Pattern Based Glaucoma Classification Approach using Statistical Texture Features

Abstract

Glaucoma is the leading eye disorder that may cause irreversible vision loss if not diagnosed quickly. Due to its invisible symptoms, it is very hard to detect glaucoma in the early stages hence increasing its impact and leads to blindness. Due to the limitations with the available medical tests, glaucoma diagnosis is preferred with computer-aided design (CAD) approach. Hence it is necessary to propose a model to diagnose glaucoma with retinal color fundus images. This paper proposed a new methodology based on local directional texture pattern (LDTP) descriptor and statistical texture features and classified using various machine learning schemes. The proposed method is validated on Drishti-GSI and ACRIMA datasets with 101 and 705 images respectively and evaluated performance with 10-fold cross validation and 70:30 split ratio approach and reported results with sufficient performance metric values. From the obtained simulation results and metrics, we state that our approach achieves good classification performance compared to other existing approaches.