EVALUASI KUALITAS CITRA X-RAY TORAK MENGGUNAKAN BERBAGAI TEKNIK ENHANCEMENT SEBAGAI PRA-PEMROSESAN DETEKSI TUBERKULOSIS Evaluation of Chest X-ray Image Quality Using Various Enhancement Techniques as Pre-processing for Tuberculosis Detection

Eka Pramesti, Amanda Restu (2025) EVALUASI KUALITAS CITRA X-RAY TORAK MENGGUNAKAN BERBAGAI TEKNIK ENHANCEMENT SEBAGAI PRA-PEMROSESAN DETEKSI TUBERKULOSIS Evaluation of Chest X-ray Image Quality Using Various Enhancement Techniques as Pre-processing for Tuberculosis Detection. Undergraduate thesis, Universitas 17 Agustus 1945 Surabaya.

Preview	Text Abstrak.pdf Download (1MB) | Preview
	Text Bab I.pdf Restricted to Repository staff only Download (110kB)
	Text Bab II.pdf Restricted to Repository staff only Download (562kB)
	Text Bab III.pdf Restricted to Repository staff only Download (950kB)
	Text Bab IV.pdf Restricted to Repository staff only Download (1MB)
	Text Bab V dan Daftar Pustaka.pdf Restricted to Repository staff only Download (114kB)
	Text Lampiran.pdf Restricted to Repository staff only Download (1MB)
Preview	Text Jurnal dan alih publish.pdf Download (503kB) | Preview
	Text Turniti.pdf Restricted to Repository staff only Download (455kB)

Abstract

Tuberculosis (TB) remains a global health threat and ranks second in terms of case prevalence in Indonesia. Chest X-ray images serve as a primary diagnostic tool, yet their quality is often compromised by noise, low contrast, or uneven illumination. Pre-processing via image enhancement techniques is crucial for improving the visibility of TB lesions prior to analysis by medical professionals or artificial intelligence-based systems. This study aims to implement and evaluate the effectiveness of various non-deep learning enhancement techniques, both in isolation and in combination with a deep learning-based denoising model (DnCNN). Eight enhancement techniques were tested across 16 scenarios (single filters and filter-plus-DnCNN combinations). Evaluation was conducted using objective metrics: RMSE, PSNR, MSSIM, ENL, Pratt’s Figure of Merit (FOM), and R². Comprehensive results demonstrate that integrating DnCNN significantly enhances the performance of traditional filters. Overall, the combination of Anisodiff_Perona-Malik and DnCNN proved to be the superior solution, achieving the lowest RMSE (1.4533) and the highest PSNR (44.92) and R² (0.9995) values. However, the Guided method achieved the highest MSSIM (0.9999), while CLAHE combined with DnCNN recorded the highest ENL (5.39). These findings confirm that combining Anisodiff_Perona-Malik with DnCNN offers an optimal balance across most critical metrics, making it the most recommended pre-processing technique for TB X-ray images.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	image enhancement, chest X-ray (CXR), TBC, pra-pemrosesan, evaluasi matrik
Subjects:	Q Science > Q Science (General)
Divisions:	Fakultas Teknik > Program Studi Teknik Informatika
Depositing User:	1462200220 Amanda Restu Eka Pramesti
Date Deposited:	18 Jun 2026 02:49
Last Modified:	26 Jun 2026 04:30
URI:	http://repository.untag-sby.ac.id/id/eprint/45847

Actions (login required)

View Item