基于统计特性的小波噪声抑制在低剂量CT中的应用

Statistically-based Wavelet Denoising for Low-dose CT Sinogram

较高的照射剂量限制了X线断层成像(computed tomography,CT)技术在筛查及体检中的应用,目前临床常采用降低剂量的解决方案,但CT图像质量亦有明显下降。为提高低剂量CT的重建质量,提出了一种基于投影数据统计特性的小波去噪算法。通过分析低剂量投影数据的噪声特性,发现在投影域其噪声均值和方差接近非线性高斯分布,根据非平稳噪声在平稳小波域中的性质,结合贝叶斯估计方法对小波系数进行基于最小均方误差的自适应滤波,实现了图像信噪分离的目的。滤波完成后,采用常规滤波反投影(FBP)法重建CT图像。较传统算法,该方法具有较高的信噪比,实验结果表明,该算法能够有效地抑制噪声,且较好地保留图像细节。

The high radiation dosage of computed tomography limits its further applications to mass screening. Clinleally,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 statlstically-based wavelet denolsing 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.