Reduksi Noise Pada Citra Magnetic Resonance Imaging (MRI) Otak Menggunakan Convolutional Autoencoder (CAE)

Siagian, Pangestu Sandya Etniko (2025) Reduksi Noise Pada Citra Magnetic Resonance Imaging (MRI) Otak Menggunakan Convolutional Autoencoder (CAE). Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 21081010180_Cover.pdf Download (742kB)
	Text (Bab 1) 21081010180_Bab1.pdf Download (135kB)
	Text (Bab 2) 21081010180_Bab2.pdf Restricted to Repository staff only until 18 June 2027. Download (566kB)
	Text (Bab 3) 21081010180_Bab3.pdf Restricted to Repository staff only until 18 June 2027. Download (1MB)
	Text (Bab 4) 21081010180_Bab4.pdf Restricted to Repository staff only until 18 June 2027. Download (1MB)
	Text (Bab 5) 21081010180_Bab5.pdf Download (128kB)
	Text (Daftar pustaka) 21081010180_Daftar Pustaka.pdf Download (147kB)
	Text (Lampiran) 21081010180_Lampiran.pdf Restricted to Repository staff only until 18 June 2027. Download (945kB)

Abstract

Medical imaging, particularly brain MRI scans, plays a critical role in the diagnosis of various medical conditions. However, the quality of MRI images is often compromised by the presence of Noise, which can obscure important details and potentially reduce diagnostic accuracy. Therefore, enhancing the quality of medical images is essential to improve the effectiveness of the diagnostic process. This study aims to reduce Noise in brain MRI images using the Convolutional Autoencoder (CAE) method, with a focus on improving the quality of degraded medical images caused by different types of Noise. The types of Noise applied in this study include Speckle, Salt, Pepper, and Salt & Pepper Noise, with varying intensity levels of 0.025, 0.05, 0.075, 0.1, 0.125, 0.15, 0.175, and 0.2 These intensity levels sufficiently represent a range of Noise conditions without excessively degrading the images to the point of diagnostic irrelevance. In its implementation, the CAE model is designed to perform denoising using a convolutional network architecture composed of an encoder and a decoder. The evaluation results are assessed using the Mean Squared Error (MSE) and Peak Signal-to-Noise Ratio (PSNR) metrics. MSE provides a clear numerical measure of piksel-level errors, making it particularly useful for model optimization. It quantifies the extent of deviation between the restored image and the original image, effectively guiding piksel-level error minimization. On the other hand, PSNR serves as a more perceptually intuitive measure of visual quality, as it compares the original signal to the Noise-induced distortion resulting from model errors. The CAE model successfully enhanced image quality, achieving a highest PSNR of 31,17 dB and a lowest MSE of 0,00086 on brain MRI images corrupted with 2,5% pepper Noise. This study concludes that the CAE method is effective in reducing Noise in brain MRI images and can be utilized to enhance medical image quality, thereby supporting more accurate diagnostic processes.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Diyasa, I Gede Susrama Mas	NIDN0019067008	igsusrama.if@upnjatim.ac.id
Thesis advisor	Puspaningrum, Eva Yuliaf	NIDN0005078908	evapuspaningrum.if@upnjatim.ac.id

Subjects:

T Technology > T Technology (General)
T Technology > T Technology (General) > T385 Computer Graphics
T Technology > T Technology (General) > T58.6-58.62 Management Information Systems

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Pangestu Sandya Etniko Siagian

Date Deposited:

19 Jun 2025 02:15

Last Modified:

19 Jun 2025 02:15

URI:

https://repository.upnjatim.ac.id/id/eprint/38533