Beyond Relevance: Bayesian Evidence Acquisition for Agentic Whole-Slide Image Reasoning

2026-08-06

AI Summary

This work addresses the limitations of existing training-free agents in whole-slide image (WSI) reasoning, which rely on semantic retrieval of image patches and struggle to identify diagnostically discriminative evidence, often resulting in redundant observations and unresolved uncertainty. The study introduces, for the first time, a Bayesian evidence acquisition framework into WSI reasoning, formulating a training-free agent that maintains posterior beliefs over diagnostic hypotheses and actively selects image patches based on expected information gain. A multi-level evidence controller dynamically decides whether to answer, continue evidence collection, or increase resolution. Built upon off-the-shelf foundation models without fine-tuning, the method achieves state-of-the-art zero-shot performance across five WSI-VQA benchmarks, significantly improving evidence-gathering efficiency and demonstrating the efficacy and promise of Bayesian reasoning in computational pathology.

Abstract

Whole-slide image (WSI) reasoning requires an agent to sequentially acquire visual evidence before answering a diagnostic question. Existing training-free agentic frameworks formulate this process as iterative patch retrieval based on semantic relevance to the question. However, semantic relevance does not necessarily imply diagnostic informativeness in computational pathology, where competing diagnoses often exhibit similar and overlapping morphological patterns, making many patches semantically relevant yet diagnostically non-discriminative. Consequently, relevance-based retrieval may acquire redundant observations and leave diagnostic uncertainty unresolved. We propose BEACON, a plug-and-play agentic framework that reformulates WSI reasoning as a Bayesian evidence acquisition problem. BEACON maintains a probabilistic belief over competing diagnostic hypotheses and sequentially acquires patches by maximizing expected information gain (EIG) to reduce diagnostic uncertainty. An evidence controller then determines whether to answer, acquire additional evidence, or perform higher-resolution inspection. Built entirely from off-the-shelf foundation models, BEACON requires no additional training or fine-tuning. Extensive zero-shot experiments across five WSI-VQA benchmarks demonstrate that BEACON achieves the strongest overall performance among training-free agentic frameworks while substantially improving evidence acquisition efficiency, establishing Bayesian evidence acquisition as a principled paradigm for uncertainty-aware agentic WSI reasoning. The code is available at https://github.com/bryanwong17/BEACON

Problem

• whole-slide image reasoning
• diagnostic uncertainty
• evidence acquisition
• semantic relevance
• computational pathology

Innovation

• Bayesian evidence acquisition
• whole-slide image reasoning
• expected information gain
• diagnostic uncertainty
• training-free agentic framework