Pengembangan Sistem Klasifikasi Harapan Hidup Pasien Tumor Otak Berdasarkan Data Citra Magnetic Resonance Imaging (MRI) 3D Menggunakan Deep Learning

Ardan, Indira Salsabila (2024) Pengembangan Sistem Klasifikasi Harapan Hidup Pasien Tumor Otak Berdasarkan Data Citra Magnetic Resonance Imaging (MRI) 3D Menggunakan Deep Learning. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6026231019-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (2MB) | Request a copy

Abstract

Kanker otak termasuk kanker yang paling ganas, dengan 300 kasus terdiagnosis setiap tahunnya di Indonesia. Dalam menangani kasus tersebut, diperlukan informasi mengenai harapan hidup pasien untuk menentukan pengobatan, pemeriksaan, dan perencanaan perawatan pribadi. Namun, untuk membuat prediksi yang akurat, diperlukan keahlian dokter khusus yang jumlahnya terbatas dan tidak selalu tersedia di setiap rumah sakit. Untuk memprediksi harapan hidup pasien tumor otak, selain data citra Magnetic Resonance Imaging (MRI), data-data seperti tingkat keganasan tumor otak, serta usia pasien juga dapat dipertimbangkan. Namun, hasil analisis manual antar dokter radiologi sangat mungkin berbeda, sehingga dapat terjadi inter-observer error dan hasil yang bergantung pada pengalaman dokter. Untuk mengatasi hal tersebut, saat ini, beberapa penelitian telah dilakukan untuk membantu dokter dalam melakukan prediksi harapan hidup pasien tumor otak pada citra MRI 3D menggunakan machine learning. Namun, penelitian yang mengintegrasikan informasi mengenai citra MRI otak, tingkat keganasan tumor otak, dan usia pasien dalam melakukan prediksi harapan hidup pasien tumor otak masih belum banyak dilakukan. Oleh karena itu, penelitian ini mengusulkan sebuah sistem berbasis deep learning untuk mengklasifikasikan harapan hidup pasien tumor otak berdasarkan citra MRI 3D, tingkat keganasan tumor, dan usia pasien. Sistem ini terdiri dari beberapa proses, yaitu segmentasi tumor menggunakan U-Net dan klasifikasi tingkat keganasan tumor menggunakan Convolutional Neural Network (CNN). Output dari kedua model tersebut kemudian menjadi input bagi model klasifikasi harapan hidup, dengan informasi tambahan berupa usia pasien. Model segmentasi mencapai dice sebesar 82%, sedangkan model klasifikasi tingkat keganasan mencapai accuracy, precision, recall, dan f1-score sebesar 97%. Model klasifikasi harapan hidup mengklasifikasikan pasien ke dalam “Short-survivors”, “Mid-survivors”, dan “Long-survivors” dengan accuracy, precision, recall, dan f1-score sebesar 77%.
====================================================================================================================================
Brain cancer is one of the most malignant cancers, with 300 cases diagnosed each year in Indonesia. In dealing with such cases, information regarding the patient's life expectancy is required to determine treatment, examination, and personalized care planning. However, to make accurate predictions, specialized medical expertise is required, which is limited and not always available in every hospital. To predict the life expectancy of brain tumor patients, in addition to Magnetic Resonance Imaging (MRI) image data, data such as the degree of malignancy of the brain tumor, as well as the patient's age can also be considered. However, the results of manual analysis between radiologists may differ, resulting in inter-observer error and results that depend on the experience of the doctor. To overcome this, currently, several studies have been conducted to assist doctors in predicting the life expectancy of brain tumor patients on 3D MRI images using machine learning. However, there are still not many studies that integrate information about brain MRI images, brain tumor malignancy level, and patient age in predicting the life expectancy of brain tumor patients. Therefore, this study proposes a deep learning-based system to classify the life expectancy of brain tumor patients based on 3D MRI images, tumor malignancy level, and patient age. The system consists of several processes, namely tumor segmentation using U-Net and tumor malignancy classification using Convolutional Neural Network (CNN). The output of both models then becomes the input for the life expectancy classification model, with additional information in the form of the patient's age. The segmentation model achieved a dice of 82%, while the malignancy classification model achieved accuracy, precision, recall, and f1-score of 97%. The life expectancy classification model classified patients into "Short-survivors", "Mid-survivors", and "Long-survivors" with accuracy, precision, recall, and f1-score of 77%.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Tumor otak, Magnetic Resonance Imaging (MRI), Deep learning, Klasifikasi harapan hidup pasien, Brain tumor, Classification of patient life expectancy.
Subjects:	T Technology > T Technology (General) T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Indira Salsabila Ardan
Date Deposited:	28 Jul 2024 12:29
Last Modified:	28 Jul 2024 12:29
URI:	http://repository.its.ac.id/id/eprint/109166

Actions (login required)

View Item