PHD: Machine learning for non-invasive prostate cancer characterization
Applications are invited for a PhD position at the Centre for Medical Image Computing of the UCL Department of Computer Science CMIC combines methodological researchers from the Departments of CS and Medical Physics & Bioengineering with biomedical and clinical groups in the Faculty of Biomedicine. The interface of engineering and medicine, where CMIC specializes, is a unique and exciting place to do cutting-edge research.
The project will develop novel computational medical imaging based technologies for the examination of tissue microstructure in prostate cancer. The approach will use a combination of mathematical modelling for MRI analysis and machine learning methods to utilise available imaging and other clinical data to improve diagnostic features and prognosis.
Prostate cancer diagnosis still relies on traditional histology via biopsies. Each year in the UK 75000 men undergo this invasive procedure, which can have unpleasant side effects and poor sensitivity, as the biopsy may miss lesions. Additionally a negative biopsy result does not imply that the patient is free of cancer, and negative-biopsy patients are monitored in regular repeat exams. The on-going process increases risks of side effects and is expensive for the National Health Service.
This project develops non-intrusive imaging methods to provide diagnostic information as powerful as current invasive techniques. Primarily this will ameliorate patient suffering from the examination and the side effects, while assisting better clinical detection, staging and diagnosis by earlier and enhanced knowledge of what is happening at the tissue level.
The project will leverage the advances of a method developed in CMIC, called VERDICT, which uses diffusion MRI to provide an early demonstration of the power of such a model-based imaging technique for estimating cancer-tissue microstructure noninvasively. This project builds on these ideas utilizing machine learning to achieve the performance levels required to allow robust detection, grading and staging for establishing non-invasive imaging as the primary cancer diagnostic method.
The student will have the opportunity to work closely with the prostate-cancer team at UCL and UCL Hospitals (The Lancet 2017)
Prospective applicants should have an interest in machine learning and medical imaging. Expertise in deep learning is considered a plus. In all cases, applicants should have a good honours and/or Masters degree in a related discipline.
Supervisor: Dr. Eleftheria Panagiotaki
Funded for 3 years by EPSRC for EU\UK (EU applicants will need to have been living or working in the EU for 3 years to qualify for a full studentship)
Please apply via Prism