SISTEM KLASIFIKASI TINGKAT KEPARAHAN KATARAK MENGGUNAKAN METODE HYBRID SVM DAN HARMONY SEARCH PADA CITRA FUNDUS

Hermadiputri, Firdausa Yasmin (2024) SISTEM KLASIFIKASI TINGKAT KEPARAHAN KATARAK MENGGUNAKAN METODE HYBRID SVM DAN HARMONY SEARCH PADA CITRA FUNDUS. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 20081010105.-cover.pdf Download (3MB)
	Text (Bab 1) 20081010105.-bab1.PDF Download (1MB)
	Text (Bab 2) 20081010105.-bab2.PDF Restricted to Repository staff only until 18 July 2026. Download (7MB)
	Text (Bab 3) 20081010105.-bab3.PDF Restricted to Repository staff only until 18 July 2026. Download (9MB)
	Text (Bab 4) 20081010105.-bab4.PDF Restricted to Repository staff only until 18 July 2026. Download (23MB)
	Text (bab 5) 20081010105.-bab5.PDF Download (586kB)
	Text (Daftar Pustaka) 20081010105.-daftarpustaka.PDF Download (1MB)
	Text (Lampiran) 20081010105.-lampiran.PDF Restricted to Repository staff only Download (153kB)

Abstract

Cataract is a condition caused by a biochemical reaction in the eye that results in cloudiness. It is the world's highest incidence of visual impairment and can lead to blindness. Cataracts can be diagnosed through the fundus of the eye. However, in making this diagnosis, there is often a lack of standardization among ophthalmologists due to personal experience. Therefore, this study proposes a system that implements the support vector machine classification method and the Harmony Search metaheuristic algorithm to find the optimal weight vector w in the SVM hyperplane. The research data comes from kaggle which focuses on normal and cataract eye fundus images based on mild-moderate and severe severity. This research involves the stages of digital image processing, namely RGB to Grayscale, contrast enhancement by comparing Histogram Equalization and GLCE, and feature extraction using GLCM algorithm and Haar wavelet transformation. The research also uses the SMOTEENN data resampling method. The results showed that harmony search used to optimize the values in svm can improve model performance in all tests conducted. In binary classification, the system managed to increase accuracy by 0.17 or 17%, which was originally 0.55 to 0.72 on unbalanced data, and by 0.16 or 16%, which was originally 0.67 to 0.83 on balanced data. Then in multiclass classification, the system managed to increase accuracy by 0. 18 or 18%, which was originally 0.53 to 0.71 on unbalanced data, and by 0.13 or 13%, which was originally 0.67 to 0.80 on balanced data. Not only that, with Harmony Search the computation time is more efficient. This is because harmony search has the ability to explore space globally instead of locally.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Mandyartha, Eka Prakarsa	NIDN0725058805	eka_prakarsa.fik@upnjatim.ac.id
Thesis advisor	Rizki, Agung Mustika	NIDN0025079302	agung.mustika.if@upnjatim.ac.id

Subjects:

T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Firdausa Yasmin Hermadiputri

Date Deposited:

18 Jul 2024 08:00

Last Modified:

19 Jul 2024 07:06

URI:

https://repository.upnjatim.ac.id/id/eprint/26533