (EVD) placement. By leveraging real-time detection and tracking, precise image registration, and intuitive visualisation, the system seeks to enhance surgical accuracy, reduce complications, and address the limitations of traditional and existing image-guided techniques. To fulfil this aim, the following objectives will be pursued: 1) develop a real-time tracking system using stereo cameras and bespoke AI algorithms to capture and analyse the surgical scene, achieving high-precision tracking of the patient’s head and surgical instruments; 2) construct an efficient 6D pose estimation method for registering preoperative and intraoperative data, providing accurate guidance for surgical procedures.

Application Process

More information about the opportunity here and here.

PhD opportunity [October 2026 start] on "Intelligent deep learning neuroimaging system for guiding brain tumour treatment"

Sat, 07 Feb 2026 00:00:01 +0000

Applications are invited for a fully funded 1+3 years MRes+PhD or 4 years PhD EPSRC CDT PSI studentship (including tuition fees, annual stipend and consumables) starting in October 2026.

Award details:

Focus: Intelligent deep learning neuroimaging system for guiding brain tumour treatment
First supervisor: Jonathan Shapey
Second supervisor: Tom Vercauteren
Funding type: 4-year fully-funded EPSRC CDT PSI studentship including a stipend, tuition fees, research training and support grant (RTSG), and a travel and conference allowance.
Start date: October 2026

Artificial Intelligence based tool to assist clinicians during multidisciplinary team meetings (MDTMs). A specialist team of clinicians meet to discuss the optimal timing and mode of treatment for patients with brain tumours. This project will develop and evaluate the use of state-of-the-art AI-assisted tools for vestibular schwannoma. The tool will detect and segment the tumours and analyse imaging biomarkers to predict tumour behaviour before and after treatment.

Project description

Vestibular Schwannoma (VS) is a non-cancerous brain tumour that grows from the inner ear, towards the brain. At current rates, approximately 1 in 1000 people will be diagnosed with a VS in their lifetime. Patients with VS require individualized patient management that may include imaging surveillance, radiation treatment or surgery.

This project aims to: 1) optimise deep learning models to automatically detect and segment VS using MRI; 2) integrate the framework into a tool capable of being deployed in the clinic; and 3) conduct a prospective clinical pilot study to evaluate the clinical impact of using AI-based tool in patient management.

Modern learning-based image-registration methods will be utilised to provide robustness and computational efficiency. The clinical pilot study will provide the foundation for a future multicentre interventional study aimed at assessing clinical effectiveness and health economic impact.

Application Process

More information about the opportunity here and here.

PhD opportunity [October 2026 start] on "Immersive visuo-haptic endovascular tele-operation through AI-enabled multimodal semantic telecommunication"

Fri, 06 Feb 2026 00:00:01 +0000

Applications are invited for a fully funded 1+3 years MRes+PhD or 4 years PhD EPSRC CDT PSI studentship (including tuition fees, annual stipend and consumables) starting in October 2026.

Award details:

Focus: Immersive visuo-haptic endovascular tele-operation through AI-enabled multimodal semantic telecommunication
First supervisor: Mohammad Shikh-Bahaei
Second supervisor: Thomas Booth
Third supervisor: Tom Vercauteren
Funding type: 4-year fully-funded EPSRC CDT PSI studentship including a stipend, tuition fees, research training and support grant (RTSG), and a travel and conference allowance.
Start date: October 2026

Schematic illustration of semantic telecommunication for teleoperated mechanical thrombectomy.

Project description

The aim of this project is to develop AI-driven communication and sensing technologies to allow doctors to perform a type of stroke treatment, called mechanical thrombectomy (MT), using a robot remotely controlled with the help of AI and advanced telecommunication and robotic technologies.

Application Process

More information about the opportunity here and here.

PhD opportunity [October 2026 start] on "Language-based agentic collaboration for endovascular acute stroke treatment"

Mon, 19 Jan 2026 00:00:01 +0000

Applications are invited for a fully funded 4 years PhD CDT STaR-AI studentship (including tuition fees, annual stipend and consumables) starting in October 2026.

Award details:

Focus: Language-based agentic collaboration for endovascular acute stroke treatment
Joint first supervisor: Tom Vercauteren
Joint first supervisor: Yali Du
Clinical champion: Thomas C Booth
Funding type: We are only able to consider candidates who qualify for home fee status. 4-year fully-funded CDT STaR-AI studentship including a stipend, tuition fees, research training and support grant (RTSG), and a travel and conference allowance.
Project code: STaR-AI-15
Application closing date: 02 March 2026
Start date: October 2026

Mechanical thrombectomy procedure on a patient with symptoms of stroke as observed from the control room.

Project description

Stroke is the second most common cause of death and the third most common cause of disability (Feigin, 2025). Mechanical Thrombectomy (MT) has become the first-line treatment of acute cerebral stroke. MT consists of inserting a flexible catheter from the groin to the brain and using it to physically remove the blood clot (a.k.a. thrombus) causing the stroke. MT is a complex procedure and the number of physicians who have the expertise to perform it, i.e. interventional neuro-radiologists (INRs), is too low. Due to lack of INRs, less than 20% of eligible patients are currently treated (Aguiar de Sousa, 2019). Autonomous robotics and remote teleoperation are seen as potential solutions to this crisis. Current research in this area focuses on automating specific steps of the procedure (e.g. endovascular navigation) (Robertshaw, 2023). Yet, INRs operate in a collaborative environment involving multiple human agents. The radiographer for example plays a critical role by optimising X-Ray fluoroscopic views to follow the instruments and capture anatomical context, by coordinating the injection of contrast agent, and by ensuring accurate real-time imaging to support the INR. Increasing the autonomy of MT therefore requires the development of multi-agent systems whereby AI agent can seamlessly collaborate with human staff.

In this project, the PhD candidate will focus on the development of a collaborative framework between an INR and a radiographer agent to enable fluent endovascular navigation while maintaining strict safety constraints. Paving the way for flexible human-AI team composition (Yan, 2023), the project will develop a decentralised multi-agent system exploiting natural language as the basis for communication between the agents. The theoretical foundation draws from constrained and risk-sensitive reinforcement learning (RL) for safety-critical control (Gu, 2024), multi-agent coordination under partial observability, and human factors engineering for interventional workflows. Empirically, we will leverage high-fidelity computer simulation and physical mock labs to progressively validate agent behaviours, with clinical input shaping the safety envelopes, and evaluation metrics.

In year 1, the PhD candidate will train on safe multi-agent RL, world models for robotics (Wu, 2023), human-AI communication, and real-time fluoroscopy image analysis by performing a scoped literature review. This learning will initially be leveraged to engage with clinical collaborators, patient groups, ethics experts, and the supervisory team to refine the proposed work plan through a co-creation exercise. The candidate will also expand a state-of-the-art computer simulation engine for MT (Karstensen, 2025) to account for the actions of a radiographer agent. Improvements in imaging physics realism and anatomical variability will be combined with the addition of novel communication channels and APIs to support natural language interaction between agents.

In year 2, the project will aim to formalise shared and role-specific goals (e.g., instrument tracking, target selection, contrast timing) under safety constraints (radiation dose, vessel injury, embolic risk). We will then develop training approaches and natural language protocols for agent-to-agent and human-AI communication, including intent clarification, and fluoroscopy based grounding of the decision making.

The resulting multi-agent system will be trained and benchmarked in computer simulation with anatomically realistic vasculature and imaging physics.

In year 3, the objective is to transition the system to our physical mock interventional suite to evaluate usability, teamwork quality, and task performance, with INR/radiographer user studies. On the methodological side, the candidate will seek to extend the latest safe RL research to multi-agent coordination with constraint satisfaction, uncertainty quantification, and risk-aware exploration.

By the end of the PhD, the candidate will produce a thesis supporting by high-quality publications showing advances in safety-aware, language-mediated multi-agent systems, and demonstrating a credible path towards scalable human-AI teams in safety-critical environments such as MT.

References

Aguiar de Sousa, D., von Martial, R., Abilleira, S., Gattringer, T., Kobayashi, A., Gallofré, M., … & Fischer, U. (2019). Access to and delivery of acute ischaemic stroke treatments: a survey of national scientific societies and stroke experts in 44 European countries. European stroke journal, 4(1), 13-28.
Feigin, V. L., Brainin, M., Norrving, B., Martins, S. O., Pandian, J., Lindsay, P., … & Rautalin, I. (2025). World stroke organization: global stroke fact sheet 2025. International Journal of Stroke, 20(2), 132-144.
Gu, S., Yang, L., Du, Y., Chen, G., Walter, F., Wang, J., & Knoll, A. (2024). A review of safe reinforcement learning: Methods, theories and applications. IEEE Transactions on Pattern Analysis and Machine Intelligence.
Karstensen, L., Robertshaw, H., Hatzl, J., Jackson, B., Langejürgen, J., Breininger, K., … & Mathis-Ullrich, F. (2025). Learning-based autonomous navigation, benchmark environments and simulation framework for endovascular interventions. Computers in biology and medicine, 196, 110844.
Robertshaw, H., Karstensen, L., Jackson, B., Sadati, H., Rhode, K., Ourselin, S., … & Booth, T. C. (2023). Artificial intelligence in the autonomous navigation of endovascular interventions: a systematic review. Frontiers in Human Neuroscience, 17, 1239374.
Wu, P., Escontrela, A., Hafner, D., Abbeel, P., & Goldberg, K. (2023, March). Daydreamer: World models for physical robot learning. In Conference on robot learning (pp. 2226-2240). PMLR.
Yan, X., Guo, J., Lou, X., Wang, J., Zhang, H., & Du, Y. (2023). An efficient end-to-end training approach for zero-shot human-AI coordination. Advances in neural information processing systems, 36, 2636-2658.

Application Process

More information about the opportunity here and here.

A candidate recruited to this project will be registered in the Department of Informatics at King’s College London; for entry requirements, see the Computer Science Research MPhil/PhD information.

For information on how to apply, see https://www.kcl.ac.uk/research/star-ai. Please take care to follow the instructions on how to apply, otherwise your application may not be considered.

About STaR-AI: King’s Prize Doctoral Programme in Safe, Trusted and Responsible Artificial Intelligence

The King’s Prize Doctoral Programme in Safe, Trusted and Responsible Artificial Intelligence (STaR AI) brings together leading researchers from across King’s College London to train the next generation of experts in responsible AI. The programme equips graduates to understand the technical challenges of building safe and trustworthy AI, to engage critically with its human and societal implications, and to work confidently across disciplines to ensure AI technologies have positive impact.

King’s longstanding strength in interdisciplinarity provides a distinctive environment for studying AI and its wider consequences. Building on the success of the UKRI Centre for Doctoral Training in Safe and Trusted AI, STaR AI is supported by specialists in AI methods, human centred approaches, and legal and ethical frameworks from the Departments of Informatics and Digital Humanities, and the Dickson Poon School of Law. Students will gain both technical and non technical expertise relevant to responsible AI development across sectors, and will be well prepared for diverse careers, including in academia, research and development, and policy.

The programme welcomes applicants from a wide range of disciplinary backgrounds. Multidisciplinary supervision teams support students working on diverse application areas, enabling cohorts that combine technical, social scientific and humanities perspectives. This diversity is central to developing well rounded researchers able to meet the demands of a rapidly evolving national and international AI landscape.

A broad portfolio of PhD projects is available. Some focus on advancing core technical methods, such as new verification techniques for certifying AI system safety. Others examine human, social and legal dimensions, including the impact of AI on work and labour. Several projects span the socio technical space, for example developing new approaches to human–AI collaboration.

At least four fully funded studentships for home fee candidates are available for entry in October 2026. STaR AI students join a collaborative research community and take part in regular cohort building and training activities.

King’s Global Engagement Partnership Fund award for new surgical AI collaboration in Nigeria

Wed, 17 Dec 2025 00:00:01 +0000

Congratulations to Tom Vercauteren and Oluwatosin Olatunde Alabi, who have been awarded a King’s College London Global Engagement Partnership Fund grant for a new collaboration with Adewale Adisa, Professor of Surgery at Obafemi Awolowo University Teaching Hospital Complex in Ile-Ife, Nigeria.

This project will establish a pilot open dataset of laparoscopic cholecystectomy videos from Nigeria, with a focus on surgical safety and scene understanding. The pilot will also lay the groundwork for a long-term joint research programme in surgical AI in Sub-Saharan Africa.

The King’s Global Engagement Partnership Fund aims to further research and education collaboration with King’s international partners.

“This collaboration focuses on building a comprehensive understanding of surgical scenes from the beginning of an operation through to its completion. By developing richer, panoptic representations of real surgical workflows, we aim to support safer decision-making and surgeon training. Working closely with clinicians in Nigeria ensures that the work remains grounded in real practice and clinically meaningful.”
Oluwatosin Alabi

“This project provides an important opportunity to strengthen surgical training by combining established clinical practice with modern data-driven approaches. By contributing local expertise and real operative cases, we aim to reinforce safety-focused principles in both research and surgical education, while also showcasing high-quality surgical practice from West Africa within the global research community.”
Prof Adewale Adisa

PhD opportunity [June 2026 start] on "Anatomy localisation in X-ray fluoroscopy videos for mechanical thrombectomy" in collaboration with Telos Health

Wed, 03 Dec 2025 00:00:01 +0000

Applications are invited for a fully funded 4 years PhD EPSRC CDT PSI studentship (including tuition fees, annual stipend and consumables) starting in June 2026.

Award details:

Focus: Anatomy localisation in X-ray fluoroscopy videos for mechanical thrombectomy
First supervisor: Tom Vercauteren
Second supervisor: Thomas Booth
Industry supervisor: Konrad Leibrandt, Telos Health
Funding type: 4-year fully-funded EPSRC CDT PSI studentship including a stipend, tuition fees, research training and support grant (RTSG), and a travel and conference allowance.
Start date: June 2026

Schematic illustration of an interventional neuroradiologist performing a mechanical thrombectomy procedure with fluoroscopic images being automatically annotated with anatomical landmarks.

Project description

This project focuses on the development of a learning-based system capable of localising anatomy in X-Ray fluoroscopy video streams acquired during mechanical thrombectomy (MT). Mechanical thrombectomy are emergency procedures for acute ischaemic stroke. An endovascular catheter is navigated from the groin up to the brain under real-time X-ray fluoroscopy guidance. MT is challenging to perform in part due to the complexity of fluoroscopy image interpretation. Computer vision approaches have shown promising capabilities in surgical scene understanding across several minimally invasive surgical specialties but their development in interventional neuroradiology remains at its infancy. The PhD candidate will develop novel methods to automatically identify anatomical structures in fluoroscopy videos with the ambition of providing a better spatial understanding to the interventional neuroradiologist (INR). This project will require close collaboration with expert INRs to build annotated fluoroscopy databases and validate the performance of the proposed solutions.

Application Process

More information about the opportunity here and here.

PhD opportunity [October 2026 start] on "Patient-specific CTA-informed in-silico simulation of interventional fluoroscopy: A digital twin for stroke patients"

Tue, 02 Dec 2025 00:00:01 +0000

Applications are invited for a fully funded 4 years PhD EPSRC CDT DRIVE-Health studentship (including tuition fees, annual stipend and consumables) starting in October 2026.

Award details:

Focus: Digital twin with interventional neuroradiology applications
Primary supervisor: Tom Vercauteren
Secondary supervisor: Thomas C Booth
Third Supervisor: Rachel Sparks
Funding type: 4-year fully-funded EPSRC CDT DRIVE-Health studentship including a stipend, tuition fees, research training and support grant (RTSG), and a travel and conference allowance.
Application closing date: 12 January 2026
Start date: October 2026

Patient-specific CTA-informed in-silico simulation of interventional fluoroscopy: A digital twin for stroke patients.

Aim of project

Mechanical Thrombectomy (MT) has become the first-line treatment for acute cerebral stroke. It is however a complex procedure that would benefit from assistive technology and improved training opportunities for clinician. This project will develop and validate a computational pipeline to create digital twins of stroke patients based on pre-operative computed tomography angiography (CTA). By focusing on the synthesis of realistic fluoroscopy images with the digital twin being used to model a mechanical thrombectomy procedure, the student will provide a foundational platform on which better simulators can be created for training of clinicians and AI-agents alike. Direct interaction with interventional neuroradiologists and clinical trainees will guide the validation of the proposed methods and allow for user feedback to feed an iterative refinement of the digital twin methodology.

Project description

Stroke is the second leading cause of death (11% of total deaths) and the third leading cause of disability in the world. In the UK alone, the incidence of stroke is approximately 100 000 cases per year, or one stroke every five minutes. Ischaemic stroke, the most common type of stroke, happens when an artery is blocked by a blood clot, partly cutting off cerebral blood flow. For about 10% of stroke patients, mostly those presenting at the hospital within the first few hours after an ischaemic stroke, mechanical thrombectomy (MT) has emerged as a life-changing procedure. Patients undergoing MT will normally first get a 3D cerebral angiography (CTA) to determine the location of the clot within the vasculature. The procedure then entails inserting a catheter from the groin up to the clot and using a clot retrieval device passed through the catheter. Intra-operative navigation of the catheter and associated instrumentation is performed under real-time 2D X-ray fluoroscopy with intermittent local injection of contrast agent to improve visibility of the vasculature and its restriction by the clot.

MT is an effective but complex procedure whose adoption is currently limited by the availability of trained interventional neuroradiologists. There is a crucial need for improved clinical training programmes and assistive AI-defined interventional technology, both objectives requiring the advance of patient-specific models. This PhD project aims at developing a pipeline to create digital twins of ischaemic stroke patients. The resulting digital twins will enable realistic simulation of interventional fluoroscopy video feeds with simulated intravascular instruments and contrast injection.

Starting from the 3D CTA, a 3D model of the vasculature and location of the clot will first be created based on deep learning approaches. Emphasis will be put on respecting the expected topological characteristics of cerebral vasculature in order to support its follow-up use in the digital twin. The main focus of the PhD will then be put on the synthesis of fluoroscopy video feeds. Deep-learning based rendering approaches will be developed to generate convincing X-ray projections from the CTA. Building on existing MT simulators focusing on the mechanical interaction of catheters and the vasculature, the project will further design mechanisms to simulate fluoroscopy images with local contrast injection and anatomical deformations induced by the instruments. Emphasis will be put on achieving good temporal consistency in the simulation to avoid any flickering artefact that are commonly observed in single image-based synthesis. Close collaboration with interventional neuroradiologists and clinical trainees will be exploited to validate the proposed digital twins and incorporate user feedback in the iterative refinement of the solutions.

Expected academic background

The ideal candidate for this project will have a strong machine learning and computer vision background and a keen interest in interacting with clinicians. Experience developing or interfacing with gaming engines and/or physics simulators is not mandatory but would facilitate rapid progress.

Representative publications from supervisors

Wang, G., …, & Vercauteren, T. (2019). Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networks. Neurocomputing, 338, 34-45. https://doi.org/10.1016/j.neucom.2019.01.103
Sparks, R., … & Ourselin, S. (2017). Automated multiple trajectory planning algorithm for the placement of stereo-electroencephalography (SEEG) electrodes in epilepsy treatment. International journal of computer assisted radiology and surgery, 12, 123-136. https://doi.org/10.1007/s11548-016-1452-x
Robertshaw, …, & Booth, T. C. (2024). Autonomous navigation of catheters and guidewires in mechanical thrombectomy using inverse reinforcement learning. International Journal of Computer Assisted Radiology and Surgery, 1-10. https://doi.org/10.1007/s11548-024-03208-w

Application Process

More information about the opportunity here and here.

LBR-Stack: ROS 2 and Python Integration of KUKA FRI for Med and IIWA Robots

Sun, 23 Nov 2025 00:00:00 +0000

PhD opportunity [October 2026 start] on "Real-time intraoperative polarimetric imaging to detect nerves during brain tumour surgery"

Sun, 28 Sep 2025 00:00:01 +0000

Applications are invited for a fully funded 1+3 years MRes+PhD or 4 years PhD MRC DTP studentship (including tuition fees, annual stipend and consumables) starting in October 2026.

Award details:

Focus: Real-time intraoperative polarimetric imaging to detect nerves during brain tumour surgery
First supervisor: Jonathan Shapey
Second supervisor: Tom Vercauteren
Third supervisor: Graeme Stasiuk
Funding type: 4-year fully-funded MRC DTP studentship including a stipend, tuition fees, research training and support grant (RTSG), and a travel and conference allowance.
Application closing date: 28 October 2025
Start date: October 2026

Polarimetric imaging for better intra-operative visualisation of cranial nerves.

Project description

Vestibular schwannoma (VS) is the commonest pathology encountered in skull base neurosurgery. The goal of modern VS surgery is maximal tumour removal whilst preserving neurological function, but facial paralysis injury still occurs in 34% of patients because it is incredibly difficult for surgeons to visualise the facial nerve clearly during surgery.

Polarisation is a fundamental but under-exploited characteristic of light. Recent studies have led to new ways to identify nerves via their special polarization signature. Previous work has focussed on the characterisation of normal anatomy using devices that are challenging to integrate in the surgical workflow. This study aims to use a recently developed compact multi-polarisation-sensitive camera to achieve real-time intraoperative visualisation of nerve in the presence of tumour.

MRes 3-month research project: For the MRes project, students will compare the multi-polarisation-sensitive camera to a classical camera with a rotating polariser using standard polarised imaging scenes. Students will also attend the neurosurgical operating theatre to familiarise themselves with vestibular schwannoma pathology, surgery and the current limitations of standard microscopic visualisation.

Year 1: Characterisation of polarisation camera using phantom models Y1 will focus on setting up and characterising the imaging system based on the compact multi-polarisation-sensitive camera. A first pipeline for combining the multiple polarisation image responses will be designed. The student will research the physical and optical properties of nerve and tumour and develop realistic phantom models of these. The student will leverage the group’s extensive experience to fast-track developments.

Outcome: Imaging system for a detailed understanding of the optical polarisation of nerve and tumour

Year 2: Evaluation of optical polarisation of nerve and tumour in a murine model In Y2 the student will be trained in animal handling. Using an established murine model of sciatic nerve schwannoma, the student will fine-tune the combination algorithm and characterise performance in delineating nerve and tumour.

Outcome: Improved differentiation of nerve and schwannoma in vivo

Years 3 – 4: Development of a novel murine vestibular schwannoma model and evaluation of tissues using a polarisation camera In Y3–4, the student will collaborate with experts at the University of Manchester to develop a novel murine model of vestibular schwannoma. The student will perform stereotactic tumour implantation in the cerebellopontine angle and will then characterise nerve and tumour tissue in situ, mimicking vestibular schwannoma surgery.

Outcome: First demonstration of nerve and vestibular schwannoma differentiation in situ in a novel murine model.

Representative Publications from Supervisors

MacCormac, O., Horgan, C. C., Waterhouse, D., Noonan, P., Janatka, M., Miles, R., … & Shapey, J. (2025). Hyperspectral abdominal laparoscopy with real-time quantitative tissue oxygenation imaging: a live porcine study. Frontiers in Medical Technology, 7, 1549245
MacCormac, O., Noonan, P., Janatka, M., Horgan, C. C., Bahl, A., Qiu, J., … & Shapey, J. (2023). Lightfield hyperspectral imaging in neuro-oncology surgery: an IDEAL 0 and 1 study. Frontiers in Neuroscience, 17, 1239764
Shapey, J., Xie, Y., Nabavi, E., Ebner, M., Saeed, S. R., Kitchen, N., … & Vercauteren, T. (2022). Optical properties of human brain and tumour tissue: an ex vivo study spanning the visible range to beyond the second near‐infrared window. Journal of biophotonics, 15(4), e202100072
Li, P., Asad, M., Horgan, C., MacCormac, O., Shapey, J., & Vercauteren, T. (2023). Spatial gradient consistency for unsupervised learning of hyperspectral demosaicking: application to surgical imaging. International journal of computer assisted radiology and surgery, 18(6), 981-988
Budd, C., Qiu, J., MacCormac, O., Huber, M., Mower, C., Janatka, M., … & Vercauteren, T. (2023, October). Deep reinforcement learning based system for intraoperative hyperspectral video autofocusing. In International Conference on Medical Image Computing and Computer-Assisted Intervention (pp. 658-667). Cham: Springer Nature Switzerland
Ebner, M., Nabavi, E., Shapey, J., Xie, Y., Liebmann, F., Spirig, J. M., … & Vercauteren, T. (2021). Intraoperative hyperspectral label-free imaging: from system design to first-in-patient translation. Journal of Physics D: Applied Physics, 54(29), 294003

Application Process

More information about the opportunity here and here.

Navodini Wijethilake awarded at MRC DTP Symposium

Wed, 04 Jun 2025 00:00:01 +0000

Dr Wijethilake, who recently completed her PhD, was awarded the PPIE and Second Runner-Up awards in the iCASE Translational Fund Competition at the MRC DTP Symposium. The MRC DTP supported her throughout her PhD, including awarding her supplementary funding for PPIE activities and to support her transition from PhD to postdoc.

Navodini receives her prizes at the MRC DTP Symposium.

“Thrilled to have received the 2nd Runner-Up Prize in the MRC iCASE Translational Fund Competition and the Public and Patient Involvement and Engagement Prize at the DTP Symposium yesterday!
I’m incredibly grateful for the support I’ve received from the MRC DTP over the past 3.5 years — from fully funding my PhD, to awarding supplementary grants twice, as well as providing opportunities to present at symposiums. These experiences made my PhD journey truly rewarding and enjoyable. "
Dr Navodini Wijethilake

CholecInstanceSeg: A Tool Instance Segmentation Dataset for Laparoscopic Surgery

Tue, 03 Jun 2025 00:00:01 +0000

This paper introduces CholecInstanceSeg, the largest open-access dataset for surgical tool instance segmentation to date. It addresses key gaps in existing datasets by providing high-quality instance annotations for over 41,000 frames derived from clinical laparoscopic cholecystectomy procedures.

Visual representation of challenging annotation scenarios. Each image highlights a specific hard case: motion blur, smoke, soft tissue occlusion, tissue attachment, saturated lighting, instrument at the edge, reflection, instrument in fluid, low lighting, dirty lens, camera in port, and instrument far from the camera.

Research Assistant or Associate in Software Engineering and Surgical Imaging

Mon, 02 Jun 2025 00:00:01 +0000

Post overview:

Focus: Support surgical imaging data analysis tasks and work on computational algorithms to help streamline data annotation
Line manager: Tom Vercauteren
Salary: Grade 5, £38,482 - £43,249 or Grade 6, £44,355 – £47,882 (max SP34) per annum including LWA depending on experience
Duration: Fixed term contract until 30 Sep 2025

AI-enabled hyperspectral imaging in neurosurgery.

About Us

About the research group:

The CAI4CAI group, led by Prof Tom Vercauteren, is an academic research group focusing on Contextual Artificial Intelligence for Computer Assisted Interventions. Our engineering research aims at improving surgical and interventional sciences by exploiting learning-based approaches for multi-modal reasoning. We take a multidisciplinary, collaborative approach to solve clinical challenges.

About the department:

The mission of the Department of Surgical & Interventional Engineering (SIE) within the School of Biomedical Engineering & Imaging Sciences (BMEIS) is to bring together engineering and clinical experts to develop new surgical and interventional technologies for a wide range of clinical applications with a focus on combining diagnostic information to support image-guidance during procedures. The department boasts unique world-class facilities for SIE research. Our Validation Suite is a unique combination of an Integration Room, a versatile laboratory space where researchers can setup complex novel medical systems into fully integrated platforms, and an Intervention Room, a state-of-the-art simulated operative theatre where the technology can be deployed and tested on post-mortem models in a realistic surgical environment.

About the University:

King’s College London is an internationally renowned university delivering exceptional education and world-leading research. We are dedicated to driving positive and sustainable change in society and realising our vision of making the world a better place.

About The Role

We are seeking to recruit research software engineers to help translate the next generation of AI-assisted imaging systems for surgical guidance. The postholder, based within the Department of Surgical & Interventional Engineering at King’s College London, will play a key role in collaborative projects with King’s College Hospital. The role will involve supporting two ongoing clinical neurosurgery study. The successful candidates will join the clinical study team to help curate high-quality data.

The postholders will help support data analysis tasks and work on computational algorithms to help streamline data annotation. Image computing activities involve expanding on interactive segmentation algorithms to accelerate data annotation. The recruited individuals will complement our multidisciplinary team, undertake research in computer-assisted intervention, and consolidate our existing software infrastructure.

The post involves close and active collaboration with the surgical team, researchers, and engineers. The candidate will work closely with the rest of the team to correlate surgical data with other rich clinical data sources (e.g. surgical microscopy, histopathology).

This is a full-time post (35 hours per week), and you will be offered a fixed term contract until 30 Sep 2025.

About you

To be successful in this role, we are looking for candidates to have the following skills and experience:

Essential criteria

BSc/ MSc degree in computer vision or a closely related field
PhD in relevant subject area or equivalent industry experience (for Grade 6) at or near completion*
Excellent verbal and written communication skills
Experience of software development with Python
Experience of machine learning with PyTorch
Good knowledge of machine learning and computer vision algorithms
Ability to work on own initiative and in a team
Experience in collaborative software development (for Grade 6)

Desirable criteria

Understanding of translational research requirements for surgical imaging research

Downloading a copy of our Job Description Full details of the role and the skills, knowledge and experience required can be found in the Job Description document, provided here and after you click “Apply Now”. This document will provide information of what criteria will be assessed at each stage of the recruitment process.

Further Information

We pride ourselves on being inclusive and welcoming. We embrace diversity and want everyone to feel that they belong and are connected to others in our community. We are committed to working with our staff and unions on these and other issues, to continue to support our people and to develop a diverse and inclusive culture at King’s.

We ask all candidates to submit a copy of their CV, and a supporting statement, detailing how they meet the essential criteria listed in the advert. If we receive a strong field of candidates, we may use the desirable criteria to choose our final shortlist, so please include your evidence against these where possible.

To find out how our managers will review your application, please take a look at our ‘How we Recruit’ pages on the King’s website.

More information about the position and how to apply here or here.

Neurosurgical studies using novel light-based imaging systems hold patient involvement day

Tue, 06 May 2025 00:00:01 +0000

Two neurosurgical studies led by CAI4CAI researching novel, light based, imaging systems that can be used to greatly improve the efficacy of future neurosurgical procedures ran a successful day of patient involvement.

A patient representative manipulating a surgical microscope featuring our hyperspectral imaging technology for neurosurgery.

Each year in the UK, approximately 70,500 patients are diagnosed with a brain tumour, 5,000 of whom undergo surgery. Even with the best hands and the most advanced technology currently available, it remains difficult to determine tumour from healthy brain tissue. Due to this uncertainty and the need to balance risks, some tumour is often left behind.

Today, close to 30% of brain tumour patients require repeat surgery owing to tumour left behind during their first surgery. Further surgeries are more difficult, pose additional patient risks and lead to increased healthcare costs with often poor patient outcomes.

Newly developed AI-enabled camera systems have the potential to enhance the surgeon’s vision to reliably identify tumour and healthy brain structures. Hyperspectral imaging (HSI) is one of the most promising of such technologies. Its core ability is to provide very detailed and rich light-based information that is invisible to the human eye.

Two research projects, NeuroHSI and NeuroPPEye, assessed whether the new technology is capable of obtaining helpful images for surgical guidance in the future. The NeuroHSI research project is funded by the National Institute for Health and Care Research (NIHR) and the NeuroPPEye project is funded by Wellcome [223880/Z/21/Z].

The projects were designed with patient input and feedback from various patient groups including The Brain Tumour Charity. They involved establishing a Patient Advisory Group to advise on all stages of the research.

The group include patients with first-hand experience of neuro-oncology and neuro-vascular surgery, providing a range of perspectives pertinent to the work. This collaborative approach not only helps in recruiting suitable participants but also ensures that the research addresses real patient needs and that findings are shared in a way that make sense to everyone.

Researchers, members of the public and patient involvement group and funders, came together to review the progress of these projects. A hands-on demonstration of the technology was organised for the Patient Advisory Group at the School of Biomedical Engineering & Imaging Sciences, Surgical and Interventional Engineering mock operating theatre.

“The NeuroHSI and NeuroPPEye project are highly translational in nature. Our objective is to develop technology that addresses critical unmet clinical needs in a way that is acceptable to all stakeholders. Our supportive patient group has allowed us to go above and beyond what we would otherwise have been able to achieve in terms of lowering barriers to adoption.”
Prof. Tom Vercauteren, Professor of Interventional Image Computing

Group picture from our PPI event.

“The Patient Advisory Group have found our involvement in these exciting projects extremely interesting and rewarding. Being able to channel our experiences as patients and carers into projects which have so much potential to help those in our positions in the future is incredibly empowering.”
Andrew Plowright, Patient Advisory Group chair

Hands-on demonstration of the laparoscopic hyperspectral imaging device of Hypervision Surgical.

“This technology has the potential to significantly improve safety and accuracy of surgery. Working with our inspiring and engaging patient group has been one of the project’s highlights; they have helped to steer this research at every stage.”
Jonathan Shapey, Clinical Reader and Honorary Consultant Neurosurgeon

If you have been diagnosed with, or treated for a brain tumour and would be interested in helping to guide cutting edge research in neurosurgery and AI please contact the NeuroHSI team or NeuroPPEye team to enquire about space in the Science for Tomorrow’s Neurosurgery PPI group.

To learn more about the project, please visit our neurohsi.uk and neuroppeye.uk pages.

Tom Vercauteren to deliver keynote at MICAD 2025

Fri, 02 May 2025 00:00:01 +0000

Tom Vercauteren has been invited to be a keynote speaker at the MICAD 2025 conference (The 6th International Conference on Medical Imaging and Computer-Aided Diagnosis) to be held in London in November.

The 6th edition of the International Conference on Medical Imaging and Computer-Aided Diagnosis (MICAD 2025) will be held in London, UK, from November 19–21, 2025.

KDC Africa PhD studentship [October 2025 start preferred - no later than June 2026] on "Resource-efficient slice-to-volume MRI super-resolution reconstruction for improved meningioma management in Sub-Saharan Africa"

Tue, 28 Jan 2025 00:00:01 +0000

Applications are invited for a fully funded 4 years King’s Doctoral College London Africa PhD Studentships (including tuition fees, annual stipend and consumables) starting preferably in October 2025 and no later than June 2026.

To be eligible, student must be an African national and be permanently resident in an African country.

This project is a collaboration with the Medical Artificial Intelligence Laboratory (MAI Lab), in Lagos, Nigeria.

Award details:

Focus: MRI super-resolution reconstruction in Sub-Saharan Africa
First Supervisor: Tom Vercauteren
Second supervisor: Jonathan Shapey
Third Supervisor: Udunna Anazodo
Funding type: 4-year fully-funded PhD studentship including stipend, tuition fees, research training and support grant (RTSG), visa and HIS fees.
Application closing date: 02 March 2025
Start date: October 2025 prefered and no later than June 2026

Slice-to-volume MRI super-resolution reconstruction workflow.

Aim of project

This project aims at improving the quality of brain MRI data available for meningioma management in sub-Saharan Africa (SSA) through the development of novel resource-efficient artificial intelligence algorithms.

Project description

Access to MRI in SSA is restricted and its quality is limited by the hardware available locally. While high-resolution high-field-strength volumetric brain MRI is typically acquired in developed countries, a typical approach used in SSA relies on acquiring stacks of low-resolution low-field-strength slices as these are faster to acquire and more suitable for locally available MR scanners.

Meningiomas are the most common primary brain tumours in adults. The wide range of presentation they are associated with spans from slow-growing lesions to highly aggressive ones. This makes therapeutic approaches challenging and highly dependent on radiological findings, the latler directly depending on MRI quality.

We have shown in previous work for fetal brain and abdominal MRI (where slice-based acquisitions are used in developed countries to combat motion artifacts) that slice to volume approaches can effectively be used to reconstruct high-quality volumetric MRI from stacks of 2D MRI slices. Yet, these approaches typically necessitate important computational resources which would lead to a bottleneck for their clinical translation in SSA. In this project, the student will design efficient learning-based approaches for this problem and validate its usefulness using real-world meningioma data acquired in Nigeria. A close collaboration with local radiographers and clinicians associated with the Medical Artificial Intelligence Laboratory (MAI Lab) in Lagos will ensure practical relevance.

The supervisory team combines the wide range of inter-disciplinary expertise required to successfully deliver on this project. Tom Vercauteren (first supervisor) brings his machine learning background and track record in developing translational image computing solutions. Jonathan Shapey (second supervisor) brings his neurosurgical expertise and experience in integrating AI-defined devices in the clinical workflow. Udunna Anazodo (lead external partner) brings her quantitative neuroimaging expertise and her leadership in improving access to diagnostic imaging for global health. Below are two relevant papers from the team illustrating the strength of the track record for the project both in terms of engineering methodology and global health impact.

This project is a collaboration with the Medical Artificial Intelligence Laboratory (MAI Lab), in Lagos, Nigeria (https://mailab.io/). As part of this proposed project, the MAI Lab is committing to providing local expertise as well as meningioma data. We will also seek to organise at least one on-site visit / short placement from the PhD student to the MAI Lab in Lagos.

Expected academic background and academic eligibility critwria

Outstanding students with a solid background in data science are invited to apply for this PhD project. There is an expectation of good communication and teamwork. Experience in medical imaging, especially handling MRI scans, would be advantageous.

Academic eligibility criteria:

Bachelor’s degree with 2:1 honours in Applied Mathematics / Mathematics, Computer Science, Physics, Biomedical Engineering or another engineering-related discipline
A 2:2 Bachelor’s degree may be considered only where applicants also offer a Master’s degree with Merit or above.
Outstanding students with a solid background in data science are invited to apply for this PhD project. There is an expectation of good communication and teamwork. Experience in medical imaging, especially handling MRI scans, would be advantageous.

We welcome eligible applicants from any personal background, who are pleased to join diverse and friendly research groups.

For information on our English language requirements and whether you need to complete an English language test, please see King’s English Language requirements page.

Application Process

Please submit an application for the Biomedical Engineering and Imaging Science Research MPhil/PhD (Full-time) programme using the King’s Apply system. Please include the following with your application:

A PDF copy of your CV should be uploaded to the Employment History section.
A 500-word personal statement outlining your motivation for undertaking postgraduate research should be uploaded to the Supporting statement section.
Samples of relevant work (e.g. publications, MSc thesis, software repository) and reference letters can also be provided in support of the application.
Funding information: On the ‘Funding’ Section of your King’s Apply application, you MUST tick the box at item 5 (Award Scheme Code or Name) and enter the funding code: BMEIS_AFRICA_TV (Please copy and paste the code exactly!). Failing to include this code might result in you not being considered for this funding.

More information here.

New software developed to support surgical robotics

Wed, 27 Nov 2024 00:00:01 +0000

Researchers from the School of Biomedical Engineering & Imaging Sciences have successfully integrated KUKA’s Fast Robot Interface (FRI) with ROS 2 and Python, significantly enhancing the capabilities of the surgical robot.

Researchers working with a KUKA robotic arm.

KUKA’s LBR Med robot has been adapted to meet specific medical requirements and is perfectly suited for a wide range of assistance systems in medical technology on account of its human-robot collaboration capability.

The LBR-Stack project has developed an integration that simplifies the usage of these robots in real-time applications, providing robust support for both simulation and real hardware communication.

The potential impact of the project is substantial, particularly in the fields of medical robotics and industrial automation which is the research team’s primary interest.

“Ultimately, this software makes the robot part of a bigger and more flexible ecosystem. This enables surgeons to contribute novel and less restrictive workflows that are tailored to the patient’s specific needs. From an engineering perspective, the software allows for full hardware abstraction, which will help iterate systems quicker and drive down cost in the future.”
Dr Martin Huber, PhD Student, School of Biomedical Engineering & Imaging Sciences, King’s College London

The project includes various packages such as Python bindings and ROS 2 integration, making it the most comprehensive solution for developers and researchers working with KUKA robots.

By offering a single, unified framework that works with multiple FRI versions, the project makes building robotic applications smoother and faster.

The project has garnered significant attention, with over 150 stars on GitHub, a platform that allows developers to create, store, manage and share their code. Indicating its potential for global reach and the community’s recognition of its value.

PhD opportunity [October 2025 start] on "Machine Learning Tool for Predicting Digital Twin Trajectories of Meningioma Growth on MRI Brain Scans"

Sun, 17 Nov 2024 00:00:01 +0000

Applications are invited for a fully funded 3.5 years PhD CDT DT4Health studentship (including tuition fees, annual stipend and consumables) starting in October 2025.

Award details:

Focus: Digital twin with interventional neuroradiology applications
Primary supervisor: Thomas C Booth
Secondary supervisor: Jonathan Shapey
Third Supervisor: Tom Vercauteren
Funding type: 3.5-year fully-funded CDT DT4Health studentship including a stipend, tuition fees, research training and support grant (RTSG), and a travel and conference allowance.
Application closing date: Friday 3rd January 2025
Start date: October 2025

Machine Learning Tool for Predicting Digital Twin Trajectories of Meningioma Growth on MRI Brain Scans.

Aim of project

The aim of this project is to develop a machine learning tool that generates digital twin trajectories for meningioma growth in brain MRI scans. By leveraging deep learning, the tool will create patient-specific models (digital twins) that predict the future growth of meningiomas based on historical MRI data. This tool will support clinicians in anticipating changes in tumour size and shape, enabling proactive, personalized management of meningioma patients. In particular, the model may equip clinicians with a predictive tool that enhances decision-making, allowing for timely interventions and optimized monitoring. Additionally, this approach may lay the groundwork for digital twin applications in tracking other types of tumours, broadening its impact across oncology diagnostics.

Project description

Meningiomas are the most common primary brain tumours, originating from the meninges, the protective layers surrounding the brain and spinal cord. Although typically benign, meningiomas exhibit diverse growth rates and behaviours. Some remain indolent for years, while others progress rapidly, potentially causing severe neurological deficits such as seizures, vision loss, and cognitive impairment due to mass effects on surrounding brain structures. Understanding and predicting their growth trajectory is therefore critical for timely intervention and effective management.

Currently, monitoring meningiomas relies heavily on routine MRI scans taken at intervals to assess changes in size and morphology. However, this conventional approach is limited, as it provides only a retrospective view of tumour progression, which restricts proactive clinical decision-making. MRI scans may reveal a tumour’s current state, but they lack the predictive ability to forecast future growth patterns or sudden accelerations in growth that could impact patient outcomes. Thus, reliance solely on interval-based imaging may delay critical interventions, highlighting the need for more advanced predictive tools.

To address these limitations, this project proposes the development of a machine learning-based digital twin model tailored specifically for meningioma growth prediction. A digital twin is a dynamic, computational model that continuously updates in response to real-time data, effectively mirroring the evolving characteristics of a biological entity. For meningiomas, creating digital twins entails using historical MRI data and patient-specific clinical features to train a deep learning model that can simulate individualized tumour growth trajectories.

By using advancements in deep learning and temporal data analysis, the model built during the PhD will aim to forecast future changes in tumour volume and morphology, allowing clinicians to personalize monitoring schedules and treatment plans. The model’s ability to project future growth trajectories can provide valuable lead time for clinical interventions, particularly in cases where rapid growth is anticipated. Such a tool could transform meningioma management by supporting data-driven, proactive strategies, optimizing patient outcomes through timely and individualized care.

Skills learnt: Machine learning, MRI translational design, statistics. Also, broader topics including Careers & Employability, Communication & Impact, Personal Effectiveness, Writing & Publishing. Complemented by faculty and departmental lectures; seminars; one-to-one supervisions to develop skill in data handling and analysis.

Expected academic background

Application Process

More information about the opportunity here and here.

Science for tomorrow's neurosurgery: Patient & Public Involvement (PPI) group - Septembre 2024 group meeting

Wed, 25 Sep 2024 00:00:01 +0000

The 25th of September saw the latest meeting of “Science for Tomorrow’s Neurosurgery,” our now well established PPI group. As always, lots of exciting and valuable discussion with updates from Oscar on the (nearly complete!) NeuroHSI recruitment as well as Matt announcing the official opening of NeuroPPEYE phase 2!

Live scribe of the conversations between patients and researchers.

Neurosurgical research at King’s is going from strength to strength, resulting in lots of exciting new opportunities for patients to get involved with. However, we are always aware that being given lots of information about different trials from different people recruiting can be burdensome for patients, especially if their first contact is receiving a difficult diagnosis or are being told they will need an operation. We discussed this concern with the group and how best to approach recruitment; this generated lots of valuable discussion, resulting in going forward with a combined trial recruitment approach within our group; yet another example of ensuring everything we do remains patient focused! Our live illustrator, Jenny Leonard was on hand to provide us with a fantastic representation of our discussion points, for which we are always grateful.

As always, a hugely positive meeting and an overall experience that has been so valuable to us, we have decided to publish it! Read more here:

MacCormac, O., Elliot, M., Whittaker, L., Bahl, A., Ségaud, S., Plowright, A.J., Winslade, S., Taylor-Gee, A., Spencer, B., Vercauteren, T. and Shapey, J., 2024. Science for tomorrow’s neurosurgery: insights on establishing a neurosurgery patient group focused on developing novel intra-operative imaging techniques. Research Involvement and Engagement, 10(1), pp.1-12. doi:10.1186/s40900-024-00649-0

To learn more about the project, please visit our neurohsi.uk and neuroppeye.uk pages.

PhD opportunity [October 2025 start] on "Computational stereovision synthesis from monocular neuroendoscopy"

Tue, 24 Sep 2024 00:00:01 +0000

Applications are invited for the fully funded 1+3 years MRes+PhD or 4 years PhD MRC DTP studentship (including home tuition fees, annual stipend and consumables) starting in October 2025.

Award details:

Focus: Computer vision for surgical imaging applications
Primary supervisor: Tom Vercauteren
Secondary Supervisor: Jonathan Shapey
Third Supervisor: Dr Pavol Surda
Funding type: 4-year fully-funded MRC DTP studentship including a stipend, tuition fees, research training and support grant (RTSG), and a travel and conference allowance.
Application closing date: Tuesday 29 October 2024
Start date: October 2025

Computational stereovision synthesis from monocular neuroendoscopy.

Project Overview

This project focuses on machine learning approaches to synthetise 3D stereoscopic display from a monocular video feed captured through a standard neuroendoscope. The 3D vision capability will be tested in realistic surgical exercises to evaluate the benefit of computationally inferred depth cues.

Endoscopic surgery has become the gold standard for the treatment of several skull base and ear, nose and throat (ENT) pathologies. However, restricted access and viewing conditions make endoscopic endonasal procedures challenging and complex to learn. Most endonasal procedures are performed using monocular endoscopes which further reduces depth perception ability. Stereo-neuroendoscopy has recently entered the market. It has been shown to improve surgical training, but its adoption has been hampered by additional cost and relative reduction of image quality.

To overcome this barrier, this project proposes to exploit standard high-quality monocular endoscopy video feeds to generate stereoscopic vision. Recent developments in machine learning have demonstrated the ability of large models to generate accurate depth maps from monocular surgical images. Yet, these approaches remain at an early stage, have not been extended to real-time surgical video inference, and have not been utilised to solve any clinically relevant task. The PhD student will address these limitations by taking advantage of the multidisciplinary translational research environment nurtured by the co-supervisors.

In Year 1, the student will familiarise themselves with multi-view geometry and generative models and will design learning-based approaches exploiting physics-based constraints.

In Year 2, network distillation and lightweight generative models will be developed to enable real-time inference capabilities. Throughout the project, the student will interact closely with surgeons and member of the broader clinical team.

This leads to Years 3-4 focusing on conceiving and implementing robust useability studies to evaluate the impact of the research on clinically relevant surrogate tasks, iterating on previous developments, and writing up the thesis.

Representative Publications from Supervisors

Budd, C., & Vercauteren, T. (2024). Transferring Relative Monocular Depth to Surgical Vision with Temporal Consistency. Proc MICCAI 2024. https://arxiv.org/abs/2403.06683
Ahmad, M. A., Weiler, Y., Joyeux, L., Eixarch, E., Vercauteren, T., Ourselin, S., … & Vander Poorten, E. (2023). 3D vs. 2D simulated fetoscopy for spina bifida repair: a quantitative motion analysis. Scientific Reports, 13(1), 20951. https://doi.org/10.1038/s41598-023-47531-9
Vercauteren, T., Unberath, M., Padoy, N., & Navab, N. (2019). CAI4CAI: The rise of contextual artificial intelligence in computer-assisted interventions. Proceedings of the IEEE, 108(1), 198-214. https://doi.org/10.1109%2FJPROC.2019.2946993
Khan, D. Z., … Shapey, J., … & Babu, P. (2023). CSF rhinorrhoea after endonasal intervention to the skull base (CRANIAL): A multicentre prospective observational study. Frontiers in Oncology, 12, 1049627. https://doi.org/10.1016/j.wneu.2020.12.171
Pak, H. L., Lambru, G., Okasha, M., Maratos, E., Thomas, N., Shapey, J., & Barazi, S. (2022). Fully endoscopic microvascular decompression for trigeminal neuralgia: technical note describing a single-center experience. World Neurosurgery, 166, 159-167. https://doi.org/10.1016/j.wneu.2022.07.014
Kitchen, N., & Shapey, J. (2019). The operating theatre environment. Oxford textbook of neurological surgery. Oxford University Press, Oxford, 45-56. https://doi.org/10.1093/med/9780198746706.003.0004

Application Process

More information about the opportunity here and here.

CAI4CAI research to be presented at MICCAI 2024 - 4 full papers (including 2 early accept), one CLINICCAI oral presentation

Sun, 23 Jun 2024 00:00:01 +0000

The main MICCAI 2024 outcomes are now out and it’s a pleasure to announce that our research will be prominently featured. Congrats to the main authors from the group and all co-authors!

Our monocular depth perception work was rewarded with an early accept decision.

Main conference papers:

Charlie Budd, and Tom Vercauteren. “Transferring Relative Monocular Depth to Surgical Vision with Temporal Consistency.”
- Early accept preprint: https://arxiv.org/abs/2403.06683
- Code and model: https://github.com/charliebudd/transferring-relative-monocular-depth-to-surgical-vision
Aaron Kujawa, Reuben Dorent, Sebastien Ourselin, and Tom Vercauteren. “Label merge-and-split: A graph-colouring approach for memory-efficient brain parcellation.”
- Early accept preprint: https://arxiv.org/abs/2404.10572
Theodore Barfoot, Luis Garcia-Peraza-Herrera, Ben Glocker, and Tom Vercauteren. “Average Calibration Error: A Differentiable Loss for Improved Reliability in Image Segmentation.”
- Preprint: https://arxiv.org/abs/2403.06759
- Code: https://github.com/cai4cai/ACE-DLIRIS
Qi Li, Ziyi Shen, Qianye Yang, Dean Barratt, Matthew Clarkson, Tom Vercauteren and Yipeng Hu.“Nonrigid Reconstruction of Freehand Ultrasound without a Tracker.”

CLINICCAI oral presentation:

N.A. Cavalcanti, F. Carrillo, R. Li, K. van Assche, A. Davoodi, M. Tummers, M. Huber, F. Teyssere, J.A. Perez Velásquez, A. Massalimova, C.J. Laux, R. Sutter, M. Farshad, G. Borghesan, K. Denis, G. Morel, T. Chandanson, T. Vercauteren, E. Vander Poorten, P. Fürnstahl. “Bridging innovation and practice: the journey of FAROS from technical design to in-vivo animal validation”

Contribution to satelite events:

Dominic LaBella, Katherine Schumacher, Michael Mix, Kevin Leu, Shan McBurney-Lin, Pierre Nedelec, Javier Villanueva-Meyer et al. “Brain Tumor Segmentation (BraTS) Challenge 2024: Meningioma Radiotherapy Planning Automated Segmentation.”
- Preprint: https://arxiv.org/abs/2405.18383
TUS-REC Challenge
- Website: https://github-pages.ucl.ac.uk/tus-rec-challenge/

CAI4CAI on github

Sun, 23 Jun 2024 00:00:00 +0000

Transferring Relative Monocular Depth to Surgical Vision with Temporal Consistency

Thu, 20 Jun 2024 00:00:01 +0000

Science for tomorrow's neurosurgery: Patient & Public Involvement (PPI) group - February 2024 group meeting

Thu, 22 Feb 2024 00:00:01 +0000

On 22nd of February we had our 5th “Science for Tomorrow’s Neurosurgery” group meeting.

Live scribe of the conversations between patients and researchers.

With exciting updates from both the NeuroPPEYE and NeuroHSI studies, the group were thrilled to learn of the progress we are making. Ensuring adherence to the core principles of “Co-Learning” and “Reciprocal Relationships,” our research team broke down the complexities in a way that was both accessible and engaging for the group members. This is emphasised beautifully by the following quote from one of our group members:

The quite complex information was presented in such a way that it was easily understood by those of us with a non-medical background. The information was pitched at exactly the right level and was interesting, informative and of great interest. It also made me feel as though I was actively involved in helping people in the future. I have some fairly difficult side effects to live with since my op so knowing that it may be possible to limit or eradicate those in the future makes me feel as though I have done something to help others. Living with any type of brain tumour can be quite life changing and by being involved with the research it will hopefully benefit others in the future. It was also great to meet other people who have been in the same situation, to meet the researchers that are working towards better outcomes for patients and to have some insight as to how tumours are seen by surgeons. I found it particularly fascinating to see how AI can make the colour spectrum appear the same to each and every surgeon rather than relying on an individual’s interpretation of the colours they see themselves, which can differ greatly from person to person.

Continuing with these core principles, the research team sough to gain thoughts from our PPI group on the need to carry out an animal study in order to validate the work we are doing. This is, understandably, an emotive topic for all parties and it was extremely important to receive the support from our PPI group on this matter. It also provided the research team with reassurance that we had indeed exhausted all other possibilities and confirmed that the work is necessary.

As always, our live illustrator, Jenny Leonard was on hand to provide us with a fantastic representation of our discussion points, for which we are always grateful!

We are already preparing for our next meeting in six month time and look forward to updating afterwards; watch this space!

To learn more about the project, please visit our neurohsi.uk and neuroppeye.uk pages.

Tom Vercauteren appointed Senior Editor for Medical Image Analysis (MedIA) Journal

Mon, 01 Jan 2024 00:00:01 +0000

Tom Vercauteren has been appointed as Senior Editor for the highly-ranked Medical Image Analysis journal, an official journal of the MICCAI society. Sebastien Ourselin also features on the editorial board of the publication.

Tom Vercauteren appointed Senior Editor for MedIA Journal.

Harnessing colour and light to raise awareness of brain tumours

Mon, 04 Dec 2023 00:00:01 +0000

“The opportunity to hear from clinicians, talk to people in my situation and benefit from their feedback was what drew me to this project. I realised the extent of the number of people that are affected by brain tumours," Kate Dooley.

Kate was a participant in “Illuminating the Brain” a collaborative project from artist Kyam, the NeuroHSI and NeuroPPEye researcher group at King’s College London and patients and carers from The Brain Tumour Charity, that exhibited at the Science Gallery London.

Artwork harnessing colour and light to convey varied experiences relating to brain tumours.

There are increasing numbers of new brain tumour diagnoses each year, the current estimate is 13,000. Researchers and clinicians at King’s College London are consistently working on new and better ways to identify and extract these tumours quickly and safely. A large part of supporting diagnoses is awareness and, in this case, awareness comes in the form of art and discussion.

Informed by conversations between researchers, patients and carers, the artwork harnesses colour and light to convey the group’s varied experiences relating to brain tumours. Light and colour were chosen as they reflect the essence of the research project, which is investigating the potential of Hyperspectral Imaging to improve surgery for brain tumour removal.

“Hyperspectral imaging is quite a simple technology. Human vision is based on the detection of red, green and blue centered bands of light that are recombined by the brain to produce colors just like that can be seen in a rainbow. There are actually infinite intermediate colors, and sensing a larger number of them with a bespoke camera that can separate finer bands of light provides much richer information.

Sensing more data alone is not solving the challenge, extraction of relevant information is required and needs to be converted into visual output like images. Hyperspectral imaging has the potential to deliver more information for the surgeon to distinguish between healthy and diseased tissue in each individual patient.” Tom Vercauteren, Professor of interventional image computing at King’s College London.

Panel discussion at the showcase.

The Illuminating the Brain project sought to bring researchers, patients and carers together to facilitate sharing across the community and to create an artwork that would raise awareness of brain tumours and the associated research.

“From the workshop, common themes from all groups – patients, clinicians, researchers – were highlighted including trust, the notion of individual but also the different forms of emotional attachment,” Jonathan Shapey, Senior clinical lecturer, King’s College London & King’s College Hospital.

The strength of these discussions and the cultivation of lived experience, enabled artist Kyam to create artwork that really resonated with those who came to experience the showcase.

“The artwork was very thought provoking, especially as someone with lived experience of a brain tumour it’s the first time I’ve seen an artist attempt to capture how it feels,” a patron.

Overview of the artwork at the showcase.

“Collaboration is key in all that has been discussed including sharing practices among clinicians, raising awareness about and ensuring access to participate in research studies. This project demonstrated the positive outcome it has and it is encouraging”. Shannon Winslade, Involvement and Impact Manager at Brain Tumour Charity.

This project was coordinated by the School of Biomedical Engineering & Imaging Sciences Public Engagement team.

To learn more about the project, please visit our neurohsi.uk and neuroppeye.uk pages.

PhD opportunity [October 2024 start] on "Text promptable semantic segmentation of volumetric neuroimaging data"

Wed, 04 Oct 2023 00:00:01 +0000

Applications are invited for the fully funded 4 years full-time PhD studentship (including home tuition fees, annual stipend and consumables) starting in Ocober 2024.

Award details:

Focus: Vision-language models for neuroimaging
Primary supervisor: Tom Vercauteren
Secondary supervisor: Alexander Hammers
Funding type: 4-year fully-funded MRC DTP studentship including a stipend, tuition fees, research training and support grant (RTSG), and a travel and conference allowancep.
Application closing date: 8 November 2023
Start date: October 2024

Text promptable semantic segmentation of volumetric neuroimaging data.

Project Overview

Semantic segmentation of brain structures from medical images, in particular Magnetic Resonance Imaging (MRI), plays an important role in many neuroimaging applications. Deep learning based segmentation algorithms are now achieving state-of-the-art segmentation results but currently require large amounts of annotated data under predefined segmentation protocols and data inclusion/exclusion criteria. The rigidity of such approaches forbids natural interactions by humans and thus limits the usefulness for non-routine questions.

In this project, we will develop a novel AI agent able to segment neuroanatomical structures based on a textual prompt describing the structure to be segmented. Similar to how a senior clinician would provide feedback to a trainee, we will further allow providing the AI agent with additional textual information relating to the case at hand and any previous segmentation proposal by the AI agent.

Large language models have recently enabled agile human-AI interactions trough textual prompting but their translation to medical imaging question remains limited. A key challenge in designing vision-language models for volumetric neuroimaging tasks is the lack of pre-existing suitable foundation AI models. The project will take advantage of pre-trained medical language models as well as prior neuroanatomical knowledge captured through brain atlases, publications, and medical records to constrain the problem sufficiently. We will then develop latent image representations aligned with frozen latent representations extracted from medical language.

In year 1, the student will focus on assembling a collection of datasets of brain MRIs annotated according to a wide range of established protocols. A bespoke segmentation pipeline will be developed to handle the large number of potentially overlapping classes. In year 2, the focus will move to the development of a neuroimaging vision-language model. In year 3, agile text prompting and refinement will be designed. Year 4 will be dedicated to validation studies and thesis write-up.

Representative Publications from Supervisors

Zhou, Z., Alabi, O., Wei, M., Vercauteren, T., & Shi, M. (2023). Text Promptable Surgical Instrument Segmentation with Vision-Language Models. Proc. NeurIPS 2023. https://doi.org/10.48550/arXiv.2306.09244
Shapey, J., Kujawa, A., Dorent, R., Wang, G., Dimitriadis, A., Grishchuk, D., … & Vercauteren, T. (2021). Segmentation of vestibular schwannoma from MRI, an open annotated dataset and baseline algorithm. Scientific Data, 8(1), 286. https://doi.org/10.1038/s41597-021-01064-w
Dorent, R., Booth, T., Li, W., Sudre, C. H., Kafiabadi, S., Cardoso, J., … & Vercauteren, T. (2021). Learning joint segmentation of tissues and brain lesions from task-specific hetero-modal domain-shifted datasets. Medical image analysis, 67, 101862. https://doi.org/10.1016/j.media.2020.101862
Ioannou S, Chockler H, Hammers A, King AP, Alzheimer’s Disease Neuroimaging Initiative. A study of demographic bias in CNN-based brain MRI segmentation. Machine Learning in Clinical Neuroimaging: 5th International Workshop, MLCN 2022, Held in Conjunction with MICCAI 2022, Singapore, September 18, 2022, Proceedings, https://doi.org/10.1007/978-3-031-17899-3_2
Heckemann RA, Hajnal JV, Aljabar P, Rueckert D, Hammers A. Automatic anatomical brain MRI segmentation combining label propagation and decision fusion. Neuroimage 2006, 33(1): 115-126. https://doi.org/10.1016/j.neuroimage.2006.05.061.
Steinbart D, Yaakub SN, Steinbrenner M, Guldin LS, Holtkamp M, Keller SS, Weber B, Rüber T, Heckemann RA, Ilyas-Feldmann M, Hammers A; Alzheimer’s Disease Neuroimaging Initiative. Automatic and manual segmentation of the piriform cortex: Method development and validation in patients with temporal lobe epilepsy and Alzheimer’s disease. Hum Brain Mapp. 2023 Jun 1;44(8):3196-3209. https://doi.org/10.1002/hbm.26274

Application Process

More information about the opportunity here and here.

Science for tomorrow's neurosurgery: Patient & Public Involvement (PPI) group - September 2023 group meeting

Thu, 21 Sep 2023 00:00:01 +0000

On 21st September we held our fourth ‘Science for Tomorrow’s Neurosurgery’ PPI group meeting online, with presentations from Oscar, Matt and Silvère. Presentations focused on an update from the NeuroHSI trial, with clear demonstration of improvements in resolution of the HSI images we are now able to acquire; this prompted real praise from our patient representatives, which is extremely reassuring for the trial going forward. We also took this opportunity to announce the completion of the first phase of NeuroPPEYE, in which we aim to use HSI to quantify tumour fluorescence beyond that which the human eye can see. Discussions were centered around the theme of “what is an acceptable number of participants for proof of concept studies,” generating very interesting points of view that ultimately concluded that there was no “hard number” from the patient perspective, as long as a thorough assessment of the technology had been carried out. This is extremely helpful in how we progress with the trials, particularly NeuroPPEYE, which will begin recruitment for its second phase shortly. Once again, the themes and discussions were summarized in picture format by our phenomenal illustrator, Jenny Leonard (see below) and we are already making plans for our next meeting in February 2024!

Live scribe of the conversations between patients and researchers.

To learn more about the project, please visit our neurohsi.uk and neuroppeye.uk pages.

Jonathan Shapey delivers the Hunterian Lecture at the Society of British Neurological Surgeons autumn congress

Sat, 16 Sep 2023 00:00:01 +0000

Jonathan Shapey had the great honour to deliver the Hunterian Lecture at the Society of British Neurological Surgeons autumn congress (SBNS London 2023). Jonathan presented his work in developing a label-free real-time intraoperative hyperspectralimaging system for neurosurgery.

Hunterian Lectureships are highly regarded and prestigious awards awarded by The Royal College of Surgeons of England.

Jonathan Shapey delivering the Hunterian Lecture at SBNS London 2023.

MICCAI 2023 Presentation for Deep Homography Prediction for Endoscopic Camera Motion Imitation Learning

Fri, 15 Sep 2023 09:00:00 +0000

This video presents work lead by Martin Huber. Deep Homography Prediction for Endoscopic Camera Motion Imitation Learning investigates a fully self-supervised method for learning endoscopic camera motion from readily available datasets of laparoscopic interventions. The work addresses and tries to go beyond the common tool following assumption in endoscopic camera motion automation. This work will be presented at the 26th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2023).

Presentation Video for IEEE IUS

Sat, 26 Aug 2023 09:00:00 +0000

This video presents work lead by Mengjie Shi focusing on learning-based sound-speed correction for dual-modal photoacoustic/ultrasound imaging. This work will be presented at the 2023 IEEE International Ultrasonics Symposium (IUS).

You can read the preprint on arXiv: 2306.11034 and get the code from GitHub.

Patient Centered Vestibular Schwannoma Management: Patient & Public Involvement (PPI) group - June 2023 Group Meeting

Tue, 22 Aug 2023 00:00:01 +0000

Recently, we organized a Public and Patient Involvement (PPI) group with Vestibular Schwannoma patients to understand their perspectives on an patient-centered automated report. Partnering with the British Acoustic Neuroma Association (BANA), we recruited participants by circulating a form within the BANA community through their social media platforms.

Live scribe of the conversations between patients and researchers.

This approach garnered interest from a wide pool of participants across the UK, with over 200 VS patients expressing their willingness to take part. From this group, 12 individuals were thoughtfully selected to ensure diverse representation in terms of age, gender, and treatment paths, thus enriching the workshop’s insights. Further, to ensure broad participation from patients across the UK, this event was conducted online.

During the patient involvement event, Jonathan Shapey, consultant neurosurgeon, commenced the session by providing a clinical context on Vestibular Schwannoma and discussing the potential implications of automation in clinical care. Subsequently, Navodini Wijethilake presented a concise overview, outlining the automated report generation process and the pivotal role that Artificial Intelligence plays in addressing the clinical challenges.

The workshop then transitioned into small focus group discussions, each led by the impartial facilitator (William Hunter), Jonathan, and Navodini. These groups, comprising of 4 participants, facilitated candid conversations around several pertinent themes. We explored participants’ prior knowledge of the process, their feelings and concerns about automated measurements, and their preferences regarding visual aids for tracking tumour progression.

The workshop ended with a large group discussion that brought together the main ideas from the focus group sessions. An impartial facilitator guided the workshop, and an artist captured visual notes of the day’s activities and discussion points. This event was funded by the MRC DTP Flexible Supplement Fund.

Research Associate / Research Fellow in "Computational Hyperspectral Imaging"

Tue, 11 Jul 2023 00:00:01 +0000

Post overview:

Focus: Translational research on hyperspectral-based quantitative fluorescence imaging linked with a neurosurgery clinical study
Line manager: Tom Vercauteren
Clinical collaborator: King’s College Hospital
Industry collaborator: Hypervision Surgical
Salary: Grade 6, £41,386-£48,414 or Grade 7 £49,737-£55,306 per annum, including London Weighting Allowance

State-of-the-art fluorescence imaging system for neurosurgical guidance.

Job description

We are seeking an interventional image computing researcher to design and translate the next generation of AI-assisted hyperspectral imaging systems for surgical guidance using quantitative fluorescence. The postholder, based within the Department of Surgical & Interventional Engineering at King’s College London, will play a key role in NeuroPPEye a collaborative project with King’s College Hospital and work closely with the project’s industrial collaborator Hypervision Surgical, a King’s spin-out company. A clinical neurosurgery study underpins this collaboration. The successful candidate will work on the resulting neurosurgical data as well as controlled phantom data. They will also have the opportunity to provide insight on how to best acquire prospective data.

Brain tumour surgery involves removing as much of the tumour as safely as possible. However, even with the best hands and the most modern technology currently available, it is often not possible to reliably identify tumour during surgery. Hyperspectral imaging (HSI) has the potential to enhance the surgeon’s vision to reliably identify tumour and healthy brain structures through the use of quantitative fluorescence. HSI data is nonetheless complex, high-dimensional and thus requires advanced computer-processing before it can be visualised and interpreted by the surgical team.

Key activities relate to the processing of hyperspectral imaging, from low-level image reconstruction to deep-learning based tissue property estimation and semantic segmentation of brain and tumour tissue. The recruited individual will complement our multidisciplinary team and undertake research on image computing, machine learning, and artificial intelligence for computer-assisted interventions.

The post involves close and active collaboration with researchers, engineers, and clinicians. Working with established platforms and building on the software already present within our teams is of paramount importance to ensure project cohesion and strong links with the members of the team. The close collaboration between King’s College London, King’s College Hospital and Hypervision Surgical Ltd will ensure a fast-tracked conversion from the research development into products achieving accelerated patient and public benefit.

The successful candidate will design, develop, and translate modular software components for hyperspectral image computing, machine learning and visualisation. They will also interface those with existing software and hardware components including the operative surgical microscope. Specifically, the candidate will develop algorithms for quantitative fluorescence estimation, super-resolution, tissue differentiation and informative visualisation. The candidate will work closely with the rest of the team to correlate the result of their work with rich clinical data (e.g. surgical microscopy, histopathology) and validate the overall imaging system.

This post will be offered on an a fixed-term contract until March 2025. This can be a full-time or part-time post, 50-100% full-time equivalent.

Key responsibilities

Develop, validate and integrate algorithms for interventional hyperspectral imaging
Contribute to project management tasks
For Research Fellows: Contribute to junior research member supervision
Maintain accurate and up-to date technical and user documentation of the delivered software
Contribute to the dissemination of the research through publications, open-source software and public engagement activities

The above list of responsibilities may not be exhaustive, and the post holder will be required to undertake such tasks and responsibilities as may reasonably be expected within the scope and grading of the post.

Skills, knowledge, and experience

Essential criteria

PhD or equivalent industrial experience in Computer Assisted Intervention or a closely related field
Good knowledge of machine learning and computer vision algorithms
Solid knowledge of and experience using the Python programming languages
Experience with scientific software packages such as PyTorch, Pandas, SciPy, NumPy, SciKit’s, OpenCV, etc.
A demonstrable record of publications in top-ranked peer-reviewed conference proceedings and scientific journals in the field
Ability to work with a variety of people
For Research Fellows: Experience in planning research projects and coordinating the work of other staff

Desirable criteria

Understanding of image acquisition and hardware components relevant to real-time data acquisition and processing of computational biophotonics imaging
Experience in real-time computing optimization (Parallel computing, GPGPU programming, deep learning inference engines such as TensorRT, etc.).

Please note that this is a PhD level role but candidates who have submitted their thesis and are awaiting award of their PhDs will be considered. In these circumstances the appointment will be made at Grade 5, spine point 30 with the title of Research Assistant. Upon confirmation of the award of the PhD, the job title will become Research Associate and the salary will increase to Grade 6.

More information about the position and how to apply here or here.

Research Associate / Research Fellow in "Trustworthy Artificial Intelligence for Surgical Imaging and Robotics"

Thu, 18 May 2023 00:00:01 +0000

Post overview:

Focus: Design principled approaches for trustworthy AI-enabled surgical assistance with a focus on real-time imaging data
Line manager: Tom Vercauteren
Salary: Grade 6 £41,386-£48,414 or Grade 7 £49,737-£55,306 per annum, including London Weighting Allowance

A view on a lumbar microdiscectomy surgery (Source: DVIDS Public Domain Archive).

Job description

We are seeking a Post-doctoral Research Associate to develop novel trustworthy artificial intelligence (AI) algorithms able to extract actionable information from surgical imaging data.

Candidates should have demonstrable experience in AI applied to imaging data in the field of surgical data science or related fields, in working with robotic-related application, and in translating research into surgical workflows. Candidates are expected to have demonstrable experience with machine learning libraries, such as PyTorch, and robotics frameworks such as ROS. Familiarity with software engineering, version control software, and experience working within a multi-developer team is desirable.

The post involves close and active collaboration with researchers, engineers and clinicians. The post-holder will take part in the FAROS European Project, the Wellcome NeuroPPEye project and contribute to the Wellcome / EPSRC Centre for Medical Engineering - Trustworthy AI for Sensory-rich Surgical Robotics research pillar. Working with established platforms and building on the software and mechatronics infrastructure already present within our teams if of paramount importance to ensure project cohesion and strong links with the members of the consortium.

The successful candidate is expected to disseminate their research through presenting at scientific conferences, publishing in peer-reviewed journals, and providing open-source software tools. Candidates are expected to have a strong track record (for their career stage) of scientific publications in machine learning and/or computer-assisted intervention-related journals or equivalent conference publications. Candidates should have strong written and oral presentation skills.

The successful candidate will be based in the Department of Surgical and Interventional Engineering reporting to Prof Tom Vercauteren. The successful candidate will have the ability to collaborate with other researchers, supervise junior researchers (for Research Fellows), attend regular seminars, and apply their algorithms to other related clinical problems.

This post will be offered on an a fixed-term contract for up to 21 months until March 2025.

Key responsibilities

Develop, validate and integrate trustworthy algorithms for computer-assisted intervention (CAI)
Contribute to project management
For Research Fellows: Contribute to junior research member supervision
Maintain accurate and up-to date technical and user documentation of the delivered software
Contribute to the dissemination of the research through publications, open-source software and public engagement activities

Skills, knowledge, and experience

Essential criteria

PhD or equivalent industrial experience in Computer Assisted Intervention or a closely related field
Good knowledge of machine learning and computer vision algorithms
Solid knowledge of and experience using the Python programming language
Experience with scientific software packages such as PyTorch, Pandas, SciPy, NumPy, SciKit’s, OpenCV, ROS2, etc.
A demonstrable record of publications in top-ranked peer-reviewed conference proceedings and scientific journals in the field
Ability to work with a variety of people
For Research Fellows: Experience in planning research projects and coordinating the work of other staff

Desirable criteria

Understanding of image acquisition and hardware components relevant to real-time data acquisition and processing from existing and medical devices including stereo cameras, force sensors, and robot encoders.
Solid knowledge of and experience using the C++ programming language

Please note that this is a PhD level role but candidates who have submitted their thesis and are awaiting award of their PhDs will be considered. In these circumstances the appointment will be made at Grade 5, spine point 30 with the title of Research Assistant. Upon confirmation of the award of the PhD, the job title will become Research Associate and the salary will increase to Grade 6.

More information about the position and how to apply here or here.

Presentation Video for OpTaS

Mon, 08 May 2023 09:00:00 +0000

This video presents work lead by Christopher E. Mower. OpTaS is an OPtimization-based TAsk Specification library for trajectory optimization and model predictive control. This work will be presented at the 2023 IEEE International Conference on Robotics and Automation (ICRA).

Our crossMoDA challenge to be held MICCAI 2023 is now live!

Sat, 15 Apr 2023 00:00:01 +0000

CAI4CAI members and alumni are leading the organization of the new edition of the cross-modality Domain Adaptation challenge (crossMoDA) for medical image segmentation Challenge, which will runs as an official challenge during the Medical Image Computing and Computer Assisted Interventions (MICCAI) 2023 conference.

Task: Intra- and extra-meatal vestibular schwannoma and cochlea segmentation.

Domain Adaptation (DA) has recently raised strong interests in the medical imaging community. By encouraging algorithms to be robust to unseen situations or different input data domains, Domain Adaptation improves the applicability of machine learning approaches to various clinical settings. While a large variety of DA techniques has been proposed, most of these techniques have been validated either on private datasets or on small publicly available datasets. Moreover, these datasets mostly address single-class problems. To tackle these limitations, the crossMoDA challenge introduced the first large and multi-class dataset for unsupervised cross-modality Domain Adaptation.

Compared to the previous crossMoDA instance, which made use of multi-institutional data acquired in controlled conditions for radiosurgery planning and focused on a 2 class segmentation task (tumour and cochlea), the 2023 edition extends the segmentation task by including multi-institutional, heterogenous data acquired for routine surveillance purposes and introduces a sub-segmentation for the tumour (intra- and extra-meatal components) thereby leading to a 3 class problem.

More information about the challenge here.

Hypervision Surgical awarded Cutlers' Surgical Prize for HyperSnap hyperspectral imaging system

Thu, 09 Mar 2023 00:00:01 +0000

The four co-founders of Hypervison Surgical, a King’s spin-out company, have been awarded the Cutlers’ Surgical Prize for outstanding work in the field of instrumentation, innovation and technical development.

Hypervision Surgical receives the Cutlers’ Surgical Prize.

The Cutlers’ Surgical Prize is one of the most prestigious annual prizes for original innovation in the design or application of surgical instruments, equipment or practice to improve the health and recovery of surgical patients.

The prize was awarded for the team’s work on HyperSnap, a hyperspectral imaging system designed to improve surgical precision and patient safety during surgery.

The HyperSnap system is a compact, lightweight, Artificial Intelligence (AI)-powered hyperspectral imaging (HIS) camera system that enables contact-free and contrast-agent-free tissue characterisation to provide real-time surgical guidance.

I am proud and honoured to see Hypervision Surgical recognized by this highly prestigious award. Hypervision Surgical is one of the first spin-outs from our St Thomas’ MedTech Hub, and to see how it is already garnering awards of this calibre is a testament to the ecosystem we have developed at King’s.
- Professor Sebastien Ourselin, Head of School of Biomedical Engineering & Imaging Sciences (BMEIS) and Non-Executive Director, Hypervision Surgical

At Hypervision Surgical, we are deeply committed to driving advancements in surgery through cutting-edge technology and advanced surgical vision. Winning this prestigious prize is a testament to our unwavering dedication to improving patient outcomes and transforming the future of surgery. We are thrilled and honoured to receive this recognition, and we remain fully committed to pushing the boundaries of surgery through advanced surgical vision.
- Dr Michael Ebner, CEO and Co-Founder, Hypervision Surgical

Hypervision Surgical were recommended for the prize by The Royal College of Surgeons, who make recommendations every year to The Worshipful Company of Cutlers.

The Worshipful Company of Cutlers has a long connection with the manufacture of surgical instruments and for the last one hundred years has maintained its interest by fostering apprenticeships in the surgical instrument-making industry.

Hypervision Surgical oversees the development of an interventional hyperspectral imaging system that seamlessly integrates into the surgical workflow to provide real-time tissue information during surgery currently invisible to the human eye.

Using safe light and Artificial Intelligence, their system distinguishes between tissues, such as cancerous and non-cancerous tissue, in addition to providing quantitative information for blood perfusion and tissue oxygenation.

In collaboration with King’s College London and King’s Health Partners hospitals, Hypervision Surgical’s technology is currently involved in clinical evaluation studies for colorectal and neurosurgery.

“Hyperspectral imaging captures tissue characteristics that are invisible to the naked eye. Our artificial intelligence based algorithms allows us to reveal information which will help guide surgeons and support them in making better informed decision in real-time to improve patient outcomes. It is truly an honour to receive the Cutler’s prize at the time when our technology makes the transition from the research labs to the operating theater through Hypervision Surgical, our spin-out company."
- Prof Tom Vercauteren, Interventional Image Computing at the School of Biomedical Engineering & Imaging Sciences and Chief Scientific Officer at Hypervision Surgical

“It is a great honour to be awarded the prestigious Cutlers’ Surgical Prize for Surgical Innovation. Our technology will benefit patients by supporting surgeons to perform safer, more precise surgery by providing real-time tissue characterisation that is currently invisible to the surgeon.”
- Mr Jonathan Shapey, Consultant Neurosurgeon, Senior Clinical Lecturer and Clinical Lead at Hypervision Surgical

Hypervision Surgical Founders.

The award presentation was made at a dinner at Cutler’s Hall, London, with the winning team receiving an elegant mounted Victorian silver-gilt medal and an award of £5,000.

Video for OpTaS

Mon, 06 Mar 2023 09:00:00 +0000

This video presents work lead by Christopher E. Mower. OpTaS is an OPtimization-based TAsk Specification Python library for trajectory optimization and model predictive control. The code can be found at https://github.com/cmower/optas. This work will be presented at the 2023 IEEE International Conference on Robotics and Automation (ICRA).

CAI4CAI Researcher awarded funding from the King's Global Engagement Partnership Fund

Wed, 01 Mar 2023 00:00:01 +0000

We congratulate CAI4CAI post-doctoral researcher, Christopher E. Mower, on receiving funding from the prestigious King’s Global Engagement Partnership Fund.

This award will support Chris during two Visiting Scholar positions at KU Leuven, Belgium and Balgrist University Hospital, Switzerland. During his first visit to KU Leuven, Chris will join the Robot-Assisted Surgery group with Professor Emmanuel Vander Poortenand and develop AI approaches for robotic spine surgery. Later, Chris will join the Research in Orthopedic Computer Science (ROCS) group with Professor Dr. Philipp Fürnstahl to integrate his work into realistic clinical setups.

We wish Chris good luck during these positions and thank the King’s Global Engagement Partnership Fund for its support.

Science for tomorrow's neurosurgery: Patient & Public Involvement (PPI) group - February 2023 group meeting

Fri, 10 Feb 2023 00:00:01 +0000

We are working to develop new technologies that combine a new type of camera system, referred to as hyperspectral, with Artificial Intelligence (AI) systems to reveal to neurosurgeons information that is otherwise not visible to the naked eye during surgery. Two studies are currently bringing this “hyperspectral” technology to operating theatres. The NeuroHSI study uses a hyperspectral camera attached to an external scope to show surgeons critical information on tissue blood flow and distinguishes vulnerable structures which need to be protected. The NeuroPPEye study is developing this technology adapted for surgical microscopes, to guide tumour surgery.

Live scribe of the conversations between patients and researchers.

Both studies are guided by the Science for Tomorrow’s Neurosurgery patient involvment group, which explores the complex issues of bringing artificial intelligence and new technologies into patient surgeries. The group had its third meeting on the 10th February 2023 to discuss progress on the hyperspectral research projects, how machine learning can be used to bring complex pictures into focus for surgeons and potential opportunities for future research.

Charlie Budd, a research software engineer and Jianrong Qiu, a postdoctoral research fellow held talks on the development of autofocus systems for hyperspectral imaging. Jianrong introduced some of the hardware used – including how it is tested and validated to make sure it gives surgeons the sharpest images at the correct time, plus how researchers test new equipment in the lab to make sure it’s safe before it’s introduced into surgery. The group were particularly interested in how new hardware can be used to maximise the quality of information passed to the surgical team. Charlie discussed how artificial intelligence algorithms – particularly “neural networks” can use deep learning to predict where surgeons will need to focus and give clear images more quickly and reliably than traditional camera systems. The complex issue of communicating these developments with patients before surgery was discussed, with the point raised repeatedly that it’s important to have surgeons and clinicians that can read the patient – providing accessible information that’s relevant to their decision making in the right way at the right time.

A second live scribe of the conversations between patients and researchers.

Oscar MacCormac and Matt Elliot – clinical research fellows in neurosurgery gave an update on the progress of the Neuro HSI and PPEye studies. Oscar gave an update on the provisional results of Stage 1 of Neuro HSI, including how based on early results the team have begun to investigate how to attach the new imaging systems to surgical microscopes. This could provide a stable, non-intrusive way and safe way to image delicate brain structures during surgery. The group were very interested to see it in action in future meetings - and fully supportive of integrating it into the research. Matt gave an update on the PPEye study – detailing how use of surgical tumour samples is being used to build and calibrate hyperspectral systems to determine tumour boundaries in surgery.

Matt also gave a talk on future research work looking into different ways to use hyperspectral imaging in fluorescence guided tumour surgery- specifically looking into how best to cause tumours to give off light (fluorescence) that distinguishes them from the healthy brain. He discussed different methods currently in use to make tumour tissue glow under controlled light conditions, the affect this may have on hyperspectral camera systems and some of the challenges of testing this in a controlled manner. All agreed on the importance of testing different methods in the lab before introducing them to surgery, and different methods of testing ranging from liquid models to animal experiments were discussed. There was unanimous agreement that the methods used needed to mirror real life surgical situations as closely as possible before introducing new techniques into the operating theatre – and the group supported for using animal models for this purpose when strictly necessary.

All in the group were interested to hear how the current research work is progressing and were able to guide not only the current projects but feed into new and complementary work designed to improve patient safety and outcomes in neurosurgery!

To learn more about the project, please visit our neurohsi.uk and neuroppeye.uk pages.

Video for the ROS-PyByllet Interface

Fri, 02 Dec 2022 09:00:00 +0000

This video presents work lead by Christopher E. Mower. The ROS-PyBullet Interface is a framework between the reliable contact simulator PyBullet and the Robot Operating System (ROS) with additional utilities for Human-Robot Interaction in the simulated environment. This work was presented at the Conference on Robot Learning (CoRL), 2022. The corresponding paper can be found at PMLR.

[Job] Research Coordinator - King's College Hospital NHS Foundation Trust

Thu, 01 Dec 2022 00:00:01 +0000

We are seeking a motivated research nurse/coordinator to support our NeuroHSI and NeuroPPEye project.

CRN King’s Neurosurgery research coordinator

The post will be a Band 6 level Neurosurgery affiliated research nurse/coordinator to work within the neuroscience division at KCH. This is a full-time post, initially until end of August 2023 with a view to be extended by 6 - 12 months. The successful applicant will work across several neurosurgery sub-specialities with a particular focus on neuro-oncology and translational healthcare technology in neurosurgery. The applicate will work on research and clinical trials listed in the Department of Heath national portfolio, principally involving the development and evaluation of advanced smart camera technology for use during surgery. The post holder will work under the supervision of Mr Jonathan Shapey (Senior Clinical Lecturer and Consultant Neurosurgeon), Professor Keyoumars Ashkan (Professor of Neurosurgery) and the management of Alexandra Rizos, Neuroscience Research Manager. Some experience in clinical research and knowledge of good clinical practice would be beneficial.

If you require further information, please contact Alexandra Rizos (a.rizos@nhs.net), Jonathan Shapey (jshapey@nhs.net) or Sabina Patel (spatel21@nhs.net).

King’s College Hospital NHS Foundation Trust.

Main duties of the job

The post holder will be responsible for co-ordinating an agreed portfolio of commercial and non-commercial clinical trials at King’s College Hospital, Clinical Neuroscience division. Responsibilities will include helping recruit patients to portfolio adopted non commercial and if required commercial studies, identifying patients through MDT’s/screening notes and increasing referrals through participating local hospitals; collecting data; interviewing; supporting and monitoring patients, applying clinical scales to patients in clinic and data entry. The department is currently running two observational trials evaluating the use of a new advance smart camera system for use during neurosurgery. The post holder will recruit patients to these studies and be involved in follow up and completion of relevant trial specific paperwork.

The post holder will work closely with the Study PIs, Research Manager, medical staff, theatre staff and the Clinical Nurse Specialist Teams at King’s College Hospital.

This post might involve travel between hospitals based in South London (e.g. St Thomas’ hospital) where site set up activities will be carried out and subsequent recruitment to clinical studies and trials. The post holder must be prepared to travel and must have a good knowledge of clinical trial site set up procedures.

Learn more

The full job details and application process can be found at jobs.kch.nhs.uk.

To learn more about the projects this post will support, please visit our neurohsi.uk and neuroppeye.uk pages.

Open-source package for fast generalised geodesic distance transform

Mon, 28 Nov 2022 00:00:01 +0000

Muhammad led the development of FastGeodis, an open-source package that provides efficient implementations for computing Geodesic and Euclidean distance transforms (or a mixture of both), targetting efficient utilisation of CPU and GPU hardware. This package is able to handle 2D as well as 3D data, where it achieves up to a 20x speedup on a CPU and up to a 74x speedup on a GPU as compared to an existing open-source library that uses a non-parallelisable single-thread CPU implementation. Further in-depth comparison of performance improvements is discussed in the FastGeodis documentation.

Raster scan data propagation psses in FastGeodis.

Our publication in Journal of Open Source Software details the summary, statement of need, and high-level implementation details of our FastGeodis package. The FastGeodis package is implemented using PyTorch (Paszke et al., 2019), utilising OpenMP for CPU- and CUDA for GPU-parallelisation of the algorithm. It is accessible as a Python package that can be installed across different operating systems and devices. Comprehensive documentation and a range of examples are provided for understanding the usage of the package on 2D and 3D data using CPUs or GPUs. Two- and three-dimensional examples are provided for Geodesic, Euclidean, and Signed Geodesic distance transforms as well as for computing Geodesic Symmetric Filtering (GSF), the essential first step in implementing the interactive segmentation method described in Criminisi et al. (2008). A further in-depth overview of the implemented algorithm, along with evaluation on common 2D/3D data input sizes, is provided in the FastGeodis [documentation]((https://fastgeodis.readthedocs.io/).

FastGeodis: Fast generalised geodesic distance transform

Mon, 28 Nov 2022 00:00:00 +0000

PhD opportunity [February 2024 start] on "Incorporating Expert-consistent Spatial Structure Relationships in Learning-based Brain Parcellation"

Tue, 15 Nov 2022 00:00:01 +0000

Project overview:

Title: Incorporating Expert-consistent Spatial Structure Relationships in Learning-based Brain Parcellation
First supervisor: Tom Vercauteren
Second supervisor: Rachel Sparks
Clinical Supervisor: Jonathan Shapey
Start date: February 2024

Human-AI trust can be defined as the belief that the AI system will satisfy a set of contracts of trust. This project will establish contracts of trust about the spatial relationships across brain structures.

Aim of the PhD Project

Develop trustworthy deep learning-based brain segmentation/parcellation.
Formalise trustworthiness as contracts on spatial relationships between labels that the algorithm must fulfil.
Establish mathematical/algorithmic frameworks to guarantee that the proposed segmentation/parcellation respect the contracts of trust.
Implement, validate, and disseminate the proposed algorithms using open-access datasets.

Project summary

Automated segmentation and labelling of brain structures from medical images, in particular Magnetic Resonance Imaging (MRI), plays an important role in many applications ranging from surgical planning to neuroscience studies. For example, in Deep Brain Stimulation (DBS) procedures used to treat some movement disorders, segmentation of the basal ganglia and structures such as the subthalamic nucleus (STN) can help with precise targeting of the neurostimulation electrodes being implanted in the patient’s brain. Going beyond segmentation of a few discrete structures, some applications required a full brain parcellation, i.e., a partition of the entire brain into a set of non-overlapping spatial regions of anatomical or functional significance. Brain parcellation have notably been used to automate the trajectory planning of multiple intracranial electrodes for epilepsy surgery or to support the assessment of brain atrophy patterns for dementia monitoring.

Deep learning based segmentation algorithms are now achieving state-of-the-art segmentation results and, in some specific cases, have demonstrated performance that can, on average, be on par with trained radiologists. Yet, automated algorithms sometimes make spectacular mistakes that violate basic expert knowledge. An example of such violation could be the automated identification of white matter spots within the external layer of the brain where only grey matter is expected. In addition to achieving suboptimal results, failures that contradict basic expert knowledge may lead to a lack of trust by the experts in the automated algorithm.

In this project, we aim to improve the trustworthiness of automated brain segmentation/parcellation algorithms by embedding expert knowledge within data-driven deep learning approaches. Trust of human experts in artificial intelligence (AI) can be defined as the belief of the human that the AI system will satisfy the criteria of a set of contracts of trust. Such a contract-based approach provides computational researchers with an actionable definition of trustworthiness. We will here focus on establishing contracts of trust that incorporate expert knowledge about the spatial relationships across the brain structures of interest.

A solid mathematical background and a keen interest in understanding expert knowledge in neuroanatomy will be essential for the successful PhD candidate to translate clinical knowledge about brain structure spatial relationships into formal contracts of trust and to propose novel mathematical and algorithmic frameworks that can respect such contracts.

As illustrated by its prominence in the European Union Artificial Intelligence Act, AI trust is a core question surrounding the deployment of AI in the clinic. This project is thus timely and of high relevance. Implementation of the research will follow open research principles to maximise the impact of the research within the multiple target audiences: computational researchers, clinical researchers, research and development engineers, etc.

This project aims at making AI trustworthiness actionable.

More information about the PhD project and how to apply on the website of the EPSRC Centre for Doctoral Training in Smart Medical Imaging.

PhD opportunity on "Artificial intelligence-driven radiosurgery planning for brain metastases"

Mon, 14 Nov 2022 00:00:01 +0000

Project overview:

Title: Artificial intelligence-driven radiosurgery planning for brain metastases
First supervisor: Jonathan Shapey
Second supervisor: Tom Vercauteren
Clinical Supervisor: Ian Paddick
Start date: October 2023

Automated detection and segmentation of brain metastases using MRI for radiosurgery planning.

Aim of the PhD Project

Implement learning-based registration to curate a spatially-normalised dataset of MR images previously used to deliver stereotactic radiosurgery to brain metastases
Develop data-driven deep learning frameworks to automatically detect and segment brain metastases while allowing for interactive corrections
Develop imaging biomarkers to predict tumour response and behaviour following treatment

Project summary

Approximately 25,000 patients are diagnosed with a brain tumour every year in the UK. Brain metastases affect up to 40% of patients with extracranial primary cancer. Furthermore, although there are presently no reliable data, metastatic brain tumours are thought to outnumber primary malignant brain tumours by at least 3:1. Patients with brain metastases require individualized patient management and may include surgery, stereotactic radiosurgery, fractionated radiotherapy and chemotherapy, either alone or in combination.

Brain metastases most commonly occur in patients with lung, breast, kidney, melanoma or bowel cancer. Surgery, stereotactic radiosurgery (SRS), and whole-brain radiotherapy (WBRT) continue to be the mainstay of treatment for brain metastasis. In particular, SRS has emerged as an important modality for treating intracranial metastases. The most important criteria for choosing SRS or WBRT is the overall number of metastases. SRS is the preferred treatment modality for patients with 1-10 metastases although the number of clinicians treating multiple metastases is increasing and several centres routinely treat over 10-20+ metastases. The planning of SRS treatment is significantly impacted by the number of metastases. The presence of small micro-metastases can greatly increase the length of planning time, as a result of having to carefully scrutinise the imaging data to detect and segment all the tumours. Furthermore, lifelong radiological follow-up is required for all patients with brain metastases, even after they undergo treatment, placing an additional burden on healthcare resources. An automated segmentation tool could significantly improve clinical workflow during the planning of SRS. By using an AI segmentation tool as an initialisation step in a clinician-driven interactive process, we will improve workflow and operational efficiency.

Artificial intelligence (AI) refers to computing technologies that mimic processes associated with human intelligence. We have previously developed a fully-automated AI framework to segment a vestibular schwannoma (another type of brain tumour) from MRI achieving state-of-the-art results.

This project now aims to develop deep learning models to: 1) detect and automatically segment brain metastases using MRI while allowing for user-driven corrections; and 2) develop composite clinical and imaging biomarkers to predict tumour response and behaviour following gamma knife treatment. The dataset and learning from this project will provide the foundation of an ambitious research programme aiming at translating such tools in clinical practice by making the AI tools flexible enough to seamlessly integrate into the clinical workflow. This will require the design of interactive corrections, the provision of interpretability means, and the development of proven AI trustworthiness features.

A broad understanding of the mathematical basis of deep learning and a keen interest in developing expert knowledge in neuroanatomy, neuropathology and radiosurgical treatment will be essential for the successful PhD candidate. Implementation of the research will follow open research principles to maximise the impact of the research within the multiple target audiences: computational researchers, clinical researchers, research and development engineers, etc.

More information about the PhD project and how to apply on the website of the EPSRC Centre for Doctoral Training in Smart Medical Imaging.

PhD opportunity [February 2024 start] on "Accurate automated quantification of spine evolution — it’s about time!"

Sun, 13 Nov 2022 00:00:01 +0000

Project overview:

Title: Accurate automated quantification of spine evolution — it’s about time!
First supervisor: Marc Modat
Second supervisor: Tom Vercauteren
Clinical Supervisor: Amanda Isaac
Start date: February 2024

Aiming at characterising and quantifying the changes occurring in an individual’s spine over time.

Aim of the PhD Project

Complement radiological expertise with automated analysis of longitudinal spine changes
Development of longitudinal registration algorithms to align spine images from different imaging modalities
Development of processing tools robust to the presence of metal artefact and various fields of view
Extraction of imaging biomarkers of spine degeneration.

Project summary

Back pain presents because of a wide range of conditions in the spine and is often multifactorial. Spine appearances also change after surgery, and postoperative changes depend on the specific interventions offered to patients and human factors such as healing and mechanical adaptations, which are also unique to each patient. When reviewing medical images for diagnostic, monitoring or prognosis purpose, radiologists are required to evaluate multiple structures, including bone, muscles, and nerves, as well as the surrounding soft tissues and any instrumentation used in surgery. They must use their expertise to assess how each structure has evolved over time visually. Such readings are, therefore, both time-consuming and require dedicated expertise, which is limited to large regional spine centres throughout the UK.

This project aims to develop a tool to characterise and quantify the changes occurring in an individual’s spine over time. It will take advantage of the many medical images, mostly computer tomography scans and magnetic resonance images, that are acquired per patient as they go through several surgical procedures. Imaging biomarkers will then be extracted and, used in conjunction with other information about the patient, will assist clinical teams in providing improved care to their patients.

A large database of images from patients who underwent surgery at Guys’ and St Thomas’ Trust hospital between 2012 and 2022 will be used to develop, train, and validate machine learning models capable of tracking the changes in a patient’s spine. These models will be required to perform well in the presence of metal artefact induced by implants. They will also need to cope with the different imaging modalities and the various fields of view of the acquired images, where some cover the whole spine and others only a subset. A dedicated geometrical model must also be tailored to the geometry of the spine, including articulated rigid structures, the vertebrae, surrounded by soft tissues such as discs, muscles, fat or nerves.

Additionally, a better understanding of surgery-related changes will provide valuable information to guide and assess interventions. Faster and more objective demonstration of changes in spine depicted on imaging may also lead to more efficient diagnostic pathways, which are patient-specific at a more value-based, time and cost-efficient model to an overburdened limited diagnostic national service.

The candidate should have a background in one of the following: biomedical engineering, applied mathematics/physics or computer science. The candidate should also have a keen interest in medical image processing, advanced machine learning and modelling.

More information about the PhD project and how to apply on the website of the EPSRC Centre for Doctoral Training in Smart Medical Imaging.

PhD opportunity [February 2024 start] on "Physically-informed learning-based beamforming for multi-transducer ultrasound imaging"

Thu, 03 Nov 2022 00:00:01 +0000

Project overview:

Title: Physically-informed learning-based beamforming for multi-transducer ultrasound imaging
First supervisor: Laura Maria Peralta Pereira
Second supervisor: Tom Vercauteren
Clinical Supervisor: Dean Huang
Start date: February 2024

Multiple transducer delay-and-sum beamforming scheme (Peralta et al, 2019). It requires accurate transducer geometry to calculate the time-of-flight between each element and focal point, and apply proper time delays to each radio-frequency channel.

Aim of the PhD Project

Pursue sparse solutions to handle the channel count required to coherently operate multiple ultrasound transducer and design and implement machine learning strategies to avoid the sparsity-related artefacts in the images.
Develop advanced beamforming techniques using machine learning approaches informed by ultrasound physics to address the concern of the flexible geometry in a multi-transducer imaging system and achieve unprecedented image quality.
Explore the application of the techniques on healthy volunteers.

Project description

Medical Ultrasound (US) is a low-cost imaging method that is long-established and widely used for screening, diagnosis, therapy monitoring, and guidance of interventional procedures. However, the usefulness of conventional US systems is limited by physical constraints mainly imposed by the small size of the handheld probe that lead to low-resolution images with a restricted field of view and view-dependent artefacts.

Currently, new opportunities to improve US imaging are emerging with high throughput US systems allowing full control over a high number of channels [1, 2]. This presents the possibility of improving US diagnostic capability by using a hyper-aperture, i.e. a larger aperture made up of lots of conventional transducers flexibly linked together. Hyper-aperture can achieve a large overall area and yet remain flexible to conform to the patient’s body [3, 4]. However, due to its flexible geometry and large discontinuous overall aperture, such a system will challenge established US beamforming approaches designed for small and continuous apertures. This means that existing and established beamforming methods are likely to be inefficient and sub-optimal for this new approach. In addition, how to leverage the richer US data offered by a hyper-aperture system and effectively process it to enhance US diagnostic capability is a new and open question to address. To maximize its potential, new approaches of US excitation, transmit/receive sequences, and processing are needed to deal with the challenges associated with sparsity and the flexible and large aperture while keeping the channel count within feasible bounds.

This project aims to, first, explore these issues, second propose feasible transmit/receive sequences, and third develop novel acquisition strategies and smart end-to-end beamforming methods to maximize the potential of a hyper-aperture. This will include a synergistic combination of US physics, sparse array approaches, and AI. Finally, these methods will be taken in vivo and the new imaging approach will be tested in a first pilot study on healthy volunteers.

The project is suited to students who have engineering/physics/mathematics background, and who would like to develop in-depth knowledge and experimental skills in signal processing and medical ultrasound. They will be expected to use and develop their programming skills through the project.

The aim of the project is to develop advanced beamforming techniques to maximise the potential of a hyper-aperture made up of multiple ultrasound transducers. Artificial Intelligent techniques along with ultrasound physics will be used to design acquisition and reconstruction strategies that will lead to superior quality images.

References

Boni, E., Bassi, L., Dallai, A., Guidi, F., Meacci, V., Ramalli, A., … & Tortoli, P. (2016). ULA-OP 256: A 256-channel open scanner for development and real-time implementation of new ultrasound methods. IEEE transactions on ultrasonics, ferroelectrics, and frequency control, 63(10), 1488-1495.
Mazierli, D., Ramalli, A., Boni, E., Guidi, F., & Tortoli, P. (2021). Architecture for an ultrasound advanced open platform with an arbitrary number of independent channels. IEEE Transactions on Biomedical Circuits and Systems, 15(3), 486-496.
Peralta, L., Gomez, A., Luan, Y., Kim, B. H., Hajnal, J. V., & Eckersley, R. J. (2019a). Coherent multi-transducer ultrasound imaging. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 66(8), 1316-1330.
Peralta, L., Ramalli, A., Reinwald, M., Eckersley, R. J., & Hajnal, J. V. (2020b). Impact of Aperture, Depth, and Acoustic Clutter on the Performance of Coherent Multi-Transducer Ultrasound Imaging. Applied Sciences, 10(21), 7655.

More information about the PhD project and how to apply on the website of the EPSRC Centre for Doctoral Training in Smart Medical Imaging.

Hospital of the Future heralded a success at New Scientist Live 2022 with strong contributions from CAI4CAI

Sat, 22 Oct 2022 00:00:01 +0000

Thousands of guests visited the futuristic medical technologies stand The Hospital of The Future by the School of Biomedical Engineering & Imaging Sciences, at this year’s New Scientist Live Festival. The exhibit was heralded a success by academics, clinicians, scientists and all interested in the future of healthcare engineering.

The Hospital of the Future by the School of Biomedical Engineering & Imaging Sciences.

At the top of the 18mx10m stand, was Superhuman Vision for Surgery led by Jonathan Shapey with by his team of clinical students and researchers. The three-part station involved perfusion imaging, laparoscopic surgery and fluorescence imaging. The novel technologies were based on work by School spin-out Hypervision Surgical.

Read the full story here.

Open-access Spatio-temporal Atlas of the Developing Fetal Brain with Spina Bifida Aperta

Wed, 19 Oct 2022 00:00:01 +0000

Lucas led on the development of the first fetal brain atlas for spina bifida aperta (SBA). This first-time atlas will allow researchers to perform measurements of the brain’s anatomy and to study its development in a large population of unborn babies with SBA

A Spatio-temporal Atlas of the Developing Fetal Brain with Spina Bifida Aperta.

Our publication in Open Research Europe details the methodology and software tools behind our open access atlas of the developing brain in fetuses with spina bifida aperta (SBA) between 21 weeks and 34 weeks of gestation. This first-time atlas will allow researchers to perform measurements of the brain’s anatomy and to study its development in a large population of unborn babies with SBA.

Brain atlases are used to study common trends and variations in the brain anatomy of a population. They provide a model of a population of brain MRIs that represents the average brain anatomy of a population and allow the comparison of measurements in cohort studies.

The unprecedented amount of data collected in this study for such a rare population, played a key role in enabling the team to create this condition-specific fetal atlas.

A total of 90 high resolution 3D Magnetic Resonance Images (MRI) scans taken of fetuses with SBA in the womb were collected, analysed and processed to develop the atlas. Developed by Dr Lucas Fidon, the atlas can allow researchers to perform more accurate measurements in fetuses with SBA.

The effect of SBA on the development of the fetal brain is complex and is not yet fully understood. Developing an understanding of SBA is fundamental to improving diagnosis and management for babies born with this condition.

However, current fetal brain atlases only correspond to normal fetal brain development.

The researchers selected the 21–34 week period of the development of the fetal brain in SBA as surgery performed while the baby is still in the womb is currently completed prior to 26 weeks of gestation.

The proposed atlas could therefore support research on the effect of the surgery on the fetal brain anatomy.

We expect that condition-specific atlases will be created for other developmental diseases in the future. As our atlas is open source, it will enable researchers outside of our group to extend the scope of our data set. – Lead researcher, Dr Lucas Fidon, School of Biomedical Engineering & Imaging Sciences

SBA is one of the most common congenital malformations. Approximately five per 10,000 babies born in Europe suffer from SBA.

SBA is a birth defect that occurs when the spinal column of the fetus fails to close during the first month of pregnancy.

It can impact the development of the fetal brain, resulting in lifelong disabilities such as cognitive impairment, difficulties with mobility, and a reduced life expectancy.

After birth and later in life, children and adults with SBA are known to have also smaller hippocampus (the part of the brain responsible for learning and memory) abnormal cortical thickness and gyrification and smaller deep grey matter volume and total brain volume. The latter could be negatively associated with cognitive and motor functions.

In a small pilot study, it has been observed that fetal brain volume and shape is different after spina bifida repair compared to controls.

In their follow-up work the researchers are developing a deep learning algorithm for the automatic segmentation of fetal brain MRI.

Image segmentation is one of the most important tasks in medical image analysis and is often the first step to derive anatomical biomarkers.

More information here and in the scientific publication.

A spatio-temporal atlas of the developing fetal brain with spina bifida aperta

Wed, 19 Oct 2022 00:00:00 +0000

Navodini W. and Muhammed A. won the second place of WIM-WILL Competition at MICCAI 2022

Mon, 17 Oct 2022 00:00:01 +0000

WiM-WILL is a digital platform that provides MICCAI members to share their career pathways to the outside world in parallel to MICCAI conference. Muhammad Asad (interviewer) and Navodini Wijethilake (interviewee) from our lab group participated in this competition this year and secured the second place. Their interview was focused on overcoming challenges in research as a student. The link to the complete interview is available below and on youtube.

Announcing moment of winners at MICCAI 2022 Award Ceremony.

In MICCAI, Navodini also presented her work on ‘Boundary Distance Loss for Intra-/Extra-meatal Segmentation of Vestibular Schwannoma’ at the 5th workshop on Machine Learning in Clinical Neuroimaging (MLCN). In this work, we propose a staged approach, with the first stage performing the whole tumour segmentation and the second stage performing the intra-/extra-meatal segmentation using the T2 MRI along with the mask obtained from the first stage. To improve on the accuracy of the predicted meatal boundary, we introduce a task-specific loss which we call Boundary Distance Loss.

PhD opportunity on "Artificial intelligence-driven management of brain tumours"

Sun, 16 Oct 2022 00:00:01 +0000

Post overview:

Title: Artificial intelligence-driven management of brain tumours
Project ID: BE-MI2023_15
First supervisor: Jonathan Shapey
Second supervisor: Tom Vercauteren
Start date: October 2023
Application deadline: Wednesday 9 November 2022, 1:00 PM (GMT)

Exemplar cases and task for AI-driven management of brain tumours.

Project description

Approximately 25,000 patients are diagnosed with a brain tumour every year in the UK. Meningiomas and pituitary adenomas are the first and third most common primary tumour, accounting for over 50% of all primary brain tumours. Brain metastases affect up to 40% of patients with extracranial primary cancer.

Patients with brain tumours require individualized patient management. The automated detection and segmentation of brain tumours could help personalise and standardise patient management and significantly improve clinical workflow.

We have previously developed a fully-automated AI framework to segment a vestibular schwannoma (a type of brain tumour) from MRI achieving state-of-the-art results. This project aims to develop deep learning models to: 1) detect and automatically segment various non-glial brain tumours (meningioma, pituitary adenoma, brain metastases) using MRI; and 2) develop compositive clinical and imaging biomarkers to predict tumour growth and behaviour (pituitary adenoma, meningioma).

In Year 1 the student will learn modern image-registration and domain-adaptation methods to curate a complete dataset of registered magnetic resonance images using the patient’s available sequences. Year 2 will include investigation of radiomic data extraction and automated tumour classification methods using machine learning. In Year 3, the student will develop data-driven deep learning frameworks combining longitudinal clinical and imaging data to create radiomic biomarkers to predict tumour growth and behaviour. The workplan may be adjusted to accommodate the student’s background.

The student will receive specialised training in modern deep learning approaches and will acquire an in-depth understanding of how to apply these methods in medical imaging analysis.

One representative publication from each co-supervisor

Shapey J, Wang G, Dorent R, Dimitriadis A, Li W, Paddick I, Kitchen N, Bisdas S, Saeed SR, Ourselin S, Bradford R, Vercauteren T. An artificial intelligence framework for automatic segmentation and volumetry of vestibular schwannoma from contrast-enhanced T1-weighted and high-resolution T2-weighted MRI. J Neurosurg 6:1-9 (2019) doi:10.3171/2019.9.JNS191949
Dorent R, Booth T, Li W, Sudre CH, Kafiabadi S, Cardoso J, Ourselin S, Vercauteren T. Learning joint segmentation of tissues and brain lesions from task-specific hetero-modal domain-shifted datasets. Med Image Anal. 2021 Jan;67:101862. doi: 10.1016/j.media.2020.101862. Epub 2020 Oct 9. PMID: 33129151; PMCID: PMC7116853. doi:10.1016/j.media.2020.101862

Eligibility information

This project is offered as part of the MRC Doctoral Training Partnership (DTP) in Biomedical Sciences.

Home, EU and international students are all eligible to apply. The MRC DTP has funding available for up to 30% of its cohort for EU and international applicants. That is approximately 10 studentship places per year. All MRC DTP studentships are fully funded; including tuition fees, stipend and bench fee. If you are unsure if you meet the eligibility criteria, you can read the FAQ and contact the MRC DTP Admissions Team for clarification.

More information about the position here and how to apply here.

Science for tomorrow's neurosurgery: Patient & Public Involvement (PPI) group - September 2022 group meeting

Wed, 21 Sep 2022 00:00:01 +0000

Live scribe of the conversations between patients and researchers.

Both studies are guided by the Science for Tomorrow’s Neurosurgery patient involvment group, which explores the complex issues of bringing artificial intelligence and new technologies into patient surgeries. The group met on the 21st of September 2022 to discuss how information about new technologies can best be discussed with patients in their treatment journey – and the difficult question about how artificial intelligence systems should be validated for patient use.

Rick Debnath, a post doctoral researcher in machine learning, gave a presentation on how machine learning systems are “trained” and “tested” using known information from previous patients. This facilitated a lively discussion on consent for research, particularly relating to the use of anonymised information to train computing systems. The group felt given the importance of the work, a system of opt out consent could be considered – allowing information to be used in research, but making sure ultimate control remains with the patient!

A second live scribe of the conversations between patients and researchers.

The team went on to discuss how new systems are validated for use, and how to decide when it is ready for use guiding surgeries. All involved felt that it was important that new imaging and AI systems acted as additional tools but that the operating surgeon remained the ultimate decision maker. The group emphasised preparing for brain surgery is a stressful, difficult time – and most felt that at that stage the decision on introducing new technologies should be with the surgeon. Raising the interesting point of how hyperspectral data can be fed into the neurosurgeons’ toolkit as smoothly as possible – and how to roll out training to make beneficial technologies available to as many surgeons as possible.

All were interested to review the progress being made in the NeuroHSI study and were able to guide further study developments, reviewing new protocols to bring hyperspectral cameras even closer to neurosurgeries using endoscopes that enter safe “approach corridors” in the brain.

If you have been diagnosed with, or treated for a brain tumour and would be interested in helping to guide cutting edge research in neurosurgery and AI please contact the NeuroHSI team or NeuroPPEyeI team to enquire about space in the Science for Tomorrow’s Neurosurgery PPI group.

To learn more about the project, please visit our neurohsi.uk and neuroppeye.uk pages.

Yijing Xie awarded Royal Academy of Engineering Research Fellowship

Wed, 31 Aug 2022 00:00:01 +0000

Yijing has been awarded the prestigious Royal Academy of Engineering Research Fellowship for her research in the development of tools to help neurosurgeons during surgery.

Dr Yijing Xie, Research Fellow & recipient of the Royal Academy of Engineering Research Fellowship.

Dr Xie says that currently there is a lack of effective ways to assess brain functions in real-time, particularly during brain surgery. During such a procedure, the surgeon must remove all cancerous tissue while preserving surrounding brain tissue and regions that serve important functions.

The current ‘gold standard’ is to perform functional brain mapping while the patient is awake and performing a specific task, but even a highly experienced surgeon using the latest surgical technologies will experience difficulties including prolonged surgery time and higher patient risk.

Dr Xie is looking into multi-spectral imaging which current evidence suggests that it can detect tumor fluorescence quantitatively while also revealing brain functions. But as this is currently implemented in 2D, Dr Xie will develop a compact 3D multispectral optical imaging platform that is specialized for use in the surgical environment.

At the end of the research programme, she will aim to have a working prototype with validated performance ready for onward translation to surgical testing.

Dr Xie says the technology will be highly targeted and will also be scalable and transferable to benefit many other research fields presenting similar challenges, within and beyond biomedical research.

More information here.

NeuroPPEye

Sat, 23 Jul 2022 00:00:00 +0000

Research Associate in "Biomedical Optics - Hyperspectral Imaging"

Mon, 11 Jul 2022 00:00:01 +0000

Post overview:

Focus: Translational research in biomedical optics for hyperspectral imaging based quantitative fluorescence linked with an in-patient neurosurgery clinical study
Line manager: Tom Vercauteren
Clinical collaborator: King’s College Hospital
Industry collaborator: Hypervision Surgical
Salary: Grade 6, £38,826 - £45,649 per annum, including London Weighting Allowance

AI-assisted hyperspectral imaging systems for surgical guidance using quantitative fluorescence.

Job description

We are seeking a biomedical optics researcher to design and translate the next generation of hyperspectral imaging systems for surgical guidance using quantitative fluorescence. The postholder, based within the Department of Surgical & Interventional Engineering at King’s College London, will play a key role in a collaborative project with King’s College Hospital and work closely with the project’s industrial collaborator Hypervision Surgical, a recently founded King’s spin-out company. A clinical neurosurgery study has been set up to underpin this collaboration.

Brain tumour surgery involves removing as much of the tumour as safely as possible. However, even with the best hands and the most modern technology currently available, it is often not possible to reliably identify tumour during surgery. Hyperspectral imaging (HSI) has the potential to enhance the surgeon’s vision to reliably identify tumour and healthy brain structures through the use of quantitative fluorescence. Acquiring HSI in real-time through the surgical microscope (the workhorse of modern neurosurgical practice) nonetheless remains complex and requires careful optimisation of the imaging pipeline as well as an improved understanding of light-tissue interaction.

Key activities relate to the design and characterisation of the HSI image acquisition system to be integrated on the surgical microscope as well as the development of improved biophotonics light-tissue interaction models to achieve accurate quantitative fluorescence extraction. Phantom work will also be used for validation purposes. The recruited individual will complement our multidisciplinary team and undertake research on biophotonics for computer-assisted interventions.

The post involves close and active collaboration with AI researchers, biomedical engineers, and clinicians. The close collaboration between King’s College London, King’s College Hospital and Hypervision Surgical Ltd will ensure a fast-tracked conversion from the research development into products achieving accelerated patient and public benefit.

This post will be offered on a fixed-term contract for 18 months. This is a full-time or part-time post – 50-100% full time equivalent.

Key responsibilities

Develop hyperspectral imaging instruments
Contribute to project management tasks
Maintain accurate and up-to date technical documentation of the delivered hardware
Contribute to the dissemination of the research through publications, conferences and public engagement activities

Skills, knowledge, and experience

Essential criteria

PhD or equivalent industrial experience in optics or a closely related field
A demonstrable record of publications in peer-reviewed conference proceedings and scientific journals
Experience working on system integration tasks as part of multidisciplinary teams
Expertise in computational modelling and simulation for biomedical optics
Experience in collecting and analysing experimental data

Desirable criteria

Knowledge of and experience of software development for biophotonics
Previous experience developing a biomedical imaging system for surgery
Experience with supervision of graduate and undergraduate students

More information about the position and how to apply here or here.

Multi-label Scribbles Support in MONAI Label v0.4.0

Thu, 16 Jun 2022 00:00:01 +0000

Recent release of MONAI Label v0.4.0 extends support for multi-label scribbles interactions to enable scribbles-based interactive segmentation methods.

CAI4CAI team member Muhammad Asad contributed to the development, testing and review of features related to scribbles-based interactive segmentation in MONAI Label.

The following video shows a short demo annotating CT volume using multi-label scribbles in MONAI Label:

FAROS Integration Week at Balgrist University Hospital

Fri, 13 May 2022 00:00:01 +0000

Our team is getting ready to test FAROS technology in the operating room. From left to right: Tom, Martin, Anisha, and Matt.

The FAROS consortium had a fantastic and highly productive time working at the labs of Balgrist Campus AG and the operating room at Balgrist University Hospital this week.

FAROS technology in the hands of our surgeon collaborators.

We indeed had the opportunity to run a new consortium-wide integration week together with FAROS engineers, scientists and clinicians.

FAROS investigators in the Operating Theatre at Balgrist University Hospital.

All hands were on deck in this briefing session prior to starting an experimental procedure.

A closer look into the FAROS robotic and non-visual sensing prototype system.

PhD opportunity [June 2022 start] on "Computational approaches for quantitative fluorescence-guided neurosurgery"

Thu, 24 Mar 2022 00:00:01 +0000

Applications are invited for the fully funded 4 years full-time PhD studentship (including home tuition fees, annual stipend and consumables) starting on 1st June 2022.

Award details:

Focus: Translational research on hyperspectral-based quantitative fluorescence imaging linked with a neurosurgery clinical study
Primary supervisor: Tom Vercauteren
Secondary supervisor: Jonathan Shapey
Industry collaborator: Hypervision Surgical
Funding type: Tuition fee, stipend (Home Fee status only).
Application closing date: 11 April 2022
Start date: June 2022

AI-assisted hyperspectral imaging systems for surgical guidance using quantitative fluorescence.

Aim of the project

This project aims at enabling wide-field and real-time quantitative assessment of tumour-specific fluorescence by designing novel deep-learning-based computational algorithms. The project will leverage a compact hyperspectral imaging (HSI) system developed by Hypervision Surgical Ltd initially designed for contrast-free imaging.

The success of brain tumour surgery is largely dependent on how much of the tumour can be safely removed during surgery. Using HSI as an advanced optical imaging technique, early research results with slow benchtop HSI systems have shown that it is possible to extract quantitative information about fluorophore concentration and hence about tumour burden.

The primary hypothesis being tested in this project is that HSI-based quantitative fluorescence can be done in real-time with a device suitable for integration into the surgical workflow thanks to learning-based computational approaches.

Project details

Background. Brain tumours pose a significant public health burden, with 70,500 patients diagnosed with brain tumours in the UK each year. The success of brain tumour surgery is largely dependent on two key factors: how early the tumour is detected, and how much of the tumour can be safely removed during surgery. However, determining brain tumour from healthy brain tissue can be exceedingly difficult during surgery. To significantly improve patient outcomes, neurosurgeons need a way to reliably identify early-stage low-grade gliomas during surgery.

Recent developments in smart camera systems such as hyperspectral imaging can enhance the surgeon’s vision. Hyperspectral imaging could help neurosurgeons to detect the low levels of fluorescence generated by early-stage, low-grade gliomas and guide the removal of the tumour to improve patient outcomes. However, the data generated by hyperspectral imaging cameras is complex and requires advanced computational processing to be useful for surgical guidance.

Addressing these challenges and limitations of the previous generation of smart cameras, this project aims to develop real-time quantitative fluorescence HSI (qFHSI) algorithms that will eventually deliver a step-change in the treatment of LGG. It will involve collaborations with neurosurgeons, hardware and software engineers from KCL and the project’s industrial partner, Hypervision Surgical Ltd.

Year 1: The student will receive appropriate skills training to enable them to conduct the research project. Part of this aim will be achieved by performing a literature review on the existing quantitative fluorescence approaches. Building on previous work in the group on developing a benchtop HSI-based system for quantitative fluorescence (Xie 2017), the student will get familiar with the particularities of real-time intraoperative HSI (Ebner 2021) and adapt the existing state-of-the-art algorithms to this platform.

Year 2: While the main technical deliverable of Year 1 is a first system for real-time quantitative fluorescence using a snapshot imaging device, it would have been developed by adapting computational approaches initially designed for high-resolution spectral data acquired across a wide wavelength range. While of interest, this initial adaptation is nonetheless expected to lead to suboptimal results. In Year 2, advanced data-driven computational biophotonics approaches will allow the design of a tissue-optics model that can compensate for scattering and absorption at wavelengths not directly captured by the captured fluorescence bands. Physics-informed deep learning will be combined with experimental measurements and Monte Carlo simulations to achieve accurate and robust extrapolation. To support this goal, the student will spend a 3-month placement at Hypervision Surgical where they will get support on the physics of acquisition of the imaging system. In addition, the student will get hands-on experience with the development of algorithms and imaging systems as part of a regulated medical device development.

Year 3: Further refinement of the quantitative fluorescence model will be made in Year 3. In particular, the computational model will be extended to incorporate tissue autofluorescence and account for spatio-spectral subsampling. Integration and optimisation of the algorithm will be key to achieve real-time quantitative fluorescence measurement. Validation of the system will be performed in bespoke phantoms to assess the limitations of the developed system.

Year 4: On the technical side, the final year will focus on consolidating the previous developments and addressing the limitations identified in Year 3 whenever feasible in the time frame of the PhD. User-centred validation will be performed by designing bespoke evaluation protocols implemented in a realistic surgical environment. Finally, time is set aside for contingency planning and for the write-up of the thesis. This will ensure the student finishes their PhD within 4 years.

Further information

Informal email enquiries from interested students to the supervisor are encouraged (contact details below).

Prof. Tom Vercauteren: tom.vercauteren@kcl.ac.uk

Award value

The studentship is fully funded for 4 years. This includes home tuition fees, stipend and generous project consumables.

Stipend: Students will receive a tax-free stipend at the UKRI rate of ca £20,109 per year as a living allowance.

Research Training Support Grant (RTSG): A generous project allowance will be provided for research consumables and for attending UK and international conferences. Eligibility criteria

Prospective candidates should have a 1st or 2:1 M-level qualification in Biomedical Engineering, Physics, Engineering, Computer Science, Mathematics, or a related programme.

Preference will be given to candidates with a background conducive to multidisciplinary research and preferably programming skills.

Candidates who meet the eligibility requirements for Home Fee status will be eligible to apply for this project. Home students will be eligible for a full UKRI award, including fees and stipend, if they satisfy the UKRI criteria below, including residency requirements. To be classed as a Home student, candidates must meet the following criteria:

be a UK National (meeting residency requirements), or
have settled status, or
have pre-settled status (meeting residency requirements), or
have indefinite leave to remain or enter.

We welcome eligible applicants from any personal background, who are pleased to join diverse and friendly research groups.

Applicable level of study: Postgraduate research

Application process

A PDF copy of your CV should be uploaded to the Employment History section.
A 500-word personal statement outlining your motivation for undertaking postgraduate research should be uploaded to the Supporting statement section.
Funding information: Please choose Option 5 “I am applying for a funding award or scholarship administered by King’s College London” and under “Award Scheme Code or Name” enter MRC_TV. Failing to include this code might result in you not being considered for this funding

More information about the opportunity here or here.

Science for tomorrow's neurosurgery: Patient & Public Involvement (PPI) group

Wed, 09 Feb 2022 00:00:01 +0000

We are actively involving patients and carers to make our research on next generation neurosurgery more relevant and impactful. Early February 2022, our research scientists from King’s College London and King’s College Hospital organised a Patient and Public Involvement (PPI) meeting with support from The Brain Tumour Charity.

Live scribe of the conversations between patients and researchers.

It was a pleasure to take part in conversations between patients and carers and researchers about developments in brain tumour surgery. Valuable insights was captured into the importance of communication, trust of AI and patient priorities.

Thanks to freelance artist Jenny Leonard for the live scribe of the conversation.

A second live scribe of the conversations between patients and researchers.

To learn more about the project, please visit our neurohsi.uk page.

NeuroHSI

Mon, 07 Feb 2022 00:00:00 +0000

Congratulations to Dr. Rémi Delaunay!

Tue, 18 Jan 2022 00:00:01 +0000

A great milestone today for Rémi Delaunay who passed his PhD viva with minor corrections! His thesis is entitled “Computational ultrasound tissue characterisation for brain tumour resection”.

Thanks to Pierre Gélat and Greg Slabaugh for their role examining the thesis.

To learn more about Rémi’s PhD work (Supervised by Tom Vercauteren and Yipeng Hu) while the thesis corrections are being prepared, plese look at Rémi’s publication track record or his MICCAI video presentation:

Meet Anisha, Martin and Mengjie, CAI4CAI PhD students from the CDT SIE cohort

Sat, 15 Jan 2022 00:00:01 +0000

The Surgical & Interventional Engineering Centre for Doctoral Training delivers translational research to transform patient pathways. Meet some of our talented PhD students is this programme who are engineering better health!

Meet many CDT SIE students on the CDT youtube channel.

Meet Anisha Bahl who is working on a hyperspectral imaging system to extract data in surgical settings to provide a real-time view of critical information for surgeons.

Meet Martin Huber who is working on automating laparoscopic camera motion, a type of minimally invasive surgery.

Meet Mengjie Shie who is looking at how to apply machine and deep learning for signal processing to improve imaging performance in photo acoustic imaging.

Research Associate in "Real-time Computational Hyperspectral Imaging"

Wed, 12 Jan 2022 00:00:01 +0000

Post overview:

Focus: Translational research on real-time computing for hyperspectral imaging linked with an active neurosurgery clinical study
Line manager: Tom Vercauteren
Industry collaborator: Hypervision Surgical
Clinical collaborator: King’s College Hospital
Salary: Grade 6, £38,826 - £45,649 per annum, including London Weighting Allowance

Hyperspectral Imaging (HSI) in the operating room.

Job description

We are seeking an interventional image computing researcher to design and translate the next generation of real-time AI-assisted hyperspectral imaging systems for surgical guidance. The postholder, based within the Department of Surgical & Interventional Engineering at King’s College London, will play a key role in a collaborative project with King’s College Hospital and Hypervision Surgical, a recently founded King’s spin-out company. A clinical neurosurgery study has been set up to underpin this collaboration. The successful candidate will work on the resulting neurosurgical data as well as retrospective data. They will also have the opportunity to provide insight on how to best acquire prospective data.

Brain tumour surgery involves removing as much of the tumour as safely as possible. However, even with the best hands and the most modern technology currently available, it is often not possible to reliably identify tumour during surgery. Hyperspectral imaging (HSI) has the potential to enhance the surgeon’s vision to reliably identify tumour and healthy brain structures. HSI data is nonetheless complex, high-dimensional and thus requires advanced computer-processing before it can be visualised and interpreted by the surgical team.

Key activities relate to real-time processing of hyperspectral imaging, from low-level image reconstruction to deep-learning based tissue property estimation and semantic segmentation of brain tissue and tumour types. The recruited individual will complement our multidisciplinary team and undertake research on real-time image computing, machine learning, and artificial intelligence for computer-assisted interventions.

The successful candidate will design, develop, and translate real-time modular software components for hyperspectral image computing, machine learning and visualisation. They will also interface those with existing software and hardware components. Specifically, the candidate will develop algorithms for super-resolution, physiological parameter estimation, tissue differentiation and informative visualisation. The candidate will work closely with the rest of the team to correlate the result of their work with rich clinical data (e.g. surgical microscopy, histopathology) and validate the overall imaging system.

This post will be offered on an a fixed-term contract for 24 months in the first instance. This can be a full-time or part-time post – 50-100% full time equivalent

Key responsibilities

Develop, validate and integrate real-time algorithms for interventional hyperspectral imaging
Contribute to project management tasks
Maintain accurate and up-to date technical and user documentation of the delivered software
Contribute to the dissemination of the research through publications, open-source software and public engagement activities

Skills, knowledge, and experience

Essential criteria

Honours degree (2:1 or above) or equivalent in Mathematics, Engineering, Physics, Computer Science or related numerate discipline
PhD or equivalent industrial experience in Computer Assisted Intervention or a closely related field
Good knowledge of machine learning and computer vision algorithms
Solid knowledge of and experience using the Python and C++ programming languages
Experience with scientific software packages such as PyTorch, Pandas, SciPy, NumPy, SciKit’s, OpenCV, etc.
Experience in standard software engineering practices including version control systems and software testing methodologies
Experience working on system integration tasks
Ability to work with a variety of people

Desirable criteria

A demonstrable record of publications in peer-reviewed conference proceedings and scientific journals
Experience in real-time computing optimization (Parallel computing, GPGPU programming, deep learning inference engines such as TensorRT, etc.)
Understanding of image acquisition and hardware components relevant to real-time data acquisition and processing of computational biophotonics

More information about the position and how to apply here or here.

Research Associate in "Computational Hyperspectral Imaging"

Tue, 11 Jan 2022 00:00:01 +0000

Post overview:

Focus: Translational research on hyperspectral-based quantitative fluorescence imaging linked with a neurosurgery clinical study
Line manager: Tom Vercauteren
Clinical collaborator: King’s College Hospital
Industry collaborator: Hypervision Surgical
Salary: Grade 6, £38,826 - £45,649 per annum, including London Weighting Allowance

AI-assisted hyperspectral imaging systems for surgical guidance using quantitative fluorescence.

Job description

We are seeking an interventional image computing researcher to design and translate the next generation of AI-assisted hyperspectral imaging systems for surgical guidance using quantitative fluorescence. The postholder, based within the Department of Surgical & Interventional Engineering at King’s College London, will play a key role in a collaborative project with King’s College Hospital and work closely with the project’s industrial collaborator Hypervision Surgical, a recently founded King’s spin-out company. A clinical neurosurgery study has been set up to underpin this collaboration. The successful candidate will work on the resulting neurosurgical data as well as controlled phantom data. They will also have the opportunity to provide insight on how to best acquire prospective data.

This post will be offered on an a fixed-term contract for 3 years. This can be a full-time or part-time post, 50-100% full-time equivalent.

Key responsibilities

Develop, validate and integrate real-time algorithms for interventional hyperspectral imaging
Contribute to project management tasks
Maintain accurate and up-to date technical and user documentation of the delivered software
Contribute to the dissemination of the research through publications, open-source software and public engagement activities

Skills, knowledge, and experience

Essential criteria

Honours degree (2:1 or above) or equivalent in Mathematics, Engineering, Physics, Computer Science or related numerate discipline
PhD or equivalent industrial experience in Computer Assisted Intervention or a closely related field
Good knowledge of machine learning and computer vision algorithms
Solid knowledge of and experience using the Python and C++ programming languages
Experience with scientific software packages such as PyTorch, Pandas, SciPy, NumPy, SciKit’s, OpenCV, etc.
Experience in standard software engineering practices including version control systems and software testing methodologies
Experience working on system integration tasks

Desirable criteria

A demonstrable record of publications in peer-reviewed conference proceedings and scientific journals
Experience in real-time computing optimization (Parallel computing, GPGPU programming, deep learning inference engines such as TensorRT, etc.)
Understanding of image acquisition and hardware components relevant to real-time data acquisition and processing of computational biophotonics

Please note that this is a PhD level role but candidates who have submitted their thesis and are awaiting award of their PhDs will be considered. In these circumstances the appointment will be made at Grade 5, spine point 30 with the title of Research Assistant. Upon confirmation of the award of the PhD, the job title will become Research Associate and the salary will increase to Grade 6.

More information about the position and how to apply here or here.

Research Associate or Fellow in "Real-time AI for Surgical Robot Control"

Wed, 05 Jan 2022 00:00:01 +0000

Post overview:

Focus: Real-time learning-based processing of hyperspectral imaging and its integration in complex robotic systems
Line manager: Tom Vercauteren
Related research project: FAROS
Salary: Grade 6, £38,826 - £45,649 or Grade 7, £46,934 - £50,919 per annum, including London Weighting Allowance

Overview of the H2020 FAROS project.

Job description

We are seeking a highly motivated individual to join us and work on FAROS, a European research project dedicated to advancing Functionally Accurate RObotic Surgery, https://h2020faros.eu, in collaboration with KU Leuven, Sorbonne University, Balgrist Hospital and SpineGuard.

Functional accuracy is defined as the degree to which the functional outcome of surgery conforms to the expected value for a successful complication-free operation. FAROS aims at improving functional accuracy through embedding physical intelligence in surgical robotics.

Within the FAROS consortium, research in the Department of Surgical & Interventional Engineering within the School of Biomedical Engineering & Imaging Sciences focuses on developing technology for novel image-guided interventions. Deep machine learning is being developed to interpret intraoperative data and link assembled knowledge to autonomously execute surgical actions at operating rates far beyond human response capabilities.

Key activities relate to real-time processing of hyperspectral imaging and its integration in complex robotic systems. The recruited individual will complement our multidisciplinary team and undertake research on machine learning, artificial intelligence and visual servo control of robotically controlled surgical instruments.

The post involves close and active collaboration with researchers, engineers and clinicians. Working with established platforms and building on the software and mechatronics infrastructure already present within our teams if of paramount importance to ensure project cohesion and strong links with the members of the consortium.

The successful candidate will design, develop and translate real-time modular software components for medical data processing, machine learning and visualisation, and also interface those with existing software and hardware components. Specifically, the candidate will develop algorithms that identify and track key anatomical landmarks, and autonomously guide the robot to maximise the informativeness of the captured data (active sensing). The candidate will integrate their developed software with the consortium’s surgical robots.

This post will be offered on an a fixed-term contract for 24 months (latest end date 31/12/2023) This can be a full-time or part-time post – 50-100% full time equivalent

Key responsibilities

Develop, validate and integrate real-time algorithms for computer-assisted intervention (CAI)
Contribute to project management tasks
Maintain accurate and up-to date technical and user documentation of the delivered software
Contribute to the dissemination of the research through publications, open-source software and public engagement activities

Skills, knowledge, and experience

Grade 6

Essential criteria

Honours degree (2:1 or above) or equivalent in Mathematics, Engineering, Physics, Computer Science or related numerate discipline
PhD or equivalent industrial experience in Computer Assisted Intervention or a closely related field
Good knowledge of machine learning and computer vision algorithms
Solid knowledge of and experience using the Python programming languages
Experience with scientific software packages such as PyTorch, Pandas, SciPy, NumPy, SciKit’s, OpenCV, ROS2, OROCOS, etc.
Experience in standard software engineering practices including version control systems and software testing methodologies
Experience working on system integration tasks
Ability to work with a variety of people

Desirable criteria

A demonstrable record of publications in peer-reviewed conference proceedings and scientific journals
Project management experience
Understanding of image acquisition and hardware components relevant to real-time data acquisition and processing from existing and medical devices including stereo cameras, force sensors, and robot encoders.

Grade 7

Criteria as above plus:

Active participation in the planning of research projects;
The ability to co-ordinate the work of other staff

More information about the position and how to apply here or here.

PhD opportunity on "Joint learning of stereo vision reconstruction and hyperspectral imaging upsampling for binocular surgical guidance"

Thu, 04 Nov 2021 00:00:01 +0000

Project overview:

Title: Joint learning of stereo vision reconstruction and hyperspectral imaging upsampling for binocular surgical guidance
First supervisor: Tom Vercauteren
Second supervisor: Christos Bergeles
Clinical Supervisor: Jonathan Shapey
Start date: October 2022

Hyperspectral imaging (HSI) data can now be acquired in real-time using compact devices suitable for surgery. By combining HSI, high-resolution RGB, and multi-view geometry and by designing novel machine learning approaches, this PhD project will for the first time allow optimal display of HSI-derived information for binocular guided surgeries.

Project summary

Optimal outcomes in oncology surgery are hindered by the difficulty of differentiating between tumour and surrounding tissues during surgery. Real-time hyperspectral imaging (HSI) provides rich high-dimensional intraoperative information that has the potential to significantly improve tissue characterisation and thus benefit patient outcomes. Yet taking full advantage of HSI data in clinical indication performed under binocular guidance (e.g. microsurgery and robotic surgery) poses several methodological challenges which this project aims to address. Real-time HSI sensors are limited in the spatial resolution they can capture. This further impacts the usefulness of such HSI sensors in multi-view capture settings. In this project, we will take advantage of a stereo-vision combination with a high-resolution RGB viewpoint and a HSI viewpoint. The student will develop bespoke learning-based computational approaches to reconstruct high-quality 3D scenes combining the intuitiveness of RGB guidance and the rich semantic information extracted from HSI.

Project plan

The student will design novel computational approaches to optimally embed hyperspectral imaging in binocular surgical vison, thereby providing novel tissue characterisation capabilities. Real-time mosaic sensor based hyperspectral imaging (HSI) is a camera-based optical imaging technique that split light into multiple narrow spectral bands across the pixels of the sensors. It enables the acquisition of much richer information than the RGB information that can be seen with the naked eye or standard cameras but can only be done in real-time with a restricted spatial resolution. It also faces additional challenges related to the unknown geometry of the scene which hinders some of the quantitative spectral measurement capabilities. The project plan aims at addressing these challenges. We hereby concisely list the deliverables of the PhD and associated timelines.

Year 1

Set up of a first stereo vision rig with a high-resolution(4k) RGB camera and a mosaic sensor (1080p across channels) HSI camera separated by a baseline
Geometric calibration of the imaging system
Acquisition software to capture synchronised data stream
Implementation and validation of learning-based dense stereo RGB reconstruction using the high-resolution RGB combined with RGB-from-HSI (as reconstructed using a state-of-the-art standalone HSI super-resolution pipeline)
Phantom creation for validation purposes

Year 2

Refinement of the acquisition rig
Development of RGB-guided HSI super-resolution approaches to achieve 4K HSI spatial reconstruction (factor 2 beyond the state of the art)
Implementation of the RGB-guided upsampling to improve dense depth reconstruction
Publication

Year 3

Algorithm development to use the recovered depth information to achieve better quantitative spectral measurement
Software stack optimisation and development of end-to-end approaches to achieve (near) real-time reconstruction and spectral calibration
Publication

Year 4

Embedding of HSI-derived semantic segmentation and tissue features (e.g. perfusion) in the binocular reconstruction
Phantom refinement
Thesis write-up and publication

More information about the PhD project and how to apply on the website of the Centre for Doctoral Training in Surgical & Interventional Engineering.

PhD opportunity on "Exploiting multi-task learning for endoscopic vision in robotic surgery"

Wed, 03 Nov 2021 00:00:01 +0000

Project overview:

Title: Exploiting multi-task learning for endoscopic vision in robotic surgery
First supervisor: Miaojing Shi
Second supervisor: Tom Vercauteren
Clinical Supervisor: Asit Arora
Start date: October 2022

Overview of the project objective. Laparoscopic image courtesy of ROBUST-MIS.

Project summary

Multi-task learning is common in deep learning, where clear evidence shows that jointly learning correlated tasks can improve on individual performances. Notwithstanding, in reality, many tasks are processed independently. The reasons are manifold:

many tasks are not strongly correlated, benefits might be obtained for only one or none of the tasks in joint learning;
the scalability of learning multiple tasks is limited with the number of tasks in terms of both network optimization and practical implementation.

Having a scalable and robust multi-task learning strategy however is very meaningful and of substantial potential in many real applications, i.e. endoscopic image processing. This project studies multi-task learning in endoscopic vision for robotic surgery with a particular focus on depth and optical flow estimation, surgical instrument detection and anatomy recognition, as well as surgical action recognition. The aim is to design effective multi-task learning strategies to improve the performance on all tasks.

Project description

Multi-task learning is common in deep learning: For similar tasks like detection and segmentation, or detection and counting, this has already been achieved given the supervision of one for the other. There exist clear evidence that adding one side task would help the improvement of the main task, yet it is unclear how much benefits both tasks can get in these combinations, especially if they are not strongly correlated. For this reason, multiple tasks are normally processed independently in the current fashion. Another main obstacle lies in the scalability of learning multiple tasks together in terms of both network optimization and practical implementation. To tackle this, careful designs of the conjunction of multiple tasks are needed; novel methodologies of learning paradigms are also expected. This project is placed in the endoscopic image processing domain. We aim to develop a machine learning model with general visual intelligence capacity in robotic surgery, which includes depth and optical flow estimation, surgical instrument detection and anatomy recognition, as well as surgical action recognition. Depth and optical flow estimation as well as anatomy recognition are key requirement to develop autonomous robotic control schemes that are cognizant of the surgical scene. Automatic detection and tracking of surgical instruments from laparoscopic surgery videos further plays an important role for providing advanced surgical assistance to the clinical team given the uncertainties associated with surgical robots kinematic chains and the potential presence of tools not directly manipulated by the robot. Being able to know how many and where the instruments find its applications such as: placing informative overlays on the screen; performing augmented reality without occluding instruments; visual servoing; surgical task automation; etc. Surgical action recognition is also critical to advance autonomous robotic assistance during the procedure and for automated auditing purposes.

More information about the PhD project and how to apply on the website of the EPSRC Centre for Doctoral Training in Smart Medical Imaging.

NIHR i4i product development award received for novel imaging system

Thu, 07 Oct 2021 00:00:01 +0000

A consortium of researchers, clinicians, and industry partners have received an NIHR invention for innovation (i4i) product development award to support a clinical neurosurgery study and quality management system led product development for an intraoperative hyperspectral imaging (HSI) technology, designed to make brain tumour removal surgery safer and more complete. The award will allow the team to further develop an existing intraoperative prototype system to achieve commercial readiness with patient input and feedback from patient groups, while also informing on NHS benefits.

The project receives funding from the NIHR.

Through the course of the 3-year project involving the School of Biomedical Engineering & Imaging Sciences, King’s College Hospital and King’s College London spin-out Hypervision Surgical Ltd, the team will ultimately demonstrate clinical safety of the intraoperative HSI system for neurosurgery involving 81 patients – 18 neurovascular and 63 neuro-oncology patients – undergoing brain tumour and blood vessel surgery.

For research endeavours such as ours to lead to better care for patients, ambitious multidisciplinary programmes across academic, clinical and industry partners are needed. I am delighted to get the support from NIHR to lead this project and further advance hyperspectral imaging technology and real-time artificial intelligence so as to provide new opportunities to develop advanced surgical vision technology that can assist the surgical team in delivering more precise surgery.
- Tom Vercauteren, Professor of Interventional Image Computing at King’s College London and Chief Scientific Officer at Hypervision Surgical

We are excited about this highly collaborative project to leverage world-class research and the clinical expertise at the King’s MedTech ecosystem to fast-track conversion of our technology into a commercial medical device and push the boundaries of surgical care.
- Dr Michael Ebner, CEO and Co-Founder of Hypervision Surgical Ltd

Brain surgery operations include brain tumour removal and blood vessel procedures. Each year in the UK, approximately 70,500 patients are diagnosed with a brain tumour, 5,000 of whom undergo surgery. Approximately 1,000 patients undergo blood vessel brain surgery. For surgeons, the complication lies in reliably identifying and distinguishing between tumour, healthy brain structures as well as nerve and blood vessels. Often, tumour is left behind resulting in repeat surgery.

Further surgeries are more difficult, pose additional patient risks and lead to increased healthcare costs with often poor patient outcomes.

Newly developed camera systems such as HSI can aid surgeons in more reliably and safely identifying and removing tumour as it aims to provide currently unavailable detailed information that is otherwise invisible to the human eye during surgery.

Previously in a clinical feasibility case study and ex-vivo study, it was demonstrated that the research system seamlessly integrates into an operating workflow, with minimal disruption to surgery.

Each study provided crucial insights to develop a system suitable for acquiring high-quality data in a highly dynamic and constrained environment as it is in an operating room during surgery.

This project will further advance the prototype, refine the setup and develop key computer-processing features as HSI data is very complex and requires advanced computer-processing for its interpretation.

As part of a 2-year clinical study, the team will collect comprehensive intraoperative HSI data ranging from healthy brain structures to multiple brain tumour types which can be retrospectively correlated with histological analysis of excised tumour tissue.

This will allow capture of clinical safety data of the system in addition to the creation of paired data vital for developing machine learning-based algorithms for both tissue property and type estimations.

The study also involves patient input and feedback from patient groups, forming a Patient Advisory Group to advise on all stages of the project.

This exciting project has the potential to significantly improve the care of neurosurgery patients. We will work closely with patients and the public during this award to ensure that the project meets the needs of patients and will publicise the study’s results through a variety of patient-led activities and public events.
- Mr Jonathan Shapey, Senior Clinicial Lecturer at King’s College Hospital, Consultant Neurosurgeon at King’s College Hospital and Clinical Lead at Hypervision Surgical

We are delighted to be the recipients of the i4i award. Our intra-operative hyperspectral imaging technology fills a clinical gap in neurosurgery and allows for safer, more radical surgery for our patients with brain tumours. It will really help achieve the best outcomes for our patients.
- Professor Ashkan Keyoumars, Professor of Neurosurgery and Consultant Neurosurgeon at King’s College Hospital

By the end of the project, the technology will be ready for MHRA submission and for inclusion in prospective clinical efficacy studies.

This project is funded by the National Institute for Health Research (NIHR) i4i Product Development Award. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.

NeuroHSI

Fri, 01 Oct 2021 00:00:00 +0000

NeuroPPEye

Fri, 01 Oct 2021 00:00:00 +0000

CAI4CAI presenters at MICCAI 2021

Wed, 22 Sep 2021 00:00:01 +0000

CAI4CAI will be presenting their work at MICCAI 2021, the 24th International Conference on Medical Image Computing and Computer Assisted Intervention, held from 27 September to 1 October 2021 as a virtual event.

MICCAI 2021 runs 27 September 2021 to 1 October 2021.

Reuben Dorent will be presenting at the main conference on “Inter Extreme Points Geodesics for End-to-End Weakly Supervised Image Segmentation”:

Video on vimeo
Virtual presentation on pathable
Paper on arXiv

Lucas Fidon will be presenting at the main conference on “Label-set Loss Functions for Partial Supervision: Application to Fetal Brain 3D MRI Parcellation”:

Video on vimeo
Virtual presentation on pathable
Paper on arXiv

Lucas will also be presenting at the PIPPI workshop on “Distributionally Robust Segmentation of Abnormal Fetal Brain 3D MRI”:

Paper on arXiv
Video on vimeo

He will also be presenting his contribution to the FeTA challenge with “Partial supervision for the FeTA challenge 2021”:

Video on vimeo

Helena Williams will be presenting at the main conference on “Interactive segmentation via deep learning and B-spline explicit active surfaces”:

Video on vimeo
Virtual presentation on pathable

Peichao Li will be presenting at the AE-CAI worksop on “Deep Learning Approach for Hyperspectral Image Demosaicking, Spectral Correction and High-resolution RGB Reconstruction”:

Video on vimeo
Paper on arXiv

Martin Huber will be presenting at the AE-CAI worksop on “Deep Homography Estimation in Dynamic Surgical Scenes for Laparoscopic Camera Motion Extraction”:

Video on vimeo

Reuben Dorent, Aaron Kujawa, Samuel Joutard, Jonathan Shapey and Tom Vercauteren will be running the CrossMoDA challenge with a keynote presentation by Jonathan.

Michael Ebner and Tom Vercauteren will be participating to the MICCAI Student Board Academia & Industry Event with Michael representing Hypervision Surgical:

Event on pathable

We are all looking forward to interacting with the MICCAI community!

Join us at IEEE International Ultrasonics Symposium 2021

Thu, 09 Sep 2021 00:00:01 +0000

Join us at the IEEE International Ultrasonics Symposium where CAI4CAI members will present their work.

IEEE International Ultrasonics Symposium runs 11-16 September 2021.

Christian Baker will be presenting on “Real-Time Ultrasonic Tracking of an Intraoperative Needle Tip with Integrated Fibre-optic Hydrophone” as part of the Tissue Characterization & Real Time Imaging (AM) poster session.

Mengjie Shie will be presenting on “Enhancing photoacoustic visualisation of clinical needles with deep learning”. The video of her talk can be seen on vimeo:

Look out for the upcoming proceedings paper!

TRABIT Virtual Conference 7-10 Sept 2021

Mon, 02 Aug 2021 00:00:01 +0000

Join us for the TRABIT conference (7-10 Sept 2021) with outstanding speakers and fun networking events. Registration is free but mandatory.

TRABIT conference flyer.

You can check all the videos made by the PhD students of TRABIT to present their research projects on youtube.

In particular, here is the video that Lucas prepared about his work on “AI for fetal brain MRI analysis”:

Surgical & Interventional Engineering Summer School

Wed, 16 Jun 2021 00:00:01 +0000

Join us for the Surgical & Interventional Engineering Summer School designed for postgraduate students and research professionals from academia and industry to learn more from and interact with scientists from the international community. You can engage with world renowned experts in a diverse range of topics on Instrumentation Technologies for Personalised Treatments, and Navigation Technologies for Precision Medicine.

Surgical & Interventional Engineering Summer School at King’s College London

More information here including the course flyer and the programme.

[CfP] MedIA Special Issue on Explainable and Generalizable Deep Learning Methods for Medical Image Computing

Mon, 24 May 2021 00:00:01 +0000

CAI4CAI members are contributing to guest editing the Medical Image Analysis Special Issue on Explainable and Generalizable Deep Learning Methods for Medical Image Computing. The final submission dealine is on December 1, 2021 but papers will be reviewed as they arrive.

Cover of Medical Image Analysis.

Call for Papers

Medical Image Analysis Special Issue on Explainable and Generalizable Deep Learning Methods for Medical Image Computing

Description

Deep learning has recently revolutionized the methods used for medical image computing due to automated feature discovery and superior results. However, they have significant limitations that make clinicians skeptical on their usefulness for clinical practice. Deep learning models are essentially black boxes that do not offer explainability of their decision-making process which in turn makes it hard to debug them when necessary. The poor explainability leads to distrust from clinicians who are trained to make explainable clinical inferences. In addition, their generalizability is still limited in clinical environments due to the many different imaging protocols, large variations in image-based manifestation of pathologies and rare diseases whose related data may have not been used during training. The generalizability problem becomes even more conspicuous when a deep learning model trained on data from a given medical center is deployed to other medical centers whose data have significant variations or there is a domain shift from the training set. Consequently, there is an urgent need for innovative methodologies to improve the explainability and generalizability of deep learning methods that will enable them to be used routinely in clinical practice.

Topics of Interest

To address the limitations of deep learning methods in medical image computing, this special issue solicits novel explainable/interpretable and generalizable deep learning methods for intelligent medical image computing applications. The methods should provide novel explainable/interpretable and generalizable solutions to key application domains such as disease classification and prediction, pathology detection and segmentation, image registration and reconstruction. Topics of interest include, but not limited to the following:

Explainable/interpretable deep learning models for medical image computing
Methods that offer explainability and interpretability in deep learning models for disease characterization and classification using medical images
Learning interpretable knowledge from unannotated/annotated medical images
Explainable deep learning networks for computer-aided diagnosis from medical images
Incorporation of clinical knowledge into deep learning models for interpretable medical image analytics methods
Generalizable deep learning methods when the training medical image datasets are small
Novel data augmentation, regularization and training strategies to reduce over-fitting, especially in case of rare diseases and high-dimensional images where the training set is small
Integration of prior medical knowledge into deep learning models for medical image analysis
Human interaction to improve the robustness when dealing with rare or complex cases, such as for segmentation
Learning domain-invariant features for images from different modalities, scanning protocols and patient groups
Unsupervised, weakly supervised and semi-supervised model adaptation to new domains for medical image computing
Out-of-distribution detection methods when applying a model to novel data not previously trained on
Generalizable models for images from multi-centers, multi-modalities, multi-diseases or multi-organs
Generalizable deep learning methods in cases of images with potential domain shift

Accepted papers should demonstrate the improvements offered by their methods compared to previous deep learning methods.

Important Dates

Paper submission deadline: December 1, 2021.
1st round of Reviews: Feb 1, 2022.
Revised manuscript due: April 15, 2022.
Final decision: May 15, 2022.
Camera ready version: June 15, 2022.

Guest Editors

Guotai Wang, PhD. University of Electronic Science and Technology of China
Shaoting Zhang, PhD. The University of North Carolina at Charlotte, and SenseTime Research
Tom Vercauteren, PhD. King’s College London
Xiaolei Huang
Dimitris Metaxas, PhD. Rutgers University

Our crossMoDA challenge at MICCAI 2021 is now live!

Thu, 08 Apr 2021 00:00:01 +0000

CAI4CAI members are leading the organization of the cross-modality Domain Adaptation challenge (crossMoDA) for medical image segmentation Challenge, which runs as an official challenge during the Medical Image Computing and Computer Assisted Interventions (MICCAI) 2021 conference.

Domain Adaptation (DA) has recently raised strong interests in the medical imaging community. By encouraging algorithms to be robust to unseen situations or different input data domains, Domain Adaptation improves the applicability of machine learning approaches to various clinical settings. While a large variety of DA techniques has been proposed for image segmentation, most of these techniques have been validated either on private datasets or on small publicly available datasets. Moreover, these datasets mostly address single-class problems.

Source (contrast-enhanced T1)

Target (high resolution T2)

To tackle these limitations, the crossMoDA challenge introduces the first large and multi-class dataset for unsupervised cross-modality Domain Adaptation. The goal of the challenge is to segment two key brain structures involved in the follow-up and treatment planning of vestibular schwannoma (VS): the tumour and the cochlea. Specifically, the segmentation of the tumour and the surrounding organs at risk, such as the cochlea, is required for radiosurgery, a common VS treatment. Moreover, tumour volume measurement has also been shown to be the most accurate measurements for the evaluation of VS growth. While contrast-enhanced T1 (ceT1) Magnetic Resonance Imaging (MRI) scans are commonly used for VS segmentation, recent work has demonstrated that high-resolution T2 (hrT2) imaging could be a reliable, safer, and lower-cost alternative to ceT1. For these reasons, we propose an unsupervised cross-modality challenge (from ceT1 to hrT2) that aims to automatically perform VS and cochlea segmentation on hrT2 scans. The training source and target sets are respectively unpaired annotated ceT1 and non-annotated hrT2 scans.

This challenge will be the first medical segmentation benchmark of unsupervised DA techniques and promote the development of new unsupervised domain adaptation solutions for medical image segmentation. It will also contribute to the development of new algorithms for the follow-up and treatment planning of VS using hrT2 scans only.

More information about the challenge here.

New Partnership with Moon Surgical to Develop Machine Learning for Computer-Assisted Surgery

Tue, 23 Mar 2021 00:00:01 +0000

King’s College London, School of Biomedical Engineering & Imaging Sciences and Moon Surgical announced a new strategic partnership to develop Machine Learning applications for Computer-Assisted Surgery, which aims to strengthen surgical artificial intelligence (AI), data and analytics, and accelerate translation from King’s College London research into clinical usage.

Surgical tool image compositing for machine learning training pipelines (📖).

More information here.

Moon Surgical

Tue, 23 Mar 2021 00:00:00 +0000

Image Compositing for Segmentation of Surgical Tools without Manual Annotations

Mon, 08 Feb 2021 00:00:00 +0000

Yijing Xie receives a Wellcome/EPSRC CME Research Fellowship award

Sat, 06 Feb 2021 00:00:01 +0000

Yijing will develop a 3D functional optical imaging system for guiding brain tumour resection.

Yijing presenting her work at New Scientist Live.

She will engineer two emerging modalities, light field and multispectral imaging into a compact device, and develop novel image reconstruction algorithm to produce and display high-dimensional images. The CME fellowship will support her to carry out proof-of-concept studies, start critical new collaborations within and outside the centre. She hopes the award will act as a stepping stone to enable future long-term fellowship and grants, thus to establish an independent research programme.

More information here.

New King's Public Engagement award led by Miguel Xochicale

Thu, 07 Jan 2021 00:00:01 +0000

Miguel will collaborate with Fang-Yu Lin and Shu Wang to create activities to engage school students with ultrasound-guidance intervention and fetal medicine. In the FETUS project, they will develop interactive activities with 3D-printed fetus, placenta phantoms as well as the integreation of a simulator that explain principles of needle enhancement of an ultrasound needle tracking system.

Finding a fETus with an Ultrasound Simulator (FETUS).

The aim of the project FETUS, Finding a fETus with an Ultrasound Simulator, is to increase the awareness of young people, living in Southwark and Lambeth with the involvement of The Young Persons’ Advisory Group (YPAG), on what research engineers do and the impact they make to society. For the workshops, participants will have first-hand experience through creative and interactive activities of the current and future challenges in fetal medicine and ultrasound-guidance intervention. The activities are: a) guessing the age of a baby before birth (fetus) (b) playing with a placenta phantom and (c) playing with an ultrasound simulator to tracking needles.

The anticipated outcomes of the 12-month project are the creation of mutual benefit between young audiences and research engineers by (a) encouraging young audiences to pursue a career in STEM and (b) improving communications and project managements skills of research engineers. Experiences of the activities will be shared by writing blogs and disseminating outputs within the faculty, department and schools. In addition, such activities will help to engage with other researchers and clinicians to spark collaboration, to design future events for other type of audiences as well as the potential to publish the outcomes of the project in scientific journals.

The award of King’s Public Engagement Grant is of £1000, starting in January 2021, and it is funded by the The Centre for Doctoral Studies in King’s College London and the Wellcometrust Institutional Strategic Support Fund.

Fang-Yu, Shu and Miguel thank everyone in the Ultrasound Needle Tracking and GIFT-Surg teams for their support.

Congratulations to Dr. Luis Garcia Peraza Herrera!

Tue, 15 Dec 2020 00:00:01 +0000

A great milestone today for Luis Garcia Peraza Herrera who passed his PhD viva with minor corrections! His thesis is entitled “Deep Learning for Real-time Image Understanding in Endoscopic Vision”.

Screenshot from Luis’ online PhD viva.

Thanks to Ben Glocker and Enrico Grisan for their role examining the thesis.

To learn more about Luis’s PhD work (Supervised by Seb Ourselin and Tom Vercauteren) while the thesis corrections are being prepared, plese look at Luis’ strong publication track record.

Privacy Policy

Tue, 15 Dec 2020 00:00:00 +0000

PhD opportunity on "Intraoperative hyperspectral imaging for neurosurgery: Surgical workflow optimisation and validation"

Mon, 14 Dec 2020 00:00:01 +0000

This project will evaluate and contribute to the optimisation of a pre-CE-mark and pre-commercial HSI medical device for intraoperative surgical guidance in brain tumour surgery.

First-in-patient clinical study using the hyperspectral imaging device (spine surgery).

To deliver on this primary objective, the project will pursue the following aims:

(A1) Investigate and optimise surgical integration of the intraoperative HSI (iHSI) device into the clinical workflow
(A2) Investigate correlation between in-vivo HSI and ex-vivo histological analysis of corresponding biopsied pathological tissue
(A3) Investigate correlation between in-vivo HSI and the Magnetic Resonance Imaging (MRI) appearance of imaged tissue
(A4) Investigate the accuracy of the iHSI device to differentiate tumour, normal tissue, nerves and blood vessels from in-vivo iHSI data
(A5) Disseminate the study’s outcomes and the device’s potential to patients and the public
(A6) Knowledge exchange with the project’s external partner (Hypervision Surgical Ltd)

Patients undergoing brain tumour surgery have significantly improved outcomes and increased life expectancy if complete tumour removal is achieved. Successful surgery mandates maximal safe tumour removal but even with the most advanced current techniques, it is not possible to always reliably identify tumour and critical structures during surgery.

Hyperspectral imaging (HSI) is an advanced optical imaging technique that provides a promising solution for real-time computer-assisted tissue recognition during surgery by the safe application of light alone. HSI exploits the ability to split light into multiple narrow colour bands and can provide crucial but currently invisible information about critical biological structures during surgery.

This project will evaluate and translate a novel compact intraoperative HSI device in patients undergoing brain tumour surgery. The student will optimise the device for surgical use and will perform a clinical patient study assessing its integration into the neurosurgical workflow and quantitative image analysis.

1st Supervisor: Jonathan Shapey, King’s College London

2nd Supervisor: Tom Vercauteren, King’s College London

More information about the PhD project here.

MONAI (Medical Open Network for AI): PyTorch for medical imaging

Sat, 12 Dec 2020 00:00:00 +0000

PhD opportunity on "Semi-supervised detection and tracking of instruments for robotic surgery guidance"

Thu, 03 Dec 2020 18:44:41 +0000

This project seeks to advance the state of the art in AI-based surgical tool detection and tracking by designing novel semi-supervised and weakly-supervised approaches able to achieve robust and real-time performance.

Representative sample images of robotic surgery (left) and state-of-the-art instrument segmentation results (right). True positive (white), true negative (black), false positive (magenta), and false negative (green).

Aim of the PhD Project:

Robust, real-time detection and tracking of surgical tools
Learning from combined small-scale annotated and large-scale but non-annotated datasets of robotic surgery video footages
Advancing the state of the art in combining self-supervision, week-supervision, and semi-supervision for surgical vision tasks
Designing and validating stereo-vision based learning paradigms

1st Supervisor: Tom Vercauteren, King’s College London

2nd Supervisor: Miaojing Shi, King’s College London

Clinical Champion: Prokar Dasgupta, King’s College London

More information about the PhD project here.

PhD opportunity on "Data-driven composite biomarker development for the personalised management of vestibular schwannoma"

Thu, 03 Dec 2020 17:44:41 +0000

This project aims to develop and validate deep learning models to predict, from MRI and clinical data, which tumours are likely to grow and require treatment. Thestudent will be able to focus on designing novel radiomics analysis for vestibular schwannoma while exploiting an existing fully-automated AI tumour segmentation framework. This will enable clinicians to deliver personalised and standardised management plans to individual patients and has the potential to significantly reduce the number of required surveillance scans.

Representative automated segmentation of a vestibular schwanoma.

Vestibular schwannoma (VS) is a non-cancerous tumour arising from one of the balance nerves connecting the brain and inner ear. Approximately 1 in 1000 people will be diagnosed with a VS in their lifetime. For patients with smaller tumours, lifelong observation with repeated scans is advised but timely treatment is crucial in patients with growing tumours.

The workplan will be balanced along the methodological/translational axis depending on the student’s background. In Year 1, the student will learn modern image-registration and domain-adaptation methods to curate a complete dataset of magnetic resonance (MR) images using the patient’s available sequences. Year 2 will include the investigation of radiomic data extraction and automated tumour classification methods using machine learning. In Year 3/4, the student will develop a data-driven deep learningframework combining longitudinal clinical and imaging data to create a composite predictive biomarker of VS growth.

During their research training, the student will receive specialised training in modern deep learning approaches and will acquire an in-depth understanding of how to translate these methods in clinical applications.

1st Supervisor: Jonathan Shapey, King’s College London

2nd Supervisor: Tom Vercauteren, King’s College London

More information about the PhD project here.

CDT AE-PSI