AHMBF: An Adaptive Hybrid Filtering Method for Noise Reduction in Cardiac MRI

Medical imaging is critical in diagnosing and managing heart diseases. Unfortunately, Rician noise alongside motion blur artifacts often contaminate cardiac MRI images, severely degrading their quality and hampering clinical evaluation. This paper presents an Adaptive Hybrid Median-Bilateral Filtering (AHMBF) algorithm for the medical image denoising of cardiac MRIs. The adaptive method proposed is focused on a combination of impulse noise median filtering and edge preserving smoothing bilateral filtering. Unlike classical denoising techniques, AHMBF reduces noise and preserves structure by adjusting the filtering parameters to local picture data. We assess the capabilities of AHMBF alongside more conventional denoising techniques including Gaussian filtering, Non-Local Means (NLM), and BM3D filtering using the ACDC dataset. As demonstrated, AHMBF surpasses prior techniques achieving a PSNR of 33.5 dB, SSIM 0.94, and MSE 98.7, proving effectiveness in noise suppression while detail retention. These findings suggest that enhanced detail of cardiac MRI images achieved using AHMBF improves automated and clinical diagnosis.