SAM-OCTA: Prompting segment-anything for OCTA image segmentation

Open Webpage

Xinrun Chen, Chengliang Wang, Haojian Ning, Shiying Li, Mei Shen

Published: 01 Jan 2025, Last Modified: 12 Apr 2025Biomed. Signal Process. Control. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Introduced fine-tuned model SAM-OCTA for precise local segmentation in OCTA images.•Explored prompt points’ impact on segmentation accuracy and methodology.•Achieved top performance in OCTA segmentation tasks on the OCTA-500 dataset.•Analyzed model scalability and improvement in artery–vein segmentation.