Quantification of wet Age Related Macular Degeneration in Optical Coherence Tomography Angiography images

Abstract

Age-related Macular Degeneration (AMD) is a common eye condition that leads
to about 8% of all blindness worldwide. It is leading cause of vision loss among people age 50
and older. Late stage AMD is divided into dry AMD and wet AMD. Drusen and Choroidal
Neovascularization(CNV) are the major causes of dry and wet AMD. OCTA is recent noninvasive and safer technique that provide detailed internal and external structural information
of CNVs. Detecting and analysing CNV is very effective for proper treatment and assessment
of wet AMD. The proposed work aim to provide automated method and application to detect
and analyse CNV region in OCTA images. Proposed methodology is divided into two modules
which include CNV segmentation step based on connected component labelling algorithm and
other one is CNV quantification step based on Otsu thresholding. This provides two important
quantification measures namely CNV area and CNV vessel density. Performance of proposed
system was measured using various parameters and was found better when compared with state
of art methods. The jaccard similarity score was 0.9379 ± 0.023 and false negative rate was
0.0026 ± 0.003.