Zero-shot Segmentation of Skin Conditions: Erythema with Edit-Friendly Inversion

2025-08-02

AI Summary

Skin erythema segmentation heavily relies on large-scale pixel-level annotated datasets, which are costly and scarce in clinical dermatology. Method: We propose a zero-shot image segmentation framework that requires no mask-annotated training samples. Our approach first employs diffusion model-based edit-friendly image inversion to generate erythema-free reference images; then performs pixel-wise difference modeling between the input and reference images in the LAB color space, followed by statistical thresholding for precise lesion localization. Contribution/Results: This paradigm eliminates the need for supervised training, drastically reducing data dependency while ensuring strong generalizability and clinical interpretability. Qualitative evaluation on facial erythema cases demonstrates superior accuracy and robustness compared to conventional thresholding methods. The results validate the framework’s effectiveness and practical utility in real-world dermatological辅助 diagnosis scenarios.

Technology Category

Computer Vision: SegmentationMachine Learning: Learning on the Edge & Model CompressionNatural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasetsWeb Mining and Content Analysis: Large pretrained models with web dataSecurity and Privacy: Data transparency and provenance

Abstract

This study proposes a zero-shot image segmentation framework for detecting erythema (redness of the skin) using edit-friendly inversion in diffusion models. The method synthesizes reference images of the same patient that are free from erythema via generative editing and then accurately aligns these references with the original images. Color-space analysis is performed with minimal user intervention to identify erythematous regions. This approach significantly reduces the reliance on labeled dermatological datasets while providing a scalable and flexible diagnostic support tool by avoiding the need for any annotated training masks. In our initial qualitative experiments, the pipeline successfully isolated facial erythema in diverse cases, demonstrating performance improvements over baseline threshold-based techniques. These results highlight the potential of combining generative diffusion models and statistical color segmentation for computer-aided dermatology, enabling efficient erythema detection without prior training data.

Problem

• Zero-shot segmentation of skin erythema without labeled data
• Generative editing synthesizes reference images for alignment
• Color-space analysis detects erythema with minimal user input

Innovation

• Edit-friendly inversion in diffusion models
• Generative editing synthesizes reference images
• Color-space analysis with minimal user intervention