Breast cancer detection and classification in mammogram using a three-stage deep learning framework based on PAA algorithm
作者：
Highlights：
• To our knowledge, this paper is the first to apply the PAA algorithm to analyze mammograms.
• To reduce FPPI, we cascaded an ROI detector to expand the PAA into a two-stage detector that further process the candidate regions. Besides, we added an ROI classifier to classify the detected lesions between benign and malignant.
• We introduced a threshold-adaptive post-processing algorithm that can adaptively adjust the threshold of post-processing algorithm according to the breast density.
• We added a whole mammogram classification branch to directly predict the benign and malignant mammograms, to achieve that fully automatic lesion detection and mammogram classification in a single model.
摘要
•To our knowledge, this paper is the first to apply the PAA algorithm to analyze mammograms.•To reduce FPPI, we cascaded an ROI detector to expand the PAA into a two-stage detector that further process the candidate regions. Besides, we added an ROI classifier to classify the detected lesions between benign and malignant.•We introduced a threshold-adaptive post-processing algorithm that can adaptively adjust the threshold of post-processing algorithm according to the breast density.•We added a whole mammogram classification branch to directly predict the benign and malignant mammograms, to achieve that fully automatic lesion detection and mammogram classification in a single model.
论文关键词：Breast cancer,Whole mammogram classification,Breast lesion detection,Deep learning,Object detection algorithm
论文评审过程：Received 14 October 2021, Revised 20 July 2022, Accepted 2 October 2022, Available online 13 October 2022, Version of Record 19 October 2022.
论文官网地址：https://doi.org/10.1016/j.artmed.2022.102419
