Glaucoma is a chronic optic neuropathy that can lead to irreversible blindness, making early and accurate diagnosis essential for maintaining vision health. Fundus photography and optical coherence tomography (OCT) are traditional imaging methods that provide important structural information, but they may have low sensitivity when used alone. In this study, we propose a hybrid imaging technique that combines multi-modal features and machine learning methods to improve the detection of glaucomatous changes. Using a dataset that includes demographic, clinical, and imaging parameters (e.g., age, gender, IOP – intraocular pressure, CDR – cup-to-disc ratio, RNFL thickness, and visual acuity), we trained and tested several algorithms. Results showed that traditional models achieved up to 82–90% accuracy (SVM, Random Forest, KNN), while Convolutional Neural Networks (CNN) obtained the best performance with 93% accuracy. The hybrid CNN + SVM model outperformed standalone models, achieving 95% accuracy, 94% sensitivity, 96% specificity, and a 95% F1-score. Additionally, RNFL measurement showed significant thinning in eyes with glaucomatous damage (p 0.8). The prediction accuracy for visual field defect rate was over 90% in advanced glaucoma, consistent with the disease stage of these patients. Through deep feature extraction and conventional classifiers, the hybrid approach demonstrates improved diagnostic performance over existing methods, highlighting its potential as a robust clinical decision-support tool for both early and advanced glaucoma detection.