图像域增强约束卷积稀疏编码的稀疏角度CT重建算法研究

Sparse-View CT Reconstruction Algorithm based on Convolutional Sparse Coding with Image Domain Enhancement Constrained

稀疏角度CT重建因其可以降低辐射剂量引起广泛关注,然而减少角度会降低重建图像质量,影响诊断结果分析.为解决上述问题,提出了图像域增强约束卷积稀疏编码的稀疏角度CT重建算法,该算法继承了卷积稀疏编码的优点,通过直接处理整幅图像提取特征,克服了字典学习因图像分块聚合引起的伪影.继而引入全变分正则项来增强图像域的约束,可以有效地进一步抑制噪声.通过几组稀疏角度的重建实验与不同算法对比,实验结果表明,所提算法在噪声抑制、伪影减少和图像细节恢复方面性能优越.

Sparse-view CT reconstruction has attracted widespread attention because it can reduce radiation dose.However,reducing view will decrease the quality of the reconstructed image and affect the analysis of diagnostic results.To address the above problems,this paper proposes a sparseview CT reconstruction algorithm based on convolutional sparse coding with image domain enhancement constrained.The algorithm inherits the advantages of convolutional sparse coding by directly processing the entire image to extract features,and reduces the artifacts from dictionary learning caused by image block aggregation.Then,the total variation regularization term is introduced to enhance the constraints of the image domain,which can effectively further suppress the noise.By comparing several sets of sparse view reconstruction experiments with different algorithms,the experimental results show that the proposed algorithm has excellent performance in noise suppression,artifact reduction and image detail restoration.