基于先验信息感知学习的能谱CT及物质定量智能成像算法  

A semi-supervised material quantitative intelligent imaging algorithm for spectral CT based on prior information perception learning

作　　者：段政 李丹阳 曾栋[1,2] 边兆英 马建华[1,2] DUAN Zheng;LI Danyang;ZENG Dong;BIAN Zhaoying;MA Jianhua(School of Biomedical Engineering,Southern Medical University,Guangzhou 510515,China;Guangzhou Key Laboratory of Medical Radioimaging and Detection Technology,Guangzhou 510515,China)

机构地区：[1]南方医科大学生物医学工程学院,广东广州510515 [2]广州市医用放射成像与检测技术重点实验室,广东广州510515

出　　处：《南方医科大学学报》2023年第4期620-630,共11页Journal of Southern Medical University

基　　金：国家自然科学基金(U21A6005);数学天元基金项目(12226004)。

摘　　要：目的提出一种基于先验信息感知学习的能谱CT半监督物质定量智能成像算法(SLMD-Net),以提升能谱CT及物质定量成像精度和质量,并降低数据驱动网络对标签数据的依赖性。方法算法框架包括监督子模块和自监督子模块。在监督子模块中,基于少量标签数据和均方误差损失函数学习构建从低信噪比数据到高信噪比数据的映射关系;在自监督子模块中,针对大量无标签低信噪比基物质图像数据,采用基于图像恢复模型构建损失函数,并纳入基物质图像数据的先验信息,以全变分(TV)模型刻画图像的先验信息。两个子模块合并构成SLMD-Net并通过临床仿真数据评估可行性和有效性。结果与模型驱动的物质定量成像方法(FBP-DI、PWLS-PCG、E3DTV),数据驱动的物质定量成像方法,如基于监督学习的物质定量成像方法(SUMD-Net和BFCNN),基于无监督学习的物质定量成像方法UNTV-Net以及基于半监督的循环一致性生成对抗网络(Semi-CycleGAN)相比,SLMD-Net在视觉和定量评估上均有明显优势,如在水物质定量成像结果和骨物质定量成像结果中,SLMDNet获得最高的PSNR指标(31.82和29.06)、最高的FSIM指标(0.95和0.90)以及最低的RMSE指标(0.03和0.02),且图像质量评分与其他7种对比方法分解性能的差异具有统计学意义(P<0.05)。SLMD-Net的物质定量性能可接近于使用两倍数量级标签数据训练的SUMD-Net。结论少量标签数据和大量无标签低信噪比基物质图像数据可被充分利用训练网络,有效抑制能谱CT基物质分解过程中产生的强噪声伪影,降低数据驱动网络对标签数据的依赖性,具有更广阔的临床应用前景。Objective To propose a semi-supervised material quantitative intelligent imaging algorithm based on prior information perception learning(SLMD-Net)to improve the quality and precision of spectral CT imaging.Methods The algorithm includes a supervised and a self-supervised submodule.In the supervised submodule,the mapping relationship between low and high signal-to-noise ratio(SNR)data was constructed through mean square error loss function learning based on a small labeled dataset.In the self-supervised sub-module,an image recovery model was utilized to construct the loss function incorporating the prior information from a large unlabeled low SNR basic material image dataset,and the total variation(TV)model was used to to characterize the prior information of the images.The two submodules were combined to form the SLMD-Net method,and pre-clinical simulation data were used to validate the feasibility and effectiveness of the algorithm.Results Compared with the traditional model-driven quantitative imaging methods(FBP-DI,PWLS-PCG,and E3DTV),data-driven supervised-learning-based quantitative imaging methods(SUMD-Net and BFCNN),a material quantitative imaging method based on unsupervised learning(UNTV-Net)and semi-supervised learning-based cycle consistent generative adversarial network(Semi-CycleGAN),the proposed SLMD-Net method had better performance in both visual and quantitative assessments.For quantitative imaging of water and bone materials,the SLMD-Net method had the highest PSNR index(31.82 and 29.06),the highest FSIM index(0.95 and 0.90),and the lowest RMSE index(0.03 and 0.02),respectively and achieved significantly higher image quality scores than the other 7 material decomposition methods(P<0.05).The material quantitative imaging performance of SLMD-Net was close to that of the supervised network SUMD-Net trained with labeled data with a doubled size.Conclusions A small labeled dataset and a large unlabeled low SNR material image dataset can be fully used to suppress noise amplification and artifacts in

关 键 词：能谱CT 基物质分解 半监督学习 U-Net 全变分 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]