Segmentation of Periarterial and Perivenous Capillary-Free Zones in Optical Coherence Tomography Angiography

Explore advanced deep learning approaches for automated segmentation of periarterial and perivenous capillary-free zones (CFZs) in optical coherence tomography angiography (OCTA) for early diabetic retinopathy detection in this Stanford MedAI research presentation. Learn how AI/ML Research Scientist Mansour Abtahi from UCSF developed and evaluated various deep learning models, including CNNs and Vision Transformers, to achieve precise segmentation of CFZs that serve as potential biomarkers for diabetic retinopathy monitoring. Discover the superior performance of UNet++ with EfficientNet-b7 architecture and understand how quantitative features derived from CFZ analysis can significantly improve early detection and monitoring of diabetic retinopathy. Gain insights into the intersection of computer vision, multimodal learning, and clinical applications in medical imaging, followed by interactive discussion and Q&A with the speaker who specializes in developing multimodal foundation models for cancer diagnosis and treatment planning.