DermDiff: Generative Diffusion Model for Mitigating Racial Biases in Dermatology Diagnosis

2025-03-21

AI Summary

Existing AI-based dermatological diagnosis models suffer from degraded performance on individuals with darker skin tones due to imbalanced skin tone distribution in training data, exacerbating structural bias. This paper introduces DermDiff—the first generative framework integrating conditional diffusion modeling with multimodal vision-language alignment to enable fine-grained, joint controllable generation of skin disease and skin tone attributes, thereby mitigating bias at the data source. Built upon a latent diffusion model (LDM), DermDiff incorporates CLIP-guided text conditioning and dermoscopic feature alignment to achieve high-fidelity, diverse synthetic image generation (FID < 12.5; LPIPS < 0.28). When integrated into downstream classification, DermDiff improves AUC by 14.2% for darker-skinned populations and reduces equal opportunity difference (ΔEO) to below 0.05—substantially enhancing both diagnostic accuracy and fairness across skin tones.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Ethics, Bias, and Fairness
• Philosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

Abstract

Skin diseases, such as skin cancer, are a significant public health issue, and early diagnosis is crucial for effective treatment. Artificial intelligence (AI) algorithms have the potential to assist in triaging benign vs malignant skin lesions and improve diagnostic accuracy. However, existing AI models for skin disease diagnosis are often developed and tested on limited and biased datasets, leading to poor performance on certain skin tones. To address this problem, we propose a novel generative model, named DermDiff, that can generate diverse and representative dermoscopic image data for skin disease diagnosis. Leveraging text prompting and multimodal image-text learning, DermDiff improves the representation of underrepresented groups (patients, diseases, etc.) in highly imbalanced datasets. Our extensive experimentation showcases the effectiveness of DermDiff in terms of high fidelity and diversity. Furthermore, downstream evaluation suggests the potential of DermDiff in mitigating racial biases for dermatology diagnosis. Our code is available at https://github.com/Munia03/DermDiff

Problem

• Mitigating racial biases in dermatology diagnosis using AI
• Generating diverse dermoscopic images for underrepresented skin tones
• Improving diagnostic accuracy for skin diseases across all races

Innovation

• Generative diffusion model for diverse dermoscopic images
• Text prompting enhances underrepresented group representation
• Multimodal learning improves skin disease diagnosis accuracy