Hybrid textural feature set based automated diagnosis system for Age Related Macular Degeneration using fundus images

Macula is the most sensitive component of human retina and it is responsible for sharp colored vision. Any abnormality effecting macula results in blurriness and other eye impairments. Two main abnormalities related to macula are macular edema and ARMD (Age Related Macular Degeneration). This paper focus on automated detection of ARMD using digital fundus images. The proposed technique extracts macular region automatically from input image and then analyzes texture of macular region to identify abnormal macula. A novel hybrid feature set consisting of different textural and color features have been proposed. The experiments are conducted using publicly available STARE and locally available AFIO databases. Our proposed system achieves 97.5%, 83% and 95.52% sensitivity, specificity, and accuracy respectively.