Hypertensive Retinopathy Diagnosis in Fundus Image Using Support Vector Machine

Hypertensive retinopathy (HR) is damage to the retina of the eye and the surrounding blood vessels caused by high blood pressure or hypertension, which causes thickening and narrowing of the blood vessels in the retina, inhibiting blood flow to the retina. This blood flow disturbance disrupts retinal function, which can affect vision. Most cases of hypertensive retinopathy do not show specific symptoms, so a reliable system is needed to identify reproductive health in hypertensive patients. Examining HR usually involves a fundus camera and an ophthalmoscope, through which the ophthalmologist manually detects the disease. This process is time-consuming and can cause observer error, so an automatic system is needed to detect HR. This study aims to design an automatic system to detect and classify hypertensive retinopathy in hypertensive patients using a combination of Local Binary Pattern (LBP) and Support Vector Machine (SVM) methods. The LBP method extracts texture features from fundus images, while SVM is used for classification. This study uses the CGI-HRDC 2023 dataset with the best results from this system, showing 81% accuracy, 88% precision, 74% sensitivity, 88% specificity, 61% kappa, and 81% F1 score These results indicate that the proposed system has great potential for use in computer-aided diagnosis (CAD) for assessing hypertensive retinopathy.