Segmentation Style Discovery: Application to Skin Lesion Images
This addresses variability in medical image segmentation for healthcare applications, offering a novel approach to handle annotator differences without requiring correspondence, though it is incremental in building on multi-annotator methods.
The paper tackles the problem of variability in medical image segmentation by introducing segmentation style discovery, proposing StyleSeg to learn diverse segmentation styles without annotator correspondence, and showing it outperforms competing methods on four skin lesion datasets and aligns with annotator preferences on a new dataset.
Variability in medical image segmentation, arising from annotator preferences, expertise, and their choice of tools, has been well documented. While the majority of multi-annotator segmentation approaches focus on modeling annotator-specific preferences, they require annotator-segmentation correspondence. In this work, we introduce the problem of segmentation style discovery, and propose StyleSeg, a segmentation method that learns plausible, diverse, and semantically consistent segmentation styles from a corpus of image-mask pairs without any knowledge of annotator correspondence. StyleSeg consistently outperforms competing methods on four publicly available skin lesion segmentation (SLS) datasets. We also curate ISIC-MultiAnnot, the largest multi-annotator SLS dataset with annotator correspondence, and our results show a strong alignment, using our newly proposed measure AS2, between the predicted styles and annotator preferences. The code and the dataset are available at https://github.com/sfu-mial/StyleSeg.