{"slug": "ds-gt-arc-at-imageclefmed-gans-2026-geometric-filtering-for-privacy-preserving", "title": "DS@GT ARC at ImageCLEFmed GANs 2026: Geometric Filtering for Privacy-Preserving CT Slice Generation", "summary": "Researchers from the Data Science and Analytics Group at Georgia Tech (DS@GT ARC) developed a privacy-preserving framework for synthetic lung CT slice generation for the ImageCLEFmed GANs 2026 challenge, achieving a Privacy Preservation Score of 0.549 and an FID of 0.3290. The framework combines Optimal Transport Conditional Flow Matching with a geometric filtering pipeline, but persistent patient re-identification scores show that preventing direct image copying does not eliminate deeper anatomical identity, highlighting a frontier for future work.", "body_md": "arXiv:2607.20692v1 Announce Type: new\nAbstract: We present a privacy-preserving framework for synthetic lung CT slice generation developed for the Image-CLEFmed GANs 2026 challenge. The approach combines Optimal Transport Conditional Flow Matching with privacy-oriented training and a post-generation \"Supervisor\" pipeline that filters generated candidates in learned geometric latent spaces using autoencoder embeddings, Determinantal Point Processes, and Stein Kernel Thinning. Official results show a strong realism-privacy trade-off, with the best-performing model achieving a Privacy Preservation Score of 0.549 and competitive visual fidelity with an FID of 0.3290. While the proposed geometric filtering substantially reduces nearest-neighbor memorization and membership-inference leakage, persistent patient re-identification scores indicate that preventing direct image copying is not sufficient to remove deeper patient-specific anatomical identity, highlighting an important frontier for future privacy-preserving medical image generation.", "url": "https://wpnews.pro/news/ds-gt-arc-at-imageclefmed-gans-2026-geometric-filtering-for-privacy-preserving", "canonical_source": "https://arxiv.org/abs/2607.20692", "published_at": "2026-07-24 04:00:00+00:00", "updated_at": "2026-07-24 04:28:09.286298+00:00", "lang": "en", "topics": ["artificial-intelligence", "generative-ai", "ai-safety", "ai-ethics", "computer-vision"], "entities": ["Georgia Tech", "DS@GT ARC", "ImageCLEFmed GANs 2026", "Optimal Transport Conditional Flow Matching", "Determinantal Point Processes", "Stein Kernel Thinning"], "alternates": {"html": "https://wpnews.pro/news/ds-gt-arc-at-imageclefmed-gans-2026-geometric-filtering-for-privacy-preserving", "markdown": "https://wpnews.pro/news/ds-gt-arc-at-imageclefmed-gans-2026-geometric-filtering-for-privacy-preserving.md", "text": "https://wpnews.pro/news/ds-gt-arc-at-imageclefmed-gans-2026-geometric-filtering-for-privacy-preserving.txt", "jsonld": "https://wpnews.pro/news/ds-gt-arc-at-imageclefmed-gans-2026-geometric-filtering-for-privacy-preserving.jsonld"}}