In this paper, we present a novel method for detecting and classifying breast cancer calcification and masses in a single step. The detection and classification steps of calcifications and masses identifiable with a mammogram image are typically performed independently even though their simultaneous solution may lead to a more efficient approach. Our novel method utilizes a Convolutional Neural Network (CNN) to classify the calcifications and masses of different cropped images of a mammogram. We utilize a sliding window detector to break apart full mammogram images into sub-images, and identify and classify the observable objects in the sub-images. We receive multiple probabilities for each sub-image for the different possible classifications. We rank the sub-images, displaying the coordinates of the highest ranked sub-images for each classification. The results of this process are that we detect 46% of cancer within the mammograms and properly classify 64% of the calcifications and masses identified.
Gozdzialski, Scott; Stern, Alex; Fasere, Ireti; and Engels, Daniel W.
"The Simultaneous Detection and Classification of Mass and Calcification Leading to Breast Cancer in Mammograms,"
SMU Data Science Review: Vol. 2:
1, Article 10.
Available at: https://scholar.smu.edu/datasciencereview/vol2/iss1/10
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License