DIAGNOSIS OF GALUCOMA BASED ON MEASUREMENTS ON RETINAL FUNDUS IMAGES

• Main Author: Gebriani, Fara
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/80967
• Institution: Institut Teknologi Bandung

Glaucoma is an eye disease caused by increased intraocular pressure that causes damage to the optic nerve resulting in decreased vision in the eye or even blindness. In Indonesia, 51.4% of glaucoma cases are only examined in advanced conditions where there has been significant damage to the eye. Therefore, glaucoma should be detected as early as possible so that patients can receive early treatment. Computer-assisted detection will greatly assist the glaucoma detection process. Currently, there are many developments in automatic glaucoma detection methods, one of which is the approach of quantifying Optic Cup (OC) and Optic Disc (OD) characteristics such as Cup-to-Disc Ratio (CDR), Rim-to-Disc Ratio (RDR) and measurement of neuroretinal rim thickness in the inferior, superior, nasal and temporal quadrants (ISNT Quadrant) on retinal fundus images. In a previous study, Suwandoko (2022) developed a glaucoma detection method based on the quantification of OC and OD characteristics. However, the previous research only quantified the characteristics of CDR and RDR and the method used to classify glaucoma was logistic regression. In this study, the previously developed method will be optimized by adding new features by measuring the ISNT Quadrant and using decision tree as a glaucoma classification method. The optimization of this method resulted in classification performance with accuracy, specificity, sensitivity and f1-score of 0.900, 0.722, 1.000 and 0.928 for the Drishti-GS dataset and 0.958, 0.996, 0.600 and 0.728 for the entire REFUGE dataset.