← Back to VOLUME 14, ISSUE 11, NOVEMBER 2026
This work is licensed under a Creative Commons Attribution 4.0 International License.
GLAUCOMA DETECTION USING CNN AND SVM CLASSIFICATION: A ROBUST AND MULTIMODAL FRAMEWORK
Sharada Shyam Sundar, P. Tritiksha Balaaji, Neelam Sanjeev Kumar
π 4 viewsπ₯ 0 downloads
Abstract: Glaucoma damages the optic nerve slowly and silently, and vision that has been lost cannot be recovered. Screening with colour fundus photographs is therefore attractive, but deep learning models that perform well on curated test sets often struggle with images from other cameras, other clinics or eyes that carry a second disease. This paper describes and analyses a hybrid pipeline in which a convolutional neural network (CNN) is used only to extract features and a Support Vector Machine (SVM) makes the final decision. The design combines three kinds of evidence: segmentation-based biomarkers such as the vertical cup-to-disc ratio (vCDR), holistic CNN embeddings of the optic nerve head, and optional clinical variables including intraocular pressure and age. The distance of a case from the separating hyperplane, together with a Platt-scaled probability, lets the system answer βungradableβ instead of forcing a diagnosis. We review the evidence reported by the AIROGS challenge, G-RISK, PKRT-Net, a lightweight fuzzy segmentation network and GlaBoost, define an evaluation protocol built on AUC, sensitivity at 95% specificity, Dice and the Brier score, and discuss limitations. The study is analytical: no new clinical experiments are reported, and every performance figure quoted is taken from the cited publications.
Keywords: glaucoma screening, fundus photography, convolutional neural network, support vector machine, optic cup segmentation, out-of-distribution detection, robustness, multimodal learning.
Keywords: glaucoma screening, fundus photography, convolutional neural network, support vector machine, optic cup segmentation, out-of-distribution detection, robustness, multimodal learning.
How to Cite:
[1] Sharada Shyam Sundar, P. Tritiksha Balaaji, Neelam Sanjeev Kumar, βGLAUCOMA DETECTION USING CNN AND SVM CLASSIFICATION: A ROBUST AND MULTIMODAL FRAMEWORK,β International Journal of Innovative Research in Electrical, Electronics, Instrumentation and Control Engineering (IJIREEICE), DOI: 10.17148/IJIREEICE.2026.141106
