Projects in the area of artificial intelligence (Computer vision)
- Self-funded PhD students only
- Number of awards: Multiple
- Deadline: Ongoing.
- Supervisors: Contact Professor David Hogg to discuss this project further informally.
There are many opportunities for PhD research within this theme. The list below is a selection of some of these opportunities. This list is by no means definitive and you should feel free to contact staff to discuss your own interests:
Analysis and reconstruction of Terapixel Digital histopathology Volumes
Research in Leeds has lead to the development of a system for the reconstruction of large 3D volumes from digital images of microscope slides (see here or here). Such volumes can be very large (terabytes in size). There exists the opportunity for PhD research developing methods and software for analysing the data produced by such a system. Application domains involve the study of liver disease, cardiac function, engineered tissue etc. This work would be in collaboration with researchers in the Pathology and Tumor Biology group in the School of Medicine (see here).
Realtime Analysis of In-procedure X-ray images
Certain minimally invasive medical procedures such as coronary stenting are guided using real-time x-ray based imaging. There exists opportunities for projects using computer vision to analyse these real-time images in an operating theatre based context. Applications might involve tracking anatomical structures, or medical tools. The aim of such analysis would be to either aid doctors in their tasks, or optimise the imaging protocols (e.g. minimise radiation exposure).
Measurement of blood flow in 3D from C-arm X-ray in the operating theatre
In order to quantify blood flow to organs (e.g. the heart) chemical "contrast agaents" are often introduced into the body and imaging (e.g. CT or MRI) is used to detect flow as a change in the image contrast. Blood flow can be quantified by relating the change in intensity over time to flow using physics based models. Within the operating theatre the main form of imaging is realtime X-ray. This is a "projective" imaging modality - i.e. it represents flow in a volume as a single 2D "projection image".
Quantifying blood flow in 3D from multiple 2D projection images is a challenging task and offers interesting avenues for PhD research. The problem is complicated by motion due to cardiac and respitory motion. This project is in collaboration with the academic unit of medical physics and doctors at the Leeds Teaching hospitals NHS trust.
Realtime Multicamera imaging in minimally invasive medical procedures
Researchers in the school of mechanical engineering are interested in devices and robots that perform minimally invasive medical procedures inside the human body. There are many advantages of such procedures over open surgery, including reduxced trauma and complications and faster recovery times. We are interested in incorporating cameras (possibly multiple cameras) into such devices. This would allow geometric reconstruction and insertion of pre-operative information into a real-time setting (e.g. incorportating information from pre-operative MRI scans). The PhD research would be in development of the methods for anatomy reconstruction, self localisation, and virtual content insertion.
C-elegans worm tracking in 3D media from multiple cameras
C-elegans is a microscopic worm that has gathered much attention as a "model organism" for biological study. Various groups have developed techniques for tracking the location and configuration of these worms in flat media (e.g. on a petri dish) using computer vision and cameras attached to microscopes. In Leeds we've developed a system for 3D viewing of worms based on 3 cameras and manual control. PhD works would extend this to include automatic 3D worm tracking. This is complicated by the limited depth of field of cameras in the medium. Tracking could help analyse the behaviour of worms and develop theories about the mechanisms used within the organism.
Motion correction in time series medical imaging
Perfusion imaging is a type of time series imaging used to quantify (among other things) blood flow within organs such as the heart, kidneys and liver. A contrast agent is injected into the blood stream causing areas in which blood is flowing to be enhanced (increased in brightness). Analysis of the change in intensity over time can allow quantification of blood flow at a particular location. However, motion (due to respiration and cardiac motion) complicates this process as the location of a particular anatomical feature can change over time.
Motion correction by using image based registration has potential to correct for this motion and make flow calculations more robust. PhD research would involve developing methods for image based correction for a particular application domain. These could include prior information about tehse domains, and domain specific quality assurance and correction methods. This would be primarily aimed at MRI imaginag and would be in collaboration with researchers in the academic unit of medical physics.
We would normally expect applicants for postgraduate research to have achieved one of the following in a subject relevant to the proposed research:
• a minimum of a UK upper second class honours degree (2.1), or equivalent, or
• a good performance in a Masters level course.
We also recognise relevant industrial and academic experience and special circumstances in the consideration of applicants.
How to apply
Formal applications for research degree study should be made online through the university's website.
If you require any further information please contact the Graduate School Office
e: email@example.com, t: +44 (0)113 343 8000.