Research And Application Of Algorithms For The Detection Of Architectural Distortion In Mammograms

Architectrual distortion is defined in BI-RADS as follows:The normal architecture (of the breast) is distorted with no definite mass visible. This includes speculations radiating from a point and focal retraction or distortion at the edge of the parenchyma. Architectrual distortion is a subtle abnormality in mammograms, and a source of overlooking errors by radiologists. Computer-aided diagnosis (CAD) techniques can improve the perfmance of radiologists in detecting masses and calcifications; however, most CAD systems have not been designed to detect architectural distortion.In this dissertation, we present a technique to characterize architectural distortion in mommograms.The breast contains several piecewise linear structures, such as ligaments, ducts, and blood vessels, that cause directionally oriented texture in mammograms. We present a new method to detect and localize architectural distortion by analyzing the oriented texture in mammograms. A bank of Gabor filters is used to obtain the orientation field of the given mammogram. The orientation field is filtered and downsampled, to reduce noise and also to reduce the burden of calculation. The downsampled orientation field is analyzed to produce three phase portrait maps:node, saddle and spiral. Then three phase portrait maps are thresholded and morphologically filtered to detect architectural distortion. At last the detecting results had been validated and evaluated by mammograms from the tumor hospital of Liaoning province and the tumor hospital of Tianjin. The algorithm has been successfully applied in detection of architectural distortion of MammoCAD. About 80.95% of the architectural distortion cases were detected at rate of 0.33 false positive (PF) per image.