Odin-vision.com

Helping UCL commercialise their quantum and AI technologies

I collaborate with UCL quantum information and machine learning groups to investigate quantum machine learning and the impact of quantum technologies.

Leading a program of research science across the R&D spectrum from early stage technology exploration to advanced prototypes deployed at clinical sites. Managing teams comprised of Siemens scientists and engineers, academic post docs and PhDs and clinical collaborators. Active projects include: Quantum Machine Learning- exploring the potential acceleration of machine learning with quantum computers. Cardiac interventional guidance- using computer vision, machine learning and augmented reality to better plan and guide cardiac procedures. Automating paediatric procedures- using machine learning, computer vision and robotics to optimize interventions and therapy for children.

Augmented Reality for Laparoscopic Liver Surgery Image guidance for liver surgery has great potential to reduce tumour margins and increase patients eligibility for surgery, however soft tissue deformation and limitations of imaging equipment makes translating this technology challenging. This project is focused on delivering an end to end solution to register pre-operative CT to intra operative laparoscopic images with cone beam CT and fluoroscopy as bridging modalities. This project use dense stereo reconstruction, feature tracking and matching, point cloud registration and biomechanical modelling. Machine Learning: Mitral Valve Detection in Cone Beam CTDeploying mitral valve clips to reduce regurgitation is a new and challenging procedure. Visualizing the position and orientation of the valve in the fluoroscopic coordinate system facilitates guidance and is performed by detecting the valve in cone beam CT. Machine learning algorithms and marginal space learning were used to detect the position of the valve and its anatomy.

X-Ray and TEE image fusion for cardiac interventions Cardiac interventions such as TAVI use fluoroscopy to guide tools and ultrasound (TEE) to visualize tissue. This information is visualized in separate coordinate systems making it challenging to interpret. This project fuses the two modalities by aligning them into a common coordinate system. The position and pose of the TEE probe is detected in the fluoroscopic image using machine learning algorithms and binary template matching.

I am based in the Department of Biomedical Engineering. My research is focused on investigating the use of machine learning and augmented reality to improve surgery and interventions.

Patient Specific Simulation and Pre-Operative Realistic Training for liver surgeryImprove surgical simulators by creating patient specific organ models. This work focused on extracting and modelling biological signals from laparoscopic video of the liver. This included extracting respiration signals form stereo laparoscopic data and model fitting.

Simultaneous Localization And Mapping for Minimally Invasive SurgeryFocused on developing image processing techniques for laparoscopic images. The main goal was to create techniques for 3D reconstruction of deforming tissue (heart/liver) and simultaneous laparoscope pose estimation. Specifically focusing on pattern classification techniques for feature tracking/matching and developing non-rigid SLAM (a.k.a. online structure-from-motion).