Marias, Konstantinos; (2002) Registration and quantitative comparison of temporal mammograms (with application to HRT data). Doctoral thesis (Ph.D), UCL (University College London).

Text Registration_and_quantitative_.pdf Download (30MB)

Abstract

In order to reduce the high mortality rate of breast cancer, most research in mammogram image analysis aims to develop a CAD (Computer Assisted Detection) system that can assist the clinician in the difficult task of "early" diagnosis. The comparison of temporal pairs of mammograms is believed by radiologists to be often crucial for diagnosing cancer, especially as breast tissue is highly variable across the population making it difficult to perform diagnosis reliably from a single mammogram. However, changes in the appearance of mammograms due to differences in compression and imaging conditions can limit the effectiveness of temporal comparison. In this thesis, a registration method that aligns temporal (or bilateral) mammograms is presented. This technique has the potential to assist the clinician to detect changes more efficiently (e.g. interval cancers in mammogram sequences, architectural distortions or microcalcifications in bilateral mammograms). In addition, the application of this technique to mammograms of Hormone Replacement Therapy (HRT) users is investigated. Since long-term use of HRT can increase the risk of breast cancer (as a side effect of glandular tissue regeneration), quantification of temporal tissue density changes (in addition to registration) is needed for assessing density changes locally. In order to derive quantitative measures of breast tissue change, the mammogram pairs are processed using the hint representation of interesting tissue. Since such a measure depends on the image information context, we examine the problems that arise from combining image registration and quantification, aiming to develop a robust framework for temporal tissue density change assessment.

Download activity - last month

Export as CSVJSONXML

Download activity - last 12 months

Export as JSONXMLCSV

Downloads by country - last 12 months

Export as CSVJSONXML

Archive Staff Only

View Item