1st supervisor: Andy King, King’s College London
2nd supervisor: Julia Schnabel, King’s College London

Early screening of cardiovascular disease requires robust and validated tools to detect suspected disease as early as possible and direct the patient to more thorough examinations. There is a clinical and social economic need to perform such screening using low cost, easily available imaging tools such as ultrasound.   In this project we will explore how encoding prior knowledge of anatomy, motion and other information, and deriving indicators from the statistical analysis of this data can improve the early screening of heart pathologies. The main focus will be on shape, global and local motion indices, and valve flow, as acquired from B-Mode/Doppler ultrasound and cine/tagged MR. The goal of the project is to generate an atlas of such data for healthy subjects. Pathological function will be described as a progressive deviation from the healthy atlas according to some tailored statistical metrics.

The final goal is to develop a computer-aided diagnosis (CAD) system that will take measurements from ultrasound, extract the relevant indicators and relate them to an online database for which ultrasound and MR data are available. Then, based on this information a relevance score will be computed for each pathology. The most closely related cases from the database and their disease evolution will be made available to the user. These exemplary cases will show jointly the initial ultrasound images with more advanced protocols in MR and the disease evolution. The system will also learn MR/ultrasound discrepancies for improving the reliability, treating the MR exam as a hidden variable in the detection process.

The project will build on cardiac motion and deformation quantification tools developed at Philips for ultrasound and tagged MR. Also, Philips’ early experience in constructing reference databases for cardiac motion and deformation will offer an initial setup for the database and the retrieval of shape, motion and deformation indices. In addition, work at KCL (in collaboration with Imperial College London) to develop a spatiotemporal atlas of the heart will feed into the project.