<![CDATA[ABSTRACT The quality of radiotherapy verification images on a 6 MeV Linear Accelerator (Linac) often decreases due to noise generated by high energy, potentially reducing the accuracy of irradiation positioning and endangering surrounding healthy tissues. The combination of Bilateral Filter and Non-Local Means techniques is a denoising method designed to reduce noise while preserving important image details. This study aims to evaluate the effectiveness of combining these two techniques in improving the quality of pelvis region verification images, focusing on the quantitative parameter Peak Signal-to-Noise Ratio (PSNR) and the qualitative parameter of image information. The research design used was Research and Development (RD) with 20 samples of anteroposterior (AP) and lateral projection verification images obtained from the Electronic Portal Imaging Device (EPID) of a 6 MeV Linac at the Radiotherapy Unit of RSUD Dr. Moewardi. The denoising process was performed using Python–OpenCV. PSNR values were measured before and after denoising, while image information was assessed by four expert evaluators using a 1–5 scale checklist. Statistical analysis employed paired t-test or Wilcoxon test for PSNR, and Wilcoxon test as well as Cohen’s Kappa for image information. The results showed a significant increase in PSNR values in both projections (p 0.05) with an N-gain score of 59%–64%. Image information assessment also showed a significant improvement (p 0.05), with Cohen’s Kappa indicating moderate to strong agreement among evaluators. Therefore, the combination of Bilateral Filter and Non-Local Means has been proven effective in improving the quality of pelvis region radiotherapy verification images, potentially enhancing irradiation accuracy and patient safety. Keywords: PSNR, Bilateral Filter, Non Local Means, Radiotherapy. ABSTRAK Kualitas citra verifikasi radioterapi pada Linear Accelerator (Linac) 6 MeV sering menurun akibat noise yang dihasilkan oleh energi tinggi, sehingga berpotensi mengurangi akurasi penentuan posisi penyinaran dan membahayakan jaringan sehat di sekitarnya. Kombinasi teknik Bilateral Filter dan Non-Local Means merupakan metode denoising yang dirancang untuk mengurangi noise sekaligus mempertahankan detail penting citra. Penelitian ini bertujuan mengevaluasi efektivitas kombinasi kedua teknik tersebut dalam meningkatkan kualitas citra verifikasi regio pelvis, dengan fokus pada parameter kuantitatif Peak Signal to Noise Ratio (PSNR) dan parameter kualitatif informasi citra. Desain penelitian yang digunakan Adalah Research and Development (RnD) dengan 20 sampel citra verifikasi proyeksi AP dan lateral yang diambil dari Electronic Portal Imaging Device (EPID) Linac 6 MeV di Instalasi Radioterapi RSUD Dr. Moewardi. Proses denoising dilakukan menggunakan Python–OpenCV. Nilai PSNR diukur sebelum dan sesudah denoising, sedangkan informasi citra dinilai oleh empat penilai ahli menggunakan lembar checklist skala 1–5. Analisis statistik menggunakan uji paired t-test atau Wilcoxon untuk PSNR, dan uji Wilcoxon serta Cohen’s Kappa untuk informasi citra. Hasil penelitian menunjukkan adanya peningkatan signifikan nilai PSNR pada kedua proyeksi (p0,05) dengan N-gain skor 59%–64%. Penilaian informasi citra juga meningkat signifikan (p0,05), dengan nilai Cohen’s Kappa menunjukkan kesepakatan moderat hingga kuat antar penilai. Dengan demikian, kombinasi Bilateral Filter dan Non-Local Means terbukti efektif meningkatkan kualitas citra verifikasi radioterapi regio pelvis, yang berpotensi meningkatkan akurasi penyinaran dan keselamatan pasien. Kata Kunci: PSNR, Bilateral Filter, Non Local Means, Radioterapi.]]>
