Modern Backbones Improve Multi-task DETR for Mammography Classification and Lesion Localization
2026-08-10 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial Intelligence
AI summaryⓘ
The authors studied an AI system that can both predict breast cancer from mammograms and highlight suspicious areas in the images. They used a method called multi-task DETR, where one model shares knowledge for both tasks. They tested different backbone networks and found that newer ones like ConvNeXtV2 and DINOv3 worked better than older ResNet models. ConvNeXtV2 performed best on one dataset, while DINOv3 was better on another. Their results show that choosing a good backbone is important for accurate multi-task mammography AI.
mammographymulti-task learningDETRbackbone networkConvNeXtV2DINOv3ResNetimage-level predictionlesion localizationAUC
Authors
Dinh Tan Nguyen, Quang-Hien Kha, Le-Hoang Nguyen, Minh-Toan Dinh, Xuan-Huy Nguyen, Dac Phu Ho, Cao Truong Tran, Sai Ho Ling, Lan T Ho-Pham, Liem Pham, Nguyen Quoc Khanh Le
Abstract
Joint exam-level prediction and candidate-region localization may improve the usefulness of AI support in mammography. We study this setting using a multi-task DETR framework, where shared representations support both image-level malignancy prediction and lesion localization, and evaluate its performance on OPTIMAM and a biopsy-confirmed SGM1k cohort. Across both datasets, modern backbones consistently outperformed older ResNet-style features, with ConvNeXtV2 and DINOv3 giving the strongest overall results, whereas MambaVision was less competitive. On OPTIMAM, ConvNeXtV2 achieved the best overall performance, reaching 97.96% AUC, 99.89% sensitivity, 25.08% mAP@.5, and 74.38% recall@.25. On SGM1k, DINOv3 gave the strongest overall results, with 90.97% AUC, 86.28% sensitivity, 82.00% specificity, 27.04% mAP@.5, and 77.32% recall@.25. These findings suggest that backbone quality is a critical factor in effective multi-task mammography, with ConvNeXtV2 emerging as a particularly strong and well-matched CNN backbone for mammography in this framework.