# Auditing Data Leakage in Whole-Slide Image Multimodal Benchmarks

arXiv: https://arxiv.org/abs/2607.12278 · HTML: https://mlsanity.com/paper/2607.12278/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Medical vision-language models, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 21 of its 35 references that are in the map.
- First posted on arXiv 14 Jul 2026.

## Placed in

- Multimodal LLMs / vision-language models / [Medical vision-language models](https://mlsanity.com/mllm/medical-vision-language-models/)

## Builds on (most cited first, 21 of 21)

- [Datasheets for Datasets](https://mlsanity.com/paper/1803.09010/) (2018)
- [LLaVA-Med: Training a Large Language-and-Vision Assistant for Biomedicine in One Day](https://mlsanity.com/paper/2306.00890/) (2023)
- [LongNet: Scaling Transformers to 1,000,000,000 Tokens](https://mlsanity.com/paper/2307.02486/) (2023)
- [Quilt-1M: One Million Image-Text Pairs for Histopathology](https://mlsanity.com/paper/2306.11207/) (2023)
- [Common pitfalls and recommendations for using machine learning to detect and prognosticate for COVID-19 using chest radiographs and CT scans](https://mlsanity.com/paper/2008.06388/) (2020)
- [Benchmark Data Contamination of Large Language Models: A Survey](https://mlsanity.com/paper/2406.04244/) (2024)
- [Investigating Data Contamination in Modern Benchmarks for Large Language Models](https://mlsanity.com/paper/2311.09783/) (2023)
- [Visual Language Pretrained Multiple Instance Zero-Shot Transfer for Histopathology Images](https://mlsanity.com/paper/2306.07831/) (2023)
- [Quilt-LLaVA: Visual Instruction Tuning by Extracting Localized Narratives from Open-Source Histopathology Videos](https://mlsanity.com/paper/2312.04746/) (2023)
- [Croissant: A Metadata Format for ML-Ready Datasets](https://mlsanity.com/paper/2403.19546/) (2024)
- [Codabench: Flexible, Easy-to-Use and Reproducible Benchmarking Platform](https://mlsanity.com/paper/2110.05802/) (2021)
- [Multiple Instance Captioning: Learning Representations from Histopathology Textbooks and Articles](https://mlsanity.com/paper/2103.05121/) (2021)
- [SlideChat: A Large Vision-Language Assistant for Whole-Slide Pathology Image Understanding](https://mlsanity.com/paper/2410.11761/) (2024)
- [HistGen: Histopathology Report Generation via Local-Global Feature Encoding and Cross-modal Context Interaction](https://mlsanity.com/paper/2403.05396/) (2024)
- [WSI-LLaVA: A Multimodal Large Language Model for Whole Slide Image](https://mlsanity.com/paper/2412.02141/) (2024)
- [Patho-R1: A Multimodal Reinforcement Learning-Based Pathology Expert Reasoner](https://mlsanity.com/paper/2505.11404/) (2025)
- [Current Pathology Foundation Models are unrobust to Medical Center Differences](https://mlsanity.com/paper/2501.18055/) (2025)
- [WSI-VQA: Interpreting Whole Slide Images by Generative Visual Question Answering](https://mlsanity.com/paper/2407.05603/) (2024)
- [WsiCaption: Multiple Instance Generation of Pathology Reports for Gigapixel Whole-Slide Images](https://mlsanity.com/paper/2311.16480/) (2023)
- [PathGen-1.6M: 1.6 Million Pathology Image-text Pairs Generation through Multi-agent Collaboration](https://mlsanity.com/paper/2407.00203/) (2024)
- [MLLM-HWSI: A Multimodal Large Language Model for Hierarchical Whole Slide Image Understanding](https://mlsanity.com/paper/2603.23067/) (2026)

## Built on by (most cited first, 1 of 1)

- [Contamination, Prior, or Evidence? Decomposing and Training Evidence Use in Whole-Slide Vision-Language Models](https://mlsanity.com/paper/2609.32185/) (2026)

## Abstract (arXiv)

Recent vision-language models (VLMs) for computational pathology report striking zero-shot performance on whole-slide image (WSI) visual question answering (VQA) benchmarks. We audit these claims and find them fundamentally compromised by data leakage at two hierarchical levels: patient-level leakage, where slides from the same case appear in both training and test folds, and institutional-level leakage, where different cases nonetheless share staining-batch and scanner signatures through a common Tissue Source Site (TSS). By tracing canonical slide, case, and TSS identifiers across major public resources, we document case level train test overlaps of 92.3~100% on TCGA-derived benchmarks, together with near-complete TSS overlap. We further demonstrate that both leakage levels are linearly decodable from foundation-model feature space, that they induce a measurable accuracy gap between leaked and audit-clean cases on a published checkpoint, and that across multiple published WSI VLMs, peak reported accuracies concentrate on the most heavily contaminated benchmarks. Therefore, the current WSI VQA evaluation cannot distinguish genuine multimodal reasoning from nearest-neighbor retrieval over memorized institutional and patient-specific artifacts. Finally, we outline concrete recommendations for contamination-free evaluation. By addressing benchmark construction, provenance disclosure, and automated overlap auditing, we aim to guide future research toward verifiable claims of progress.