Researchers: John Aston, Julia Carmona-Bozo, Michael Craig, Fiona Gilbert, Joana Grah (Alan Turing Institute), Yuan Huang, Elizabeth Le, Lexin Li (UC Berkeley), Ioannis Pappas, James Rudd, Carola-Bibiane Schönlieb, Mihaela van der Schaar (University of Oxford), Yu Wang, Jingjing Zou.

In UK 2,200,000 women attend mammography screening each year. In clinical practice, two-view mammograms are taken and subsequently double-read by two radiologists, as an additional 6~10% cancers are found by a second reader. However, this also leads to a large workload and challenge to the healthcare system. Moreover, given the profound intra- and inter-tumoural heterogeneity and diversity, identification of the therapeutic regimen for each patient remains unsatisfactory. Multimodal multiparametric imaging have the potential to tackle this problem but its high-dimensional nature requires appropriate statistical and computational algorithms.

This project aims to develop computer-aided detection tools with expertise from radiology, image processing, statistics and machine learning. There are two arms of the project:

(1) using data collected from the multi-Centre TOMMY trial, develop cancer detection algorithms with the consideration of patient demographic information, family history and features extracted from mammograms;

(2) using the multimodal dataset collected in Cambridge, determine the imaging phenotypes of the tumour microenvironment using MR and PET derived parameters, such hypoxia, perfusion, cellularity and fibrosis.