Newest papers

LEAF-Mamba: Feature Adaptation and Fusion for Whole Slide Image Analysis

Lizhao Gan, Jing Zhang, Mingjun Ma, Linhao Qu, Dachun Kai, Xiaoyan Sun · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.29746 · cited by 0 works (OpenAlex) · open-access PDF

French

L'analyse de l'image de diapositive complète (WSI) permet un diagnostic et une prédiction computationnels, mais les images de gigapixels et les annotations limitées rendent l'apprentissage efficace au niveau du diapositive difficile. L'apprentissage à instances multiples aborde cette tâche en agrégant les représentations de patch provenant d'encodeurs prétrainés gelés. Récemment, Mamba a permis une modélisation efficace des séquences de patch longues. Cependant, la supervision des tâches seule ne restreint pas explicitement la récupération des informations lors de l'adaptation des fonctionnalités, tandis que Mamba couple le traitement local avec une propagation à longue portée plutôt que de conserver des représentations distinctes pour la fusion. Motiv

Machine translation (Meta M2M100); the English original follows.

Abstract

Whole slide image (WSI) analysis enables computational diagnosis and prognosis, but gigapixel images and limited annotations make slide-level learning challenging. Multiple instance learning addresses this task by aggregating patch representations from frozen pretrained encoders. Recently, Mamba has enabled efficient modeling of long patch sequences. However, task supervision alone does not explicitly constrain information recovery during feature adaptation, while Mamba couples local processing with long-range propagation rather than retaining separate representations for fusion. Motivated by these challenges, we introduce LEAF-Mamba, a framework that jointly addresses latent evidence adaptation and fusion for frozen-feature WSI analysis. A reconstruction-regularized adapter learns compact, task-specific representations while reconstructing the input features. Parallel Mamba and convolutional streams then capture long-range dependencies and sequence-local interactions, with an input-dependent gate modulating their joint representation before slide-level pooling. Experiments across feature extractors demonstrate improved average performance over strong MIL baselines in both diagnostic classification and survival prediction. Component analyses further support the contributions of adaptation and fusion. These findings highlight the value of jointly designing feature transformation and contextual aggregation to translate pretrained representations into effective slide-level predictions.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Artificial intelligence, Feature (linguistics), Task (project management), Representation (politics), Feature learning

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange