Computer Aided Detection in Mammography

Abstract

Breast screening programmes, in which mammograms are examined for signs of cancer, have been implemented in many countries. In the UK, all mammograms are reviewed by two expert medical readers. Because abnormalities are variable, subtle, and infrequent, this task is difficult and prone to human error. Computer aided detection (CAD) systems aim to improve the performance of expert readers by indicating potentially abnormal regions that may otherwise have been missed. CAD can improve performance of some readers, but often at the cost of an increase in the false positive rate due to the high number of prompts on normal regions. This thesis explores the role of CAD in mammography through a series of visual search experiments, using simulated images and targets analogous to mammography screening. First, CAD was evaluated as a second reader, where the image is first viewed unaided and then once again with CAD. This initial unaided search was found to be truncated in terms of review time and the amount of the image viewed (p

Date of Award	1 Aug 2022
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Susan Astley Theodossiadis (Supervisor) & Johan Hulleman (Supervisor)