Mammography classification improves with hierarchical attention neural network
M3D-Net: Hierarchical Coordination of Spatial Context, Feature Reuse, and Differential Attention for Mammography Classification
Computer Vision and Pattern RecognitionMachine Learning
Summary
Classifying breast images accurately needs both small details and the bigger picture, which can get lost in deep computer models. The authors created M3D-Net, a neural network that keeps track of details at different scales and uses special attention methods to understand images better. They tested it on breast x-rays and ultrasound images, where it outperformed several existing models. This shows the network’s promise for improving breast image analysis, though more testing is needed.
What this means in practice
- •For medical imaging developers: Improve breast cancer screening software by integrating hierarchical attention networks that maintain detailed and contextual image features for more accurate classifications.
- •For ultrasound system engineers: Enhance ultrasound breast image analysis by combining clinical data with improved neural encoding methods for better classification accuracy.
Authors
Zheng Yu, Xinhang Li, Jiabao Gao, Boyang Wang, Xiang Li
Abstract
Breast image classification requires local detail and global tissue context, yet these cues can weaken as representations deepen. We present M3D-Net, a mammography encoder that hierarchically coordinates multi-scale coordinate attention, bounded dynamic feature reuse, and differential attention through resolution-aware operator placement. Within-stage retrieval preserves access to earlier features, coordinate-aware aggregation integrates local and global context, and differential attention operates at coarse resolutions. We evaluate image-only classification on AISSLab mammography and an adapted image--clinical model on BrEaST ultrasound. Against EdgeNeXt, RepViT, and TransXNet, the proposed implementations achieve the highest recorded validation accuracy and late-training accuracy, with the lowest endpoint cross-entropy loss. Validation accuracies reach 97.78\% and 80.39\%, respectively. These results support further evaluation of hierarchical coordination across breast imaging settings; repeated-seed, component-controlled, and independent evaluations remain necessary.