Objective area measurement technique for choroidal neovascularization from fluorescein angiography

Abstract

The purpose of this study was to develop a non-biased method of quantitatively measuring choroidal neovascularization (CNV) areas based on late-phase fluorescein angiography (FA) images. Experimental CNV was induced in Long Evans rats by laser disruption of the Bruch's membrane. FA was performed weekly for 5weeks. Multi-Otsu thresholding (MOT) was used to quantify CNV in late-phase FA images from both experimental rodent CNV and wet age-related macular degeneration (wAMD) patients. Images were automatically thresholded into three levels based on the image histogram, with the highest level containing CNV. To determine the technique's ability to quantify CNV areas, rats were given either triamcinolone acetonide or dexamethasone sodium phosphate to treat CNV and compared to untreated rats. The rat CNV lesion areas measured from 5-week histology sections from each treatment group were compared to areas measured from the corresponding FA images. MOT was able to detect statistical decreases in rodent CNV area in the treatment groups versus control from weeks 3 through 5. The ratio of CNV area measured from histology to area measured from FA images was not statistically different between groups. Finally, to determine the usefulness of MOT on pathological morphologies of CNV, MOT was performed on late-phase FA images from patients with classic and diffuse CNV. The technique was able to segment classical CNV in wAMD patients, but performed poorly with diffuse CNV. MOT provides a robust, objective, and quantifiable area measurement of CNV lesion area in both experimentally-induced and pathological CNV. The results indicate that MOT could be a useful research tool in helping evaluate the effects of therapeutics on CNV growth.