Adaptive Wavelet Transform Techniques for Optimal Noise Reduction in Computed Tomography Scans

Computed Tomography (CT) often suffers from an inevitable increase in image noise due to the pursuit of reduced dose radiation, which compromises diagnostic quality and makes the detection of subtle features challenging. Conventional noise reduction methods often face a trade-off between effectively suppressing noise and preserving crucial fine-scale anatomical details and sharp edges. This paper presents an advanced denoising approach using the Adaptive Wavelet Transform (AWT) to obtain optimal noise reduction in CT scans while maintaining high image fidelity. The proposed method employs the multi-resolution decomposition property of the wavelet transform to separate noise (typically residing in high-frequency detail coefficients) from image structures (present across all sub-bands). An adaptive thresholding strategy is introduced to adjust the coefficient shrinkage based on the Genetic Algorithm (GA), thereby ensuring an effective balance between noise suppression and structural detail preservation. The proposed approach was rigorously evaluated on a publicly available dataset, comparing its performance with standard fixed-threshold wavelet methods and conventional spatial domain filters. Quantitative results demonstrate that the proposed method achieves superior PSNR and SSIM while preserving fine structural information that is vital for clinical diagnosis.