Statistically-based Wavelet Denoising for Low-dose CT Sinogram
基于统计特性的小波噪声抑制在低剂量CT中的应用

The high radiation dosage of computed tomography limits its further applications to mass screening. Clinically,lowdose protocol has been used in data acquisition for this situation. This will increase the image noise and degrade the image quality,and thus result in difficulties in diagnosis. To improve the image quality of low-dose CT,a statistically-based wavelet denoising method in sinogram domain is proposed. The noise properties of low-dose projection data were first analyzed and modeled. It could be regarded as approximately Gaussian distributed with a nonlinear signal-dependent variance. Then the property of non-stationary noise in the stationary wavelet domain was analyzed,and the wavelet coefficients were reconstructed with the adaptive filtering based on minimum mean-squared error combined with Bayesian estimation for an optimal noise treatment. After proposed sinogram filtering,the image was reconstructed using the conventional filtered backprojection (FBP) method. Experimental results have shown that the algorithm is effective in removing noise while maintaining the diagnostic image details.