Klasifikasi Citra Fundus Retina menggunakan Metode Convolutional Neural Network (CNN) ResNet-50 untuk Deteksi Dini Stroke

Affan, Aghnia Hasya (2023) Klasifikasi Citra Fundus Retina menggunakan Metode Convolutional Neural Network (CNN) ResNet-50 untuk Deteksi Dini Stroke. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07311940000059-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (24MB) | Request a copy

Abstract

Stroke atau cedera cerebrovascular merupakan penyakit yang terjadi ketika pembuluh darah yang membawa oksigen dan nutrien ke otak mengalami penyumbatan atau pecah sehingga menyebabkan sel otak mati. Stroke merupakan penyebab kematian ke-2, stroke bertanggung jawab atas sekitar 11% dari total kematian. Hasil dari penelitian ini diharapkan dapat membantu dokter dalam mengidentifikasi atau mendiagnosa adanya stroke, serta mengantisipasi penyakit stroke lebih lanjut berdasarkan hasil klasifikasi citra retina pasien. Penelitian ini berfokus dalam menggunakan metode transfer learning Pre-Trained ResNet-50 yang akan melakukan pengklasifikasian gambar menjadi normal dan eye stroke terhadap citra fundus retina. Tahapan yang dilakukan dalam penelitian ini adalah menyiapkan dataset, preprocessing, klasifikasi dan deteksi menggunakan ResNet-50. Dalam pengujian ini, dilakukan pengujian terhadap model Pre-Trained ResNet-50 dan ResNet-50. Diperoleh hasil terbaik dari model model Pre-Trained ResNet-50 menunjukkan nilai akurasi sebesar 89,82%, presisi sebesar 92,76%, specificity sebesar 92,14%, dan F1-Score sebesar 90,08%. Hasil ini menunjukkan bahwa model yang dikembangkan memiliki kinerja yang baik dalam mendeteksi berdasarkan citra fundus retina. Dalam penelitian ini, juga dilakukan pengujian dengan menggunakan ResNet-50 tanpa proses transfer learning. Hasil terbaik dari model ini menunjukkan nilai akurasi sebesar 92,78%, presisi sebesar 97,66%, specificity sebesar 88,46%, dan F1-Score sebesar 96,68%. Hasil ini menunjukkan bahwa model yang dikembangkan memiliki kinerja yang lebih tinggi dibandingkan dengan model Pre-Trained tetapi model cenderung mengalami overfitting.
====================================================================================================================================
Stroke or cerebrovascular injury is a condition that occurs when blood vessels carrying oxygen and nutrients to the brain become blocked or ruptured, resulting in the death of brain cells. Stroke is the second leading cause of death, accounting for approximately 11% of total deaths. The results of this research are expected to assist doctors in identifying or diagnosing strokes, as well as anticipating further stroke-related diseases based on the classification of patients' retinal images. This study focuses on using the Pre-Trained ResNet-50 transfer learning method to classify images into normal and eye stroke categories based on retinal fundus images. The stages conducted in this research include dataset preparation, preprocessing, classification, and detection using ResNet-50. In this evaluation, testing was performed on both the Pre-Trained ResNet-50 model and the ResNet-50 model. The best results were obtained from the Pre-Trained ResNet-50 model, with an accuracy of 89,82%, precision of 92,76%, specificity of 92,14%, and F1-Score of 90,08%. These results indicate that the developed model performs well in detecting eye stroke based on retinal fundus images. Additionally, testing was also conducted using the ResNet-50 model without the transfer learning process. The best results from this model showed an accuracy of 92,78%, precision of 97,66%, specificity of 88,46%, and F1-Score of 96,68%. These results indicate that the developed model has higher performance compared to the Pre-Trained model, but the model tends to experience overfitting.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Stroke, Retina, Pre-Trained ResNet-50, Klasifikasi, Classification.
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Biomedical Engineering > 11410-(S1) Undergraduate Thesis
Depositing User:	Aghnia Hasya Affan
Date Deposited:	30 Jul 2023 00:55
Last Modified:	30 Jul 2023 00:55
URI:	http://repository.its.ac.id/id/eprint/99874

Actions (login required)

View Item