基于稀疏贝叶斯学习的CT图像重构

CT Image Reconstruction Based on Sparse Bayesian Learning

计算机断层成像是医学检查的常用方法,但是检查中过量的辐射可能对病人造成二次伤害。基于此提出了一种稀疏贝叶斯学习(Sparse Bayesian Learning,SBL)的肺部计算机断层成像(Computed Tomography,CT)图像重构方法,首先应用高斯随机分布矩阵对肺部图像进行测量,并建立基于小波变换的稀疏字典,最后应用稀疏贝叶斯学习算法对图像进行重构。仿真实验结果显示,该方法能够实现对肺部图像重构,当压缩率为0.6的时候,重构的肺部组织图像的峰值信噪比达到34.1809,满足医学检查的需求。该方法能够降低医学检查中的辐射伤害,具有重要的理论研究意义和实际应用价值。

Computed tomography is a common method for medical examination.However,excessive radiation may cause secondary injury to patients.A lung CT image reconstruction method based on sparse Bayesian learning is proposed.Firstly,Gaussian random distribution matrix is used to measure lung image,and a sparse dictionary based on wavelet transform is established.Finally,sparse Bayesian learning algorithm is used to reconstruct the image.Simulation experimental results show that the method can effectively reconstruct lung image.When the compression ratio is 0.6,the peak signal-to-noise ratio of the reconstructed lung tissue image reaches 34.1809,which can meet the needs of medical examination.This method can reduce radiation injury in medical examination,which has important theoretical research significance and practical application value.