FairDiff: Fair Segmentation with Point-Image Diffusion

• DOI: 10.48550/arxiv.2407.06250
• Publication Date: 2024-07-08
• Authors: Wenyi Li; Haoran Xu; Guiyu Zhang; Huan-ang Gao; Mingju Gao; Mengyu Wang; Hao Zhao
• Keywords: FOS: Computer and information sciences; Computer Vision and Pattern Recognition (cs.CV); Computer Science - Computer Vision and Pattern Recognition

Fairness is an important topic for medical image analysis, driven by the challenge of unbalanced training data among diverse target groups and societal demand equitable quality. In response to this issue, our research adopts a data-driven strategy-enhancing balance integrating synthetic images. However, in terms generating images, previous works either lack paired labels or fail precisely control boundaries images be aligned with those labels. To address this, we formulate problem joint optimization manner, which three networks are optimized towards goal empirical risk minimization fairness maximization. On implementation side, solution features innovative Point-Image Diffusion architecture, leverages 3D point clouds improved over mask through point-mask-image synthesis pipeline. This method outperforms significantly existing techniques synthesizing scanning laser ophthalmoscopy (SLO) fundus By combining real during phase using proposed Equal Scale approach, model achieves superior segmentation performance compared state-of-the-art learning models. Code available at https://github.com/wenyi-li/FairDiff.