Classification of Retinoblastoma Eye Disease on Digital Fundus Images Using Geometric Features and Machine Learning

Aug 3, 2026

Upload by Quinn

Medical image analysis supports retinoblastoma detection by enabling clinicians to evaluate morphology, vessel density, and vascular distribution. Classifying normal versus retinoblastoma-affected retinas is a key preliminary step toward effective treatment. This study proposes a geometric feature extraction pipeline combined with machine learning. The workflow includes fundus data collection, segmentation, geometric feature extraction, model building, training/testing split, classification, and confusion-matrix evaluation. Results show segmentation can identify retinoblastoma areas and their geometric features; SVM achieves 0.96 accuracy, outperforming RF (0.55) and DT (0.63), with a 4% gain versus prior work.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Abstract

Introduction

Clinical background of retinoblastoma

Role of medical image analysis

Motivation for geometric features and machine learning

JURNAL RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol. 9 No. 3 (2025) 477-486 e-ISSN: 2580-0760 | Classification of Retinoblastoma Eye Disease on Digital Fundus Images Using Geometric Features and Machine Learning
Arif Setiawan
Department of Information System, Faculty of Engineering, Muria Kudus University, Kudus, Indonesia
[arif.setiawan@umk.ac.id](arif.setiawan@umk.ac.id)
Abstract
Medical image analysis is essential for detecting retinoblastoma tumors due to the ability of this method to assist doctors in examining the morphology, density, and distribution of blood vessels. The classification of normal and retinoblastoma-affected retinas is a preliminary step in treating retinoblastoma tumors. Therefore, this study aimed to propose a new method for classifying normal andretinoblastoma-affected retinas using geometric feature extraction and machine learning. The workflow consisted of (1) fundus image data collection for retinoblastomas,(2) image segmentation,(3) feature extraction process,(4) building a classification model using machine learning,(5) splitting testing and training data,(6) classification process using machine learning methods, and (7) evaluation of classification results using a confusion matrix. The results showed that the segmentation method could detect retinoblastoma areas and extract their geometric features. The SVM method achieved an accuracy of 0.96 while theRFandDT had 0.55 and 0.63, respectively. Moreover, a comparison with previous research showed that the proposed method achieved a 4% improvement in the classification performance. This led to the conclusion that classification using geometric features combined with the SVM on digital fundus images of retinoblastoma eye disease produced the best results.
Keywords: retinoblastoma; digital fundus images; classification; geometric features; machine learning | | | --- | --- | | How to Cite: A. Setiawan,“Classification of Retinoblastoma Eye Disease on Digital Fundus Images Using Geometric Features and Machine Learning”, J. RESTI (Rekayasa Sist. Teknol. Inf.) , vol. 9, no. 3, pp. 477-486, May 2025.
Permalink/DOI: [https://doi.org/10.29207/resti.v9i3.6337](https://doi.org/10.29207/resti.v9i3.6337) | | | Received: January 29, 2025
Accepted: May 4, 2025
Available Online: May 24, 2025 | This is an open-access article under the CC BY 4.0 License Published by Ikatan Ahli Informatika Indonesia | 1. Introduction The introduction section sets the stage for your study by providing context, defining the problem, highlighting its significance, and outlining your contribution. It should engage the reader, establish the relevance of your work, and clearly articulate the objectives of the research. Retinoblastoma is an eye cancer that invades the retinal cells, damages eye tissue, and leads to blindness. It is a solid tumor characterized by the appearance of strabismus and leukocoria [1] . The treatment includes intra-arterial chemotherapy (IAC) which is capable of curing retinoblastoma with a success rate of up to 88.2% and reducing the metastasis process to 1.6% based on analyses [2] . The tumor leads to bleeding in the posterior ciliary artery of the outer retina and grows aggressively in vascular tissues [3] . An important observation is that retinoblastoma is an intraocular and malignant tumor commonly found in children under the age of eight. The average age of diagnosis is between two and five years. Moreover, it is often clinically diagnosed with the assistance of B-scan ultrasound (USG B-scan) [4] . Patients experience targeted intra-arterial therapy, supplemented with radiotherapy to inhibit tumor growth and preserve the eyeball. However, some patients experience retinal detachment after the surgery due to the medication and surgical remnants [5] . This shows the importance of medical image analysis for detecting retinoblastoma tumors. The method aids doctors in examining the morphology, density, and distribu

FAQs

What is the main goal of the proposed approach?

To classify normal and retinoblastoma-affected retinas using geometric feature extraction together with machine learning.

What steps are included in the workflow?

The method includes fundus image data collection, image segmentation, feature extraction, classification model building, training/testing split, classification, and evaluation using a confusion matrix.

Which classifier performed best and how accurate was it?

SVM performed best with an accuracy of 0.96; RF and DT achieved 0.55 and 0.63 respectively.

Aug 3, 2026

Upload by Quinn