Machine Learning based Glaucoma Detection using Electroretinography (ERG) Signals for Early Diagnosis and Monitoring

Glaucoma, a leading cause of irreversible vision loss, gradually deteriorates the optic nerve, making early detection crucial to prevent further visual impairment. Electroretinography (ERG) serves as a diagnostic test that measures the retina’s electrical response to light stimuli, offering essential insights into retinal health. This study focuses on developing a machine learning-based system aimed at automating glaucoma detection by analyzing ERG signals, thereby ensuring accurate and efficient diagnosis for healthcare providers. The proposed system employs machine learning techniques, including Support Vector Machine (SVM) and Convolutional Neural Networks (CNN), to classify ERG signals obtained from both healthy individuals and glaucoma patients. The SVM algorithm is chosen for its capability in handling linear and non-linear classifications, while CNN efficiently extracts hierarchical features from raw ERG data, enabling the identification of minute changes associated with glaucoma. The model is trained on a diverse dataset, allowing it to recognize specific retinal features indicative of glaucomatous changes. By automating the diagnostic process, the system reduces dependency on expert clinicians and enhances diagnostic efficiency. The performance of the model is evaluated using accuracy, sensitivity, and specificity, ensuring its clinical reliability. This machine learning approach to ERG signal analysis presents a robust and accessible solution for early glaucoma detection, thereby improving patient outcomes through timely intervention.