摘要
当降低X射线计算机断层成像(computer tomography,CT)的管电流强度,投影数据将变为低信噪比的,使CT图像复原变为欠定问题。提出了一种基于离散剪切波MAP-马尔可夫随机场(map Markov random field,MAP-MRF)的低剂量医学CT图像去噪算法。首先,利用离散剪切波多尺度、多方向的优点,可以精确地计算剪切系数,而且可以更稀疏的表示图像信息,结合马尔可夫理论提出了基于MAP-MRF的去噪算法,使盲复原结果在噪声消除和边缘保持上达到有效的平衡,且优化了正则化项,使算法快速收敛。图像的噪声得到了较为有效的滤除,并且图像特征得到了较好的保留,将MAP-MRF算法与其他5种对比算法进行Shepp-Logan体膜、临床脑部CT、临床低剂量肩部CT的定性及定量实验,实验结果表明,在定性实验中该方法可以获得良好的CT重建图像,保留清晰的纹理细节和结构特征;在10 mA模拟噪声及20 mA临床低管电流定量实验中峰值信噪比(PSNR)、结构相似性指数(SSIM)指标均有较大改进,均方根误差(RMSE)低于对比算法,证明了算法的有效性。
When the computer tomography(CT)image is reconstructed,the signal-to-noise ratio of the CT image becomes low.This paper presents a low-dose medical CT image denoising algorithm based on discrete shear wave(map Markov random field,MAP-MRF).Firstly,using the advantages of multi-scale and multi direction of discrete shear wave,the shear coefficient can be calculated accurately,and the image information can be expressed sparsely.Combining with Markov theory,a denoising algorithm based on MAP-MRF is proposed,which makes the blind restoration result achieve an effective balance in noise elimination and edge preservation,and optimizes the regularization term,so that the algorithm converges quickly.The image noise has been effectively filtered out,and the image features have been well preserved.The qualitative and quantitative experiments of Shepp Logan body membrane,clinical brain CT and clinical low-dose shoulder CT with MAP-MRF algorithm and other five comparison algorithms are carried out.The experimental results show that this method can obtain good CT reconstruction images and retain clear texture details and structure in the qualitative experiments In the quantitative experiments of 10 mA simulated noise and 20 mA clinical low tube current,the PSNR and SSIM indexes are greatly improved,and the root mean square error(RMSE)is lower than the comparison algorithm,which proves the effectiveness of the algorithm.
作者
刘晓培
滕建辅
费腾
孙云山
LIU Xiaopei;TENG Jianfu;FEI Teng;SUN Yunshan(School of Electrical and Information Engineering,Tianjin University,Tianjin 300072,China;School of Microelectronics,Tianjin University,Tianjin 300072,China;School of Communication,Tianjin University of Commerce,Tianjin 300134,China)
出处
《光电子.激光》
CAS
CSCD
北大核心
2021年第6期613-620,共8页
Journal of Optoelectronics·Laser
基金
天津市自然科学基金项目基于贝叶斯压缩感知的低剂量医学CT图像盲复原重建算法(16JCYBJC28800)资助项目。