Abstract
Medical images exhibit latent anatomical groupings, such as organs, tissues, and pathological regions, that standard Vision Transformers (ViTs) fail to exploit. While recent work like SBM-Transformer attempts to incorporate such structures through stochastic binary masking, they suffer from non-differentiability, training instability, and the inability to model complex community structure. We present DCMM-Transformer, a novel ViT architecture for medical image analysis that incorporates a Degree-Corrected Mixed-Membership (DCMM) model as an additive bias in self-attention. Unlike prior approaches that rely on multiplicative masking and binary sampling, our method introduces community structure and degree heterogeneity in a fully differentiable and interpretable manner. Comprehensive experiments across diverse medical imaging datasets, including brain, chest, breast, and ocular modalities, demonstrate the superior performance and generalizability of the proposed approach. Furthermore, the learned group structure and structured attention modulation substantially enhance interpretability by yielding attention maps that are anatomically meaningful and semantically coherent.
Recommended Citation
H. Cheng and X. Yu and S. Wu and L. Fang and C. Cao and J. Zhang and T. Liu and D. Zhu and W. Zhong and P. Ma, "DCMM-Transformer: Degree-Corrected Mixed-Membership Attention for Medical Imaging," Proceedings of the Aaai Conference on Artificial Intelligence, vol. 40, no. 5, pp. 3228 - 3236, Association for the Advancement of Artificial Intelligence, Jan 2026.
The definitive version is available at https://doi.org/10.1609/aaai.v40i5.37317
Department(s)
Computer Science
International Standard Serial Number (ISSN)
2374-3468; 2159-5399
Document Type
Article - Conference proceedings
Document Version
Citation
File Type
text
Language(s)
English
Rights
© 2026 Association for the Advancement of Artificial Intelligence, All rights reserved.
Publication Date
01 Jan 2026
Comments
National Institutes of Health, Grant R01GM152814