自适应加权非局部先验贝叶斯统计重构方法提升低剂量 X 射线 CT 图像质量

摘要：低剂量 X 射线计算机断层扫描 (LDCT) 是一种常用的医学成像技术，但由于较低的辐射剂量，其重构质量通常较差。为了提高 LDCT 的重构质量，本文提出了一种基于自适应加权非局部先验的贝叶斯统计重构方法。该方法利用了非局部相似性的特点，通过学习不同区域的权重系数，对重构结果进行自适应加权。同时，本文还引入了贝叶斯统计的思想，对重构过程中的噪声进行建模和处理，进一步提高了重构质量。实验结果表明，所提出的方法可以有效地提高 LDCT 的重构质量，同时保持较低的辐射剂量。\n\n关键词：低剂量 X 射线计算机断层扫描；自适应加权非局部先验；贝叶斯统计重构；重构质量；辐射剂量\n\nAbstract: Low-dose X-ray computed tomography (LDCT) is a commonly used medical imaging technique, but due to the lower radiation dose, the reconstruction quality is usually poor. In order to improve the reconstruction quality of LDCT, this paper proposes a Bayesian statistical reconstruction method based on adaptive weighted non-local prior. This method utilizes the characteristics of non-local similarity, and adaptively weights the reconstruction results by learning the weight coefficients of different regions. At the same time, this paper also introduces the idea of Bayesian statistics to model and process the noise in the reconstruction process, further improving the reconstruction quality. The experimental results show that the proposed method can effectively improve the reconstruction quality of LDCT while maintaining a lower radiation dose.\n\nKeywords: Low-dose X-ray computed tomography; adaptive weighted non-local prior; Bayesian statistical reconstruction; reconstruction quality; radiation dose.