NACDN: A Novel Approach for Noise Intensity Estimation and Adaptive Denoising in Medical Images
Maddimsetty Surya Prakash, Tamilselvan Sadasivam · International Journal of Electronics and Communication Engineering · 2025
Image denoising is a crucial task in digital image processing, aiming to remove unwanted noise while preserving important medical image details. The Noise-Adaptive Convolutional Denoising Network (NACDN) presents an innovative methodology for addressing the challenge of medical image noise. In this work, we propose a novel deep learning framework designed to estimate the intensity or level of noise present in medical images while facilitating adaptive and precise denoising strategies tailored to the specific characteristics of each image. NACDN comprises three main modules: a Noise Estimation Module, a Noise Classification Module, and an Adaptive Denoising Module. Leveraging convolutional neural networks, NACDN accurately estimates the noise intensity in medical images and categorizes noise types, enabling targeted denoising approaches. The Adaptive Denoising Module applies denoising algorithms specific to the estimated noise characteristics, ensuring optimal noise reduction while preserving image details. Through extensive experiments and evaluations, NACDN demonstrates superior performance in enhancing image quality by effectively reducing noise artifacts. This research introduces a significant advancement in image processing, offering a robust solution for noise estimation and adaptive denoising in diverse imaging applications.