Mind the Gap: Continuous Magnification Sampling for Pathology Foundation Models

2026-01-05

AI Summary

This study addresses the unclear impact of magnification-level variation and training sampling strategies on the performance of existing histopathology foundation models. It reveals that discrete uniform sampling leads to degraded performance at intermediate magnifications—a previously unreported deficiency. Framing magnification sampling as a multi-source domain adaptation problem, the work proposes a continuous magnification sampling strategy to eliminate coverage gaps across scales and derives a theoretically optimal sampling distribution. The authors introduce two new benchmark datasets (TCGA-MS and BRACS-MS) and cross-scale evaluation metrics to rigorously assess model robustness. Experimental results demonstrate that the proposed approach improves balanced accuracy at intermediate magnifications by up to 4 percentage points, establishing magnification level as a critical factor influencing model robustness in computational pathology.

Technology Category

• Computer Vision: Other Foundations of Computer Vision
• Machine Learning: Calibration & Uncertainty Quantification
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

In histopathology, pathologists examine both tissue architecture at low magnification and fine-grained morphology at high magnification. Yet, the performance of pathology foundation models across magnifications and the effect of magnification sampling during training remain poorly understood. We model magnification sampling as a multi-source domain adaptation problem and develop a simple theoretical framework that reveals systematic trade-offs between sampling strategies. We show that the widely used discrete uniform sampling of magnifications (0.25, 0.5, 1.0, 2.0 mpp) leads to degradation at intermediate magnifications. We introduce continuous magnification sampling, which removes gaps in magnification coverage while preserving performance at standard scales. Further, we derive sampling distributions that optimize representation quality across magnification scales. To evaluate these strategies, we introduce two new benchmarks (TCGA-MS, BRACS-MS) with appropriate metrics. Our experiments show that continuous sampling substantially improves over discrete sampling at intermediate magnifications, with gains of up to 4 percentage points in balanced classification accuracy, and that optimized distributions can further improve performance. Finally, we evaluate current histopathology foundation models, finding that magnification is a primary driver of performance variation across models. Our work paves the way towards future pathology foundation models that perform reliably across magnifications.

Problem

• magnification sampling
• pathology foundation models
• multi-source domain adaptation
• histopathology
• intermediate magnifications

Innovation

• continuous magnification sampling
• multi-source domain adaptation
• pathology foundation models
• magnification coverage
• optimized sampling distribution