KOMBINASI ARSITEKTUR UNET-INCEPTION DAN DROPOUT DALAM PROSES SEGMENTASI CITRA TIGA DIMENSI TUMOR OTAK DARI MAGNETIC RESONANCE IMAGING

CHANIAGO, REDINA AN FADHILA and Andriani, Yuli and Desiani, Anita (2023) KOMBINASI ARSITEKTUR UNET-INCEPTION DAN DROPOUT DALAM PROSES SEGMENTASI CITRA TIGA DIMENSI TUMOR OTAK DARI MAGNETIC RESONANCE IMAGING. Undergraduate thesis, Sriwijaya University.

	Text RAMA_44201_08011281924029.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (1MB) | Request a copy
	Text RAMA_44201_08011281924029_TURNITIN.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (7MB) | Request a copy
	Text RAMA_44201_08011281924029_0002077202_011127702_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (604kB)
	Text RAMA_44201_08011281924029_0002077202_011127702_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (499kB) | Request a copy
	Text RAMA_44201_08011281924029_0002077202_011127702_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (116kB) | Request a copy
	Text RAMA_44201_08011281924029_0002077202_011127702_04.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (731kB) | Request a copy
	Text RAMA_44201_08011281924029_0002077202_011127702_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (44kB) | Request a copy
	Text RAMA_44201_08011281924029_0002077202_011127702_06_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (132kB) | Request a copy

Abstract

Brain tumors can be identified by performing image segmentation using a Convolutional Neural Network (CNN) from Magnetic Resonance Imaging (MRI) data. The number of three-dimensional (3D) MRI images is still limited and the data size is large. 3D MRI images are cut into two-dimensional (2D) images to simplify the segmentation process. The widely used CNN architecture in 2D image segmentation is UNet. UNet has deep layers that obtained high accuracy, but UNet uses a large kernel size and results in big parameters. The Inception architecture with a smaller kernel size is used to overcome deficiencies in UNet with additions Dropout function. In this research, 3D brain tumor segmentation from MRI images was conducted using UNet-Inception and Dropout architectures. The stages carried out in this segmentation are preprocessing, training, and testing. The study results with the 2020 Brain Tumor Segmentation (BraTS) dataset obtained an accuracy value of 96.29%, a sensitivity of 99%, a specificity of 66.54%, an f1-score of 97.96%, and an Intersection over Union (IoU) of 95.99%. It concludes that the ability of the model used in this study to identify brain tumor objects and backgrounds is excellent yet the pixels of brain tumor objects are not well predicted.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Segmentasi, Tumor Otak, MRI, UNet, Inception, Dropout
Subjects:	Q Science > Q Science (General) > Q334-342 Computer science. Artificial intelligence. Algorithms. Robotics. Automation.
Divisions:	08-Faculty of Mathematics and Natural Science > 44201-Mathematics (S1)
Depositing User:	Redina An Fadhila Chaniago
Date Deposited:	24 Feb 2023 03:47
Last Modified:	24 Feb 2023 03:47
URI:	http://repository.unsri.ac.id/id/eprint/89952

Actions (login required)

View Item