作者: Yalin Zheng , Jagdeep Singh Gandhi , Alexandros N. Stangos , Claudio Campa , Deborah M. Broadbent
DOI: 10.1167/IOVS.09-4935
关键词: Reproducibility 、 Automated segmentation 、 Segmentation 、 Foveal avascular zone 、 Artificial intelligence 、 Computer vision 、 Fluorescein angiography 、 Fundus fluorescein angiography 、 Linear correlation 、 Foveal 、 Medicine
摘要: PURPOSE. To describe and evaluate the performance of a computerized automated segmentation technique for use in quantification foveal avascular zone (FAZ). METHODS. A FAZ using images from fundus fluorescein angiography (FFA) was applied to 26 transit-phase obtained patients with various grades diabetic retinopathy. The area containing first extracted original image smoothed by Gaussian kernel ( 1.5). An initializing contour manually placed inside iteratively moved program toward boundary. Five tests different curves were run on each assess reproducibility. accuracy also validated comparing results boundaries delineated medical retina specialists. Interobserver then evaluated delineations two experts. RESULTS. One-way analysis variance indicated that disparities between not statistically significant, signifying excellent reproducibility computer program. There significant linear correlation automation manual CONCLUSIONS. This can produce highly reproducible are comparable those made clinical It has potential assist detection management ischemia be integrated into grading systems. (Invest Ophthalmol Vis Sci. 2010;51:3653‐3659) DOI:10.1167/iovs.09-4935