Abstract—Amethod for the detection of the most commonly missed breast cancer anomaly, Architectural distortion, is proposed here. The distorted abnormal structures associated with Architectural distortion in suspicious regions are extracted using geometrical properties of edge features based on an energy model. Contours obtained from a modified Single Univalue Segment Assimilating Nucleus filtered mammogram, are employed for this purpose. A Pectoral muscle delineation technique is incorporated in the proposed method to reduce false positive rate.A ranking value of these potential regions based on linear and converging properties is computed to identify the probable origins of architectural distortion. Experimental analysis is performed on 100 images obtained from Lakeshore Hospital, India. The results are verified by expert radiologists. The proposed algorithm is successful in 94 mammograms and the results are found to be promising.
Index Terms—mammograms, modified SUSAN filter, Energy model, shortest centroid distance, ranking metric, Architectural distortion
Cite: Rekha Lakshmanan, Shiji T. P, Vinu Thomas, Suma Mariam Jacob, and TharaPratab, "Detection of Architectural Distortion in Mammogram," Journal of Image and Graphics, Vol. 2, No. 2, pp.123-127, December 2014. doi: 10.12720/joig.2.2.123-127
Copyright © 2012-2023 Journal of Image and Graphics, All Rights Reserved