Date: 30^th November, 2012
Time: 11:00 A.M. - 12:30 P.M.

Abstract-Architectural distortion is a subtle sign of breast cancer that could be missed in screening mammography. This seminar will present several techniques for the detection of architectural distortion in mammograms based on the analysis of oriented texture using Gabor filters, modeling of orientation fields by phase portraits, and modeling of the oriented structure of breast tissues.

Screening mammograms obtained prior to the detection of cancer could contain subtle signs of breast cancer, in particular, architectural distortion. Several methods will be described for the characterization architectural distortion based on the analysis of the angular spread of power and other characteristics, fractal analysis, texture analysis, and measures of divergence. With a dataset of 106 prior mammograms of 56 interval-cancer cases and 52 mammograms of 13 normal cases, area under the receiver operating characteristic curve of up to 0.78 has been obtained. Free-response receiver operating characteristics have indicated sensitivity of 0.80 at fewer than 4 false positives per patient.

The results indicate that the methods proposed can help in the detection of breast cancer at earlier stages than possible by visual interpretation.