Modern Backbones Improve Multi-task DETR for Mammography Classification and Lesion Localization

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: 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
• Venue: arXiv: cs.CV
• Published: 2026-08-10
• Collected: 2026-08-11
• URL: http://arxiv.org/abs/2608.09801v1
• PDF: https://arxiv.org/pdf/2608.09801v1
• DOI: UNKNOWN
• Registry group: academic_apis_and_preprints
• Evidence role: canonical_academic_index
• Language: en
• Country / region: UNKNOWN

Plain-Language Brief

The study evaluates a multi-task DETR framework that simultaneously predicts the malignancy of mammogram images and locates lesions, using modern backbones such as ConvNeXtV2 and DINOv3. The results show that these modern backbones outperform older ResNet-style features in both datasets (OPTIMAM and SGM1k). ConvNeXtV2 achieves the best overall performance on OPTIMAM, while DINOv3 performs strongest on SGM1k.

What the Document Says

• The study investigates a multi-task DETR framework for joint exam-level prediction and candidate-region localization in mammography
• The framework uses shared representations to support both image-level malignancy prediction and lesion localization
• Evaluation of the performance is conducted on OPTIMAM and a biopsy-confirmed SGM1k cohort
• Modern backbones consistently outperform older ResNet-style features, with ConvNeXtV2 and DINOv3 giving the strongest overall results
• ConvNeXtV2 achieves the best overall performance on OPTIMAM reaching 97.96% AUC, 99.89% sensitivity, 25.08% mAP@.5, and 74.38% recall@.25
• On SGM1k, DINOv3 gives the strongest overall results with 90.97% AUC, 86.28% sensitivity, 82.00% specificity, 27.04% mAP@.5, and 77.32% recall@.25
• The study suggests that the quality of the backbone is crucial for effective multi-task mammography, with ConvNeXtV2 emerging as a particularly strong CNN backbone for mammography in this framework

Key Concepts Mentioned

• multi-task DETR framework
• mammography
• image-level malignancy prediction
• lesion localization
• ConvNeXtV2
• DINOv3
• OPTIMAM
• SGM1k

Claims Made by the Document

• Modern backbones outperform older ResNet-style features in mammography
• ConvNeXtV2 is a strong and well-matched CNN backbone for mammography

Evidence Used

• Results across both datasets showing superior performance of modern backbones compared to older ones
• Performance statistics of ConvNeXtV2 on OPTIMAM and DINOv3 on SGM1k

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

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.