Klasifikasi Penyakit Otak pada Citra MRI menggunakan Convolutional Neural Network berbasis Transfer Learning

Sidiprasetija, Arnold (2023) Klasifikasi Penyakit Otak pada Citra MRI menggunakan Convolutional Neural Network berbasis Transfer Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07211940000035_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 3 October 2025. Download (6MB) | Request a copy

Abstract

Otak adalah organ kecil yang terletak di dalam batok kepala manusia yang memiliki fungsi sangat signifikan karena mengendalikan semua anggota tubuh manusia yang lain. Ketika otak terserang penyakit, maka manusia akan kehilangan kemampuan untuk menjalani kehidupan sehari-hari seperti berbicara, berpikir, bahkan menyebabkan kematian. Salah satu penyakit otak manusia adalah Alzheimer, yaitu penyakit neurodegenerative yang artinya otak manusia tidak akan bisa kembali seperti semula. Untuk mendeteksi penyakit Alzheimer, perlu dilakukan pemindaian menggunakan metode Magnetic Resonance Imaging (MRI). Setelah dilakukan pemindaian, hasil citra tersebut akan dianalisa secara manual oleh dokter atau ahli radiologi yang rentan terhadap kesalahan manusia atau human error. Penelitian telah dilakukan untuk mengurangi human error dengan menerapkan pembelajaran mesin untuk mengklasifikasikan kondisi otak normal dan otak Alzheimer. Namun, karena data citra MRI berbentuk tiga dimensi (3D)
dan jumlahnya tidak banyak menyebabkan komputasinya cukup kompleks dan kinerja yang dihasilkan kurang maksimal. Sehingga, penelitian ini bertujuan untuk mengurangi komputasi pada 3D Convolutional Neural Network (3D CNN) dengan mendasarkan pada transfer learning. Pada penelitian ini, digunakan beberapa arsitektur model CNN ternama seperti ResNet50, SE-ResNet50, DenseNet121, dan DenseNet201. Hasil akurasi tertinggi untuk metode transfer learning adalah 94% menggunakan arsitektur SE-ResNet50. Sedangkan tanpa menggunakan transfer learning akurasi paling tinggi sebesar 77% menggunakan arsitektur DenseNet121. Sehingga performa model mengalami kenaikan akurasi dengan rata-rata 19,75%.

=============================================================================================================================
The brain is a tiny organ located inside the human skull that has a very significant function because it controls all the other members of the human body. When the brain disease, then humans will lose the ability to live life such as talking, thinking, and even causing death. One of brain disease is Alzheimer, which is a neurodegenerative disease, which means that the human brain is not will be able to return as before. To detect Alzheimer’s disease, it is necessary to scan using the Magnetic Resonance Imaging (MRI) method. After doing the scan, the image results will be analyzed manually by a doctor or radiologist who vulnerable to human error. Research has been conducted to reduce human error by applying machine learning to classify
conditions. There are two types of brains: normal brains and Alzheimer’s brains. However, because the MRI image data is three-dimensional (3D), and the number is not much, causing
the computation to be quite complex and the performance to be low; produced less than the maximum. Thus, this study aims to reduce computation on a 3D Convolutional Neural Network (3D CNN) using transfer learning two-dimensional (2D). The highest accuracy result for the transfer learning method is 94% using the SE-ResNet50 architecture. Meanwhile without using transfer learning the highest accuracy is 77% using DenseNet121 architecture. So that the performance of the model has increased accuracy with an average of 19.75%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	3D CNN, Alzheimer, Magnetic Resonance Imaging, Transfer Learning.
Subjects:	Q Science > Q Science (General) Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA336 Artificial Intelligence R Medicine > R Medicine (General) > R858 Deep Learning R Medicine > RC Internal medicine > RC78.7.N83 Magnetic resonance imaging.
Divisions:	Faculty of Electrical Technology > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	Arnold Sidiprasetija
Date Deposited:	01 Aug 2023 08:25
Last Modified:	01 Aug 2023 08:25
URI:	http://repository.its.ac.id/id/eprint/101052

Actions (login required)

View Item