Breast Mass Detection and Visualization with Complementary Deep Learning Architectures

Abstract

Abstract · Purpose: Mammograms are analyzed to identify and localize breast mass lesions as an aid to clinician review. · Approach: Two complementary forms of deep learning are used to identify the regions of interest (ROIs). An object-detection algorithm, YOLO v5, analyzes the entire mammogram to identify discrete image regions likely to represent masses. Object detections exhibit high precision. A convolutional neural network (CNN) also analyzes the mammogram after it has been decomposed into subregion tiles, and is trained to emphasize sensitivity (recall). The ROIs identified by each form of analysis are highlighted in different colors to facilitate an efficient staged review. · Results and conclusion: The object-detection stage alone exhibits high precision but insufficient overall accuracy for a clinical application. The CNN stage nearly always detects tumor masses when present, but typically occupies a larger area of the image. By inspecting the high-precision regions followed by the high-sensitivity regions, clinicians can quickly identify likely lesions before completing review of the full mammogram. Even without removing pectoral muscle from the analysis, the ROIs occupy less than 20% of the tissue in the mammograms on average. As a result, the proposed system helps clinicians review mammograms with greater accuracy and efficiency.