Automated Analysis and Display of Temporal Sequences of Mammograms
Annual rept. 1 Jul 2003-30 Jun 2004
PITTSBURGH UNIV PA
Pagination or Media Count:
Although screening for breast cancer has been effective in detecting cancers, it is not clear that the diagnostic information present in the sequences of screening exams is currently being utilized. In an attempt to unprove diagnostic accuracy, this project is adapting and integrating several novel technologies, under development in our laboratory, into a system for providing mammographers with information about changing tissue patterns, and the corresponding likelihood of malignancy, derived from temporal sequences of images. Our hypotheses are 1 Sequences of screening mammograms contain information about tissue changes that is not otherwise being exploited in the diagnosis of breast cancer and, 2 Changing tissue patterns can automatically be identified, and correlated with diagnostic questions. The main objectives of the project are to provide a system, which can be employed at the discretion of mammographers, to 1 Normalize images to facilitate comparisons 2 Apply multi-image CAD methods to identify corresponding features between images 3 Detect and classify trends in temporal sequences 4 Calculate and present various kinds of parameter images and, 5 Develop a display system for efficiently presenting sequences of exams. Our expectation is that this display will improve diagnostic performance of mammographers for these cases.
- Medicine and Medical Research