Physics-Grounded Adversarial Stain Augmentation with Calibrated Coverage Guarantees
For histopathology AI deployment across hospitals, CASA provides a principled augmentation method with calibrated coverage guarantees, significantly improving worst-group accuracy.
CASA uses adversarial augmentation in the Macenko stain space with a calibrated budget to improve histopathology model robustness to stain variation, achieving 93.9% slide-level accuracy on Camelyon17-WILDS, outperforming prior methods by at least 5.5 percentage points.
Stain variation across hospitals degrades histopathology models at deployment. Existing augmentation methods perturb color spaces with arbitrary hyperparameters, lacking both a principled budget and coverage guarantees for unseen centers. We propose \textbf{C}alibrated \textbf{A}dversarial \textbf{S}tain \textbf{A}ugmentation (\textbf{CASA}), which performs adversarial augmentation in the Macenko stain parameter space with a budget calibrated from multi-center statistics via the DKW inequality. On Camelyon17-WILDS (5 seeds), CASA achieves $93.9\% \pm 1.6\%$ slide-level accuracy -- outperforming HED-strong ($88.4\% \pm 7.3\%$), RandStainNA ($85.2\% \pm 6.7\%$), and ERM ($63.9\% \pm 11.3\%$) -- with the highest worst-group accuracy ($84.9\% \pm 0.9\%$) among all 10 compared methods.