🤖 AI Summary
This study addresses the significant degradation in generalization performance of skin cancer classification models under cross-device and varying acquisition conditions. Systematically evaluating single, photometric-combined, and composite data augmentation strategies within a ConvNeXt-Large architecture and assessed via ROC-AUC on multi-source ISIC data, the work demonstrates that augmentations simulating realistic domain shifts yield greater out-of-distribution (OOD) robustness than those merely optimizing in-domain performance. The proposed composite augmentation achieves the largest gain, improving average AUC by 0.053 (p<0.001). To mitigate evaluation bias, a lesion-ID-based source separation protocol is employed, and in independent clinical testing, sensitivity increases from 0.591 to 0.818.
📝 Abstract
Background/Objectives: Dermoscopic skin lesion classifiers often lose accuracy under domain shift across imaging devices, illumination, and capture artifacts. We study how data augmentation improves the robustness of a binary malignant-versus-non-malignant classifier, with emphasis on out-of-domain (OOD) generalization. Methods: Single augmentations, photometric combinations, and composite policies were searched on a multi-source ISIC Archive collection with Derm7pt, using a ConvNeXt-Large backbone and ROC-AUC. Splits were made at the lesion-ID level, and HAM10000 and ISIC 2019-2020 were held out as a predominantly source-disjoint OOD test. Results: The largest OOD gain came from the mix policy, and photometric transformations dominated the most useful OOD operations. On an expanded pool from the same held-out sources the gain was +0.053 (95% CI +0.045 to +0.061, p<0.001), consistent across four training seeds (per-seed ROC-AUC: baseline 0.761-0.775, mix 0.806-0.829). On a small independent clinical collection, single-checkpoint sensitivity rose from 0.591 to 0.818, but this rested on 22 malignant cases and did not persist across seeds. Conclusions: Augmentations modelling real sources of domain shift can matter more than maximizing in-domain accuracy. Because the policy was selected on the same sources used to evaluate it, a source-disjoint selection protocol is needed before this effect size can be read as unbiased.