Magnification-Invariant Image Classification via Domain Generalization and Stable Sparse Embedding Signatures

TL;DR

This study demonstrates that domain-general models with sparse embedding signatures improve magnification-invariant histopathology classification, outperforming GAN augmentation and reducing feature complexity.

Contribution

The paper introduces a domain-general training approach that yields compact, transferable representations for robust histopathology classification across magnifications.

Findings

Domain-general models achieved the strongest discrimination across magnifications.

Sparse embeddings reduced feature dimensions threefold while maintaining high accuracy.

Reproducibility of signatures increased from near-zero to 0.99 between different magnifications.

Abstract

Magnification shift is a major obstacle to robust histopathology classification, because models trained on one imaging scale often generalize poorly to another. Here, we evaluated this problem on the BreaKHis dataset using a strict patient-disjoint leave-one-magnification-out protocol, comparing supervised baseline, baseline augmented with DCGAN-generated patches, and a gradient-reversal domain-general model designed to preserve discriminative information while suppressing magnification-specific variation. Across held-out magnifications, the domain-general model achieved the strongest overall discrimination and its clearest gain was observed when 200X was held out. By contrast, GAN augmentation produced inconsistent effects, improving some folds but degrading others, particularly at 400X. The domain-general model also yielded the lowest Brier score at 0.063 vs 0.089 at baseline. Sparse embedding analysis further revealed that domain-general training reduced average signature size more than three-fold (306 versus 1,074 dimensions) while preserving equivalent predictive performance (AUC: 0.967 vs 0.965; F1: 0.930 vs 0.931). It also increased cross-fold signature reproducibility from near-zero Jaccard overlap in the baseline to 0.99 between the 100X and 200X folds. These findings show that calibrated, compact, and transferable representations can be learned without added architectural complexity, with clear implications for the reliable deployment of computational pathology models across heterogeneous acquisition settings.