ANALISIS PERBANDINGAN SNR DAN CNR PADA CITRA DIGITAL RADIOGRAPHY KEPALA NON KONTRASMENGUNAKAN METODE PENGOLAHAN CITRA PHYTON

Abstract

Background: This study compares SNR and CNR in non-contrast head X-rays using Python. Head image quality is crucial due to its complex anatomical structure and the importance of small details in determining a diagnosis. Python is used to objectively analyze image clarity, read medical data, calculate SNR and CNR values, and display the results visually. This research is expected to help physicians obtain clearer images, accelerate diagnoses, and improve the quality of radiology services. Methods: This research was conducted at the Radiology Laboratory of Aisyiyah University Yogyakarta from March to May 2025. The subjects used were not live patients, but rather cranium phantoms, which are artificial models of human heads typically used for training or research to ensure safety. In this study, three different X- ray settings were used: 70 kV 10 mAs, 75 kV 12 mAs, and 80 kV 15 mAs. Each setting produced head X-rays with varying levels of brightness and sharpness. The images were saved in a medical-specific format called DICOM. Next, the images were analyzed using the Python programming language through the Google Colab platform. This analysis was carried out to calculate two important things: SNR (Signal-to-Noise Ratio), which describes how clear the image signal is compared to interference or "noise," and CNR (Contrast-to-Noise Ratio), which indicates how easy it is to distinguish two tissues or parts in the image. This calculation was carried out both before and after the image was improved through the image enhancement process. In this way, researchers can assess whether image processing actually makes head X-rays clearer and more useful for medical diagnosis. Results: The results showed that after image processing with Python, the SNR values increased in some settings, resulting in cleaner images, but the CNR values decreased in all images. This means that although the images appear sharper and clearer visually, the ability to distinguish anatomical structures is reduced. Therefore, improving visual quality through image enhancement does not always translate into improved diagnostic quality, requiring caution when applying it to radiology. Conclusion: The conclusion of this study is that processing non-contrast head radiographic images using Python can improve the SNR value in several parameter variations, so that the image appears cleaner from noise interference. However, the CNR value tends to decrease in all variations, which means the ability to distinguish anatomical structures is reduced. This shows that increasing visual acuity of the image is not always directly proportional to improving diagnostic quality, so image enhancement techniques need to be applied carefully to maintain a balance between image clarity and clarity of anatomical details to support medical diagnosis