Skripsi

PERBANDINGAN PERFORMA U-NET DAN SWIN UNETR PADA SEGMENTASI CITRA MRI TUMOR OTAK

---

Brain tumors require rapid and accurate diagnosis, while manual segmentation of Magnetic Resonance Imaging (MRI) scans is time-consuming and highly dependent on expert knowledge. This study compares the performance of two deep learning architectures for brain tumor segmentation, namely 3D U-Net and Swin UNETR, trained and validated using the BraTS dataset and evaluated on unseen cases. Model performance was assessed using the Dice Similarity Coefficient (DSC), Intersection over Union (IoU), and the 95th percentile Hausdorff Distance (HD95). Experimental results show that 3D U-Net achieved an average Dice score of 0.767, IoU of 0.673, and HD95 of 7.1 mm, while Swin UNETR obtained higher performance with an average Dice of 0.824, IoU of 0.741, and HD95 of 6.7 mm. Overall, Swin UNETR demonstrates superior segmentation performance across all evaluation metrics, particularly in challenging regions such as the enhancing tumor (ET), whereas 3D U-Net offers advantages in computational efficiency and faster inference time. These findings indicate a trade-off between accuracy and efficiency, suggesting that model selection should be tailored to specific clinical application requirements.

---

Availability

Detail Information

Series Title

-

Call Number

T1900062025

Publisher

Indralaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2025

Collation

xvi, 138 hlm.; ilus.; tab.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

616.075 480 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Teknik Informatika
Aplikasi Medis--Tumor Otak

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment