Klasifikasi Tumor Glioma dengan Modalitas Magnetic Resonance Imaging (MRI) berbasis Convolutional Neural Network (CNN) untuk Diagnosis Tingkat Keganasan Tumor Otak

Prakosa, Nadhira Anindyafitri (2024) Klasifikasi Tumor Glioma dengan Modalitas Magnetic Resonance Imaging (MRI) berbasis Convolutional Neural Network (CNN) untuk Diagnosis Tingkat Keganasan Tumor Otak. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5023201052-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (5MB) | Request a copy

Abstract

Glioma mendominasi 80% dari kasus tumor otak ganas pada pasien, menyebabkan tingginya tingkat mortalitas, disabilitas serta penurunan kualitas hidup yang signifikan. Tumor glioma diklasifikasikan menjadi Low-Grade Glioma (LGG) dan High-Grade Glioma (HGG) berdasarkan tingkat keganasannya. Klasifikasi ini sangat penting karena berdampak signifikan terhadap prognosis, keputusan pengobatan, dan perencanaan bedah. Diagnosis yang akurat oleh para ahli medis sangat penting untuk menentukan tindakan yang tepat. Computer Aided Diagnosis (CAD) dapat digunakan dalam mempermudah dokter menghasilkan diagnosis yang persisi. Metode CAD tradisional mengandalkan teknik machine learning (ML) yang memerlukan pemilihan fitur secara manual, seperti analisis tekstur. Namun, pendekatan ini dapat menyebabkan hilangnya fitur data yang esensial, yang berpotensi mengurangi akurasi diagnosis. Convolutional Neural Networks (CNN) telah merevolusi pencitraan medis dengan kemampuan luar biasa dalam mengenali dan mengidentifikasi tumor otak. Berbeda dengan metode tradisional, CNN mengintegrasikan ekstraksi fitur dalam arsitekturnya, menghilangkan kebutuhan pemilihan fitur manual. Kemampuan ini memungkinkan CNN mencapai akurasi tinggi dalam klasifikasi tumor. Untuk klasifikasi glioma, sangat penting untuk memanfaatkan informasi dari berbagai sequence MRI guna memungkinkan model CNN mengekstraksi fitur representatif secara efektif. Studi ini menggunakan pendekatan multi-sequence fusion, menggabungkan sequence MRI Flair, T1, T1ce, dan T2 dengan fine-tuned model VGG16 untuk mengklasifikasikan tingkat keganasan glioma. Model ini memanfaatkan data dari dataset BraTS 2017, 2018, 2019, dan 2020. Model yang diusulkan mencapai hasil yang luar biasa, meliputi akurasi 100%, presisi 99,04%, recall 100%, skor F1 99,52%, dan spesifisitas 99,03%. Evaluasi metrik yang luar biasa ini menunjukkan efektivitas model dalam mengklasifikasikan tingkat keganasan glioma dengan akurat. Studi ini menunjukkan kemajuan signifikan dalam mendiagnosis tingkat keganasan tumor otak, serta memperlihatkan potensi untuk mempermudah pembuatan prognosis pasien melalui klasifikasi yang akurat dan andal.
========================================================================================================================
Gliomas dominate 80% of malignant brain tumor cases in patients, presenting a significant challenge in neuro-oncology. These tumors are classified into Low-Grade Glioma (LGG) and High-Grade Glioma (HGG) based on their malignancy levels. This classification is crucial as it significantly impacts prognosis, treatment decisions, and surgical planning. Accurate diagnosis by medical experts is essential for determining the appropriate course of action. Computer-aided diagnosis (CAD) systems have emerged as valuable tools in assisting doctors with precise diagnoses. Traditional CAD methods have often relied on machine learning techniques that require manual feature selection, such as texture analysis. However, this approach can lead to the loss of essential data features, potentially compromising the accuracy of the diagnosis. Convolutional Neural Networks (CNNs) have revolutionized medical imaging by excelling in recognizing and identifying brain tumors. Unlike traditional methods, CNNs integrate feature extraction within their architecture, eliminating the need for manual feature selection. This capability allows CNNs to achieve high accuracy in tumor classification. For glioma classification, it is essential to utilize information from different MRI sequences to enable the CNN model to extract representative features effectively. This study employs a multi-sequence fusion approach, combining Flair, T1, T1ce, and T2 MRI sequences with a fine-tuned VGG16 model to classify glioma malignancy levels. The model leverages data from the BraTS 2017, 2018, 2019, and 2020 datasets. The proposed model achieves remarkable results, including 100% accuracy, 99.04% precision, 100% recall, a 99.52% F1 score, and 99.03% specificity. These exceptional performance metrics highlight the model's effectiveness in accurately classifying glioma malignancy levels. The study demonstrates significant advancements in diagnosing brain tumor malignancy levels, showcasing the potential to facilitate accurate and reliable patient prognoses through precise classification.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	CNN, Deep Learning, Glioma, Multi-sequence Fusion, VGG16
Subjects:	T Technology > T Technology (General) T Technology > T Technology (General) > T57.74 Linear programming T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Biomedical Engineering > 11410-(S1) Undergraduate Thesis
Depositing User:	Nadhira Anindyafitri Prakosa
Date Deposited:	02 Aug 2024 05:14
Last Modified:	02 Aug 2024 05:14
URI:	http://repository.its.ac.id/id/eprint/111213

Actions (login required)

View Item