Perbandingan Hasil Segmentasi Citra Computerized Tomograpghy Tumor Hati dari Model 3D U-Net dan 3D V-Net

Pasaribu, Evan Josua (2023) Perbandingan Hasil Segmentasi Citra Computerized Tomograpghy Tumor Hati dari Model 3D U-Net dan 3D V-Net. Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 06111940000060-Undergraduate_Thesis.pdf] Text
06111940000060-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2025.

Download (3MB) | Request a copy

Abstract

Tumor ganas organ hati merupakan penyakit kanker yang umum di Indonesia. Pemeriksaan awal menggunakan citra computerized tomography (CT) scan dapat digunakan untuk penanganan awal pasien. Segmentasi citra CT scan organ hati sangat penting untuk memisahkan tumor organ hati dan bagian normal dengan efisien. Penelitian ini menerapkan arsitektur jaringan 3D U-Net dan 3D V-Net untuk segmentasi citra CT scan tumor organ hati. Kedua model ini merupakan model deep-learning yang telah banyak digunakan dalam segmentasi citra medis. Penelitian ini bertujuan mengimplementasikan 3D U-Net dan 3D V-Net untuk melakukan segmentasi citra CT scan tumor organ hati dengan akurasi yang baik dengan menggunakan kedua model tersebut sebagai model untuk pembanding. Model yang diusulkan diuji menggunakan dataset publik segmentasi tumor hati dan memberikan peningkatan akurasi berdasarkan evaluasi metrik dengan akurasi terbaik sebesar 0,930 untuk dice similarity coefficient untuk kasus segmentasi hati beserta tumor dan 0,516 untuk segmentasi tumor, serta evaluasi jaccard index atau IoU dengan skor 0,8758 untuk hati beserta tumor segmentasi tumor dan 0,7944 untuk segmentasi tumor dengan menggunakan model 3D U-Net.
========================================================================================================================
Liver cancer tumors are a common disease in Indonesia. Initial examination using computerized tomography (CT) scan images can be used for early patient management. Liver CT scan image segmentation is crucial to efficiently separate liver cancer tumors from normal parts. This research applies the 3D U-Net and 3D V-Net network architectures for segmenting liver cancer tumor CT scan images. Both models are deep-learning models widely used in medical image segmentation. The aim of this study is to implement 3D U-Net and 3D V-Net for liver cancer tumor CT scan image segmentation with good accuracy, using both models as benchmarks for comparison. The proposed model was tested using a publicly available liver tumor segmentation dataset and demonstrated an accuracy improvement based on evaluation metrics, achieving the best accuracy of 0.930 for the Dice Similarity Coefficient for liver segmentation along with the tumor and 0.516 for tumor segmentation. Furthermore, the Jaccard Index (IoU) evaluation yielded a score of 0.8758 for liver along with tumor segmentation and 0.7944 for tumor segmentation when using the 3D U-Net model.

Item Type: Thesis (Other)
Uncontrolled Keywords: 3D U-Net, 3D V-Net, Segmentasi Tumor Hati , 3D U-Net, 3D V-Net, Liver Tumor Segmentation
Subjects: Q Science > QA Mathematics > QA336 Artificial Intelligence
R Medicine > R Medicine (General) > R858 Deep Learning
Divisions: Faculty of Mathematics, Computation, and Data Science > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User: Evan Josua Pasaribu
Date Deposited: 08 Aug 2023 04:20
Last Modified: 08 Aug 2023 04:20
URI: http://repository.its.ac.id/id/eprint/103424

Actions (login required)

View Item View Item