Computed tomography (CT) plays a vital role in medical diagnostics, where image quality hinges on the choice of reconstruction kernels. Sharp kernels boost spatial resolution but amplify noise, whereas soft kernels reduce noise yet blur edges. This study presents a novel Variational Mode Decomposition combined with Quaternion Bilateral Filtering (VMD-QBF) approach to transform sharp-kernel CT images into soft-kernel equivalents without losing key structural details. The method is benchmarked against standard denoising tools—including Non-Local Means, Anisotropic Diffusion, Bilateral Filtering, and Quaternion Bilateral Filtering (QBF)—across multiple kernels (B50, B46, B41, B36, B35, B31). Performance is measured using Mean Squared Error (MSE), Structural Similarity Index (SSIM), Multiscale SSIM (MS-SSIM), and Peak Signal-to-Noise Ratio (PSNR). Results show VMD-QBF outperforming conventional methods, with the lowest MSE, highest PSNR, and superior structural fidelity across all kernels. These outcomes confirm the method’s effectiveness in noise suppression while preserving vital image features, making it a promising option for enhancing post-reconstruction CT images.
Read the articleAuthors: Mahmoud Nasr, Krzysztof Brzostowski, Adam Piórkowski
Keywords: Image Quality Metrics, Computed Tomography (CT) Imaging, Reconstruction Kernels, Image Denoising, Variational Mode Decomposition (VMD), Quaternion Bilateral Filtering (QBF), Noise Reduction