A Correction to the Article “Fast Optic Disc Segmentation in Retina Using Polar Transform”

Glaucoma is one among the leading reasons of blindness and vision loss in the working population. Retinal image analysis is an important tool for early diagnosis and prognosis of glaucoma which helps in preventing vision loss. The end-to-end processing pipeline for Glaucoma detection from retinal images includes the detection of optic disc (OD), neuroretinal rim (NRR), and optic cup (OC) segmentation, feature computation from the segmented OD and OC, and estimation of Glaucoma from these features. This paper is submitted as a correction article to our previous paper, which presented only the first step of the pipeline i.e., fast and robust OD segmentation in retinal images. This correction paper presents the remaining steps of Glaucoma detection processing pipeline. The segmented OD is preprocessed to highlight the NRR and OC area. A multi-layer perceptron with 12-D feature vector is used for pixel classification-based OC segmentation. Cup-to-disc ratio and other contextual features are extracted from the segmented OD and OC. A decision tree-based random subspace ensemble classifier is used to classify the Glaucomatous and non-Glaucomatous images. Experimental evaluation shows that the proposed methodology can be reliably utilized in screening programs for early glaucoma detection.