Komparasi Metode EM-GMM (Expectation Maximization-Gaussian Mixture Model) dan FCM (Fuzzy C-Means) dalam Segmentasi Citra Otak MRI (Magnetic Resonance Imaging) di RSUD Soetomo dalam Menentukan Area Tumor Otak

Sianipar, Win Heber Goklas (2017) Komparasi Metode EM-GMM (Expectation Maximization-Gaussian Mixture Model) dan FCM (Fuzzy C-Means) dalam Segmentasi Citra Otak MRI (Magnetic Resonance Imaging) di RSUD Soetomo dalam Menentukan Area Tumor Otak. Undergraduate thesis, Intitut Teknologi Sepuluh Nopember.

[thumbnail of 1313100134-Undergraduate_Thesis.pdf]
Preview
Text
1313100134-Undergraduate_Thesis.pdf - Published Version

Download (3MB) | Preview

Abstract

Segmentasi citra sudah populer dilakukan, terutama untuk para
peneliti di bidang Biomedis maupun Teknik Informatika.
Segmentasi citra dalam bidang kesehatan, mempunyai tujuan
utama untuk menentukan atau mendeteksi dini area tumor, salah
satu citra medis yang popular saat ini adalah MRI. Penelitian ini,
akan dilakukan studi data citra MRI di Rumah Sakit Umum Daerah
Soetomo, Surabaya. Metode segmentasi yang dipakai adalah
klastering menggunakan FCM dan EM-GMM. Dimana inisialisasi
jumlah klaster ditentukan berdasarkan Silhouette Index untuk EMGMM
dan Partition Coefficient Index Untuk FCM. Selain
berdasarkan nilai index tersebut penentuan dilakukan
berdasarkan pengamatan subjektif dari pihak medis. Dalam
komparasinya berdasarkan nilai similaritynya didapatkan metode
EM-GMM lebih robust terhadap Salt and Pepper Noise dibanding
FCM dan FCM lebih robust terhadap Gaussian Noise dibanding
EM-GMM.====================================================================================================================Image segmentation has been popular, especially for researchers
in the field of Biomedical and Informatics Engineering. Image
segmentation in the health field, has the primary goal of
determining or detecting early tumor areas, one of the most
popular medical images currently is MRI. This research, will be
studying MRI image data at RSUD Soetomo, Surabaya. The
segmentation method used is clustering using FCM and EM-GMM.
Where initialisation of cluster number is determined based on
Silhouette Index for EM-GMM and Partition Coefficient Index For
FCM. In addition to the index value based on the determination is
done based on subjective observations from the medical party. In
comparation based on the value of its similarity obtained a more
robust EM-GMM method against Salt and Pepper Noise than FCM
and FCM more robust against Gaussian Noise than EM-GMM.

Item Type: Thesis (Undergraduate)
Additional Information: RSSt 519.22 Sia k
Uncontrolled Keywords: EM; FCM; GMM; Klastering; MRI; Optimasi; Partition Coefficient Index; RSUD Soetomo; Segmentasi; Silhouette Index; Similarity; Statistika Fitur Citra; Tumor
Subjects: Q Science > QA Mathematics > QA248_Fuzzy Sets
R Medicine > R Medicine (General)
Divisions: Faculty of Mathematics and Science > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User: Win Heber Goklas Sianipar
Date Deposited: 16 Jan 2018 08:16
Last Modified: 05 Mar 2019 03:34
URI: http://repository.its.ac.id/id/eprint/48535

Actions (login required)

View Item View Item