一种基于全变差先验的低剂量X线CT投影域降噪算法

A Low-Dose X-ray CT Projection Domain Denoising Algorithm Based on Total Variation Prior

随着医学影像技术快速发展,X线CT已成为二级及以上各医院临床诊断普及方法。然而,国外文献数据表明,X线CT引发的辐射占医源性辐射总数将近50%,易导致癌症、白血病、新陈代谢异常等疾病。由此低剂量X线CT扫描技术相应产生,目前采用降低管电流实现扫描剂量减少。Filtered Back of Projection算法重建中由于投影域噪声污染出现条形伪影、噪声。本文提出一种基于全变差先验的低剂量X线CT投影域降噪算法。该算法采用全变差先验融入X线CT投影域实现降噪和保持边缘,仿真数据表明该算法能够有效降低噪声、保持边缘,图像明显优于滤波反投影算法的图像。

With the rapid development of medical imaging technology,X-ray CT has become a popular method for clinical diagnosis in hospitals of Grade II and above.However,foreign literature data indicates that radiation caused by X-ray CT accounts for nearly 50%of the total iatrogenic radiation,which easily leads to diseases such as cancer,leukemia and metabolic abnormalities.The lowdose X-ray CT scanning technique is accordingly generated,and the reduction of the tube current is currently achieved by reducing the tube current.In the reconstruction of the Filtered Back of Projection algorithm,stripe artifacts and noise appear due to noise pollution in the projection domain.In this paper,a low-dose X-ray CT projection domain denoising algorithm based on total variation prior is proposed.The algorithm uses the total variation apriori to integrate the X-ray CT projection domain to achieve noise reduction and edge preservation.The simulation data shows that the algorithm can effectively reduce noise and maintain edges,and the image is obviously better than the image of the filtered back projection algorithm.