Novel Statistical Glaucoma Prediction Algorithm Based on Spectral Analysis

Glaucoma is a condition that causes damage to the eye’s optic nerve and may result in the loss of vision. About 12 million people in India suffer from glaucoma. The ML models predicting Glaucoma are computationally complex, presence of infrastructure like GPU is of high importance, and sometimes they are also not eco-friendly due to carbon footprints. As a result, the authors of this paper proposed an algorithm which based on spectral analysis with fusion of advanced statistical approach of KL divergence for predicting the presence of Glaucoma. The algorithm has been evaluated on a total of 98 images: 70 images are used for training (glaucoma—49, normal—21) and 28 images (glaucoma—18, normal—10) are used for prediction. The accuracy obtained by the proposed novel algorithm is 76.4% with sensitivity of 86.7%, F1 Score of 82.5%, and precision 78.9%. In terms of computational efficiency, the training process takes approximately 51.116 s to complete, while the testing phase requires about 11.616 s to finish.