AUTOMATED DETERMINATION OF MACULA CENTRE POINT BASED ON GEOMETRICAL AND PIXEL VALUE APPROACHES TO SUPPORT DETECTION OF FOVEAL AVASCULAR ZONE

摘要

The centre of macula can be used to support the determination of foueavascular zone which is one of the features in automated diabetic retinopathy detectionThe centre of macula detection novel approach is divided into two parts, namely geometrical and pixel value approaches. Geometrical approach is based on the orientation ocentre of macula into the optic disc centre point, whereas pixel value approach ile particular regioninterestormed by finding the centre of several dark pixels onValidation of the centre of macula detection method shows that the correlation coeffcientobtained between detected FA Z of the proposed method and FA Z of ophthalmologist s is0. 81. Correlation coefficient achieved between detected FAZ of the proposed method andthe grade prowided by MES'STDOR database is 0. 56. Finally, correlation coeficient be-teen detected FAZ of ophthalmologist's and the grade is 0.49. Thelese promsing reslt.indicate that the proposed scheme is reliable tobe implemented in the development ofa computerized system to assist ophthalmologists in monitoring and diagnosing diabeticretinopathy.