An efficiet glaucoma prediction and classification integrating retinal fundus images and clinical data using DnCNN with machine learning algorithms

Abstract

• Earlier glaucoma prediction and prevents irreversible damage. • Cup and nerves are perspective enhanced with optimized DnCNN algorithm. • Nerve damages are analyzed for the both the eye of a patient. • Predicted the glaucoma stages using optimized machine learning. Object: Early prediction of glaucoma is crucial for preserving vision and preventing irreversible damage. This condition often progresses without noticeable symptoms until significant vision loss occurs. Timely prediction and classification are essential for effective management and treatment, allowing healthcare providers to implement interventions—such as medications or surgical options—to manage intraocular pressure and protect the optic nerve. Methods: This study utilizes the PAPILA dataset, which includes detailed fundus images and clinical parameters to enhance glaucoma detection. An optimized DnCNN (ODnCNN) denoising technique is employed to improve image quality by reducing noise. Band separation enhances the fundus images, followed by the extraction of statistical features that provide relevant metrics. These features are integrated with clinical parameters like Pachymetry, axial length, and mean defect. The study uses Bayesian Optimized Multiple Linear Regression (BOMLR) and Grid Search Optimized Support Vector Regression (GS-SVR) for feature selection and optimization. Finally, a Support Vector Machine (SVM) is used for classification based on the selected features. Results: The data is categorized into three classes: Healthy (0), Glaucoma (1), and Suspicious (2). The proposed ODnCNN with machine learning method (ODnCNN-ML) achieves an accuracy of 98 % in prediction and classification by integrating clinical parameters with image-derived features. The results indicate that combining DnCNN denoising, BOMLR, GS-SVR for feature selection, and SVM classification effectively enhances early detection and classification of glaucoma.

Topics & Concepts