Study of MRI Denoising Based on TI Multiwavelet Thresholding
基于平移不变多小波方法的磁共振去噪研究

A novel multiwavelet thresholding based on translation-invariant theory is introduced to complex denoising of MR data to advance the image quality.An improved thresholding estimator is calculated by averaging estimators for a range of circulant shifts of scaled vector coefficients,and bivariate hard thresholding is adopted in the high frequency sub-bands at each scale to minimize the expected risk.Experiments show that the proposed scheme better preserves edges while suppressing noise and artifacts than some of recent effective methods and traditional TI scalar wavelet thresholding; moreover,both the SNR and RMSE are improved obviously.