Self-Supervised Attention Model for Breast Cancer Detection from Mammography

DOI

10.2991/978-94-6239-674-6_22How to use a DOI?

Keywords

Self-supervised learning; Attention mechanism; Mammography; Breast cancer detection; Contrastive pretraining; Weakly supervised localization

Abstract

Mammography-based breast cancer screening is constrained by thick tissue characteristics, obscure lesion sizes, and annotations by professionals. The proposed self-supervised attention-based deep learning model is a solution to automated malignant lesion detection. This approach uses contrastive pretraining to train strong image features on large untagged mammograms and then fine-tunes on tagged CBIS- DDSM and INbreast datasets. Two-step attention mechanism is a hybridization of channel and spatial filtering to emphasize diagnostically valuable regions without the supervision at the pixel level. The experimental outcomes show significant performance improvement over the baseline CNNs, reaching 0.931 accuracy, 0.914 sensitivity, and 0.962 AUC, and weakly supervised lesion localization by Dice coefficient of 0.689. The method minimizes the annotation dependence, increases the interpretability with attention heatmaps, and provides credible cancer detectability in the screening processes. These results suggest that self-supervision that is combined with attention mechanisms can be a direction taken in scalable and clinically relevant computer-aided diagnosis in breast imaging.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Rahul Kumar
AU  - Mohammad Shahrookh Husain
AU  - Sujeet Kumar Sahani
AU  - Rohit Kumar
AU  - Abhishek Varshney
PY  - 2026
DA  - 2026/05/28
TI  - Self-Supervised Attention Model for Breast Cancer Detection from Mammography
BT  - Proceedings of the International Conference on Sustainable Computing and Artificial Intelligence (ICSCAI 2025)
PB  - Atlantis Press
SP  - 254
EP  - 264
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6239-674-6_22
DO  - 10.2991/978-94-6239-674-6_22
ID  - Kumar2026
ER  -