任佳豪,李健,李晋,曾周末,陈世利,刘洋.基于对比源与全波形反演的肌骨组织定量超声成像[J].电子测量与仪器学报,2026,40(5):15-29
基于对比源与全波形反演的肌骨组织定量超声成像
Quantitative ultrasound imaging of musculoskeletal tissue based on contrastsource inversion and full waveform inversion
  
DOI:
中文关键词:  肌骨组织成像  对比源反演  全波形反演  成像加速
英文关键词:musculoskeletal tissue imaging  contrast source inversion  full waveform inversion  imaging acceleration
基金项目:国家自然科学基金(12474459)项目资助
作者单位
任佳豪 天津大学精密仪器与光电子工程学院天津300072 
李健 天津大学精密仪器与光电子工程学院天津300072 
李晋 天津大学精密仪器与光电子工程学院天津300072 
曾周末 天津大学精密仪器与光电子工程学院天津300072 
陈世利 天津大学精密仪器与光电子工程学院天津300072 
刘洋 天津大学精密仪器与光电子工程学院天津300072 
AuthorInstitution
Ren Jiahao School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China 
Li Jian School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China 
Li Jin School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China 
Zeng Zhoumo School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China 
Chen Shili School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China 
Liu Yang School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China 
摘要点击次数: 252
全文下载次数: 140
中文摘要:
      超声断层成像技术因具有低成本、实时性和便携性等优势,在肌骨组织成像中具有重要应用潜力。然而,肌骨组织中软组织与骨骼之间显著的声阻抗对比会引发强反射、多次散射及明显衰减,导致超声波场高度复杂,从而限制了传统超声成像方法的分辨率和定量精度。针对上述问题,提出了一种融合对比源反演与全波形反演的定量超声成像方法。该方法首先引入乘法正则化对比源反演,在强散射条件下快速重建组织的大尺度声速分布,获得物理一致且鲁棒的初始模型;随后在细网格上采用频域双参数全波形反演,实现声速与衰减参数的联合精细重建,并通过GPU并行加速提高计算效率。数值仿真与离体牛肌骨实验结果表明,在0.3 mm空间网格间距下,该方法能够有效抑制高阻抗界面引起的非线性效应与成像伪影,实现对肌骨组织声速与衰减分布的精细定量成像,整体反演时间为16.53 min。重建声速与衰减图像的结构相似性指数分别达到0.954 0和0.902 3。结果表明,该方法在成像精度与计算效率之间取得了良好平衡,为肌骨组织的高分辨率定量超声成像提供了一种高效、稳健的技术方案。
英文摘要:
      Ultrasound computed tomography has great application potential in musculoskeletal imaging owing to its advantages of low cost, real-time capability, and portability. However, the pronounced acoustic impedance contrast between soft tissue and bone in musculoskeletal structures gives rise to strong reflections, multiple scattering, and significant attenuation, resulting in a highly complex ultrasound wavefield and thereby limiting the spatial resolution and quantitative accuracy of conventional ultrasound imaging methods. To address the above issue, a quantitative ultrasound imaging method integrating contrast source inversion and full waveform inversion is proposed. This method first introduces multiplicative-regularized contrast source inversion to rapidly reconstruct the large-scale sound speed distribution of tissues under strong scattering conditions, yielding a physically consistent and robust initial model; subsequently, frequency-domain dual-parameter full waveform inversion is performed on a fine grid to achieve joint high-resolution reconstruction of sound speed and attenuation parameters, with computational efficiency enhanced through GPU-based parallel acceleration. Numerical simulations and ex vivo bovine musculoskeletal experiments demonstrate that, with a spatial grid spacing of 0.3 mm, the proposed method can effectively suppress nonlinear effects and imaging artifacts induced by high-impedance interfaces, enabling refined quantitative imaging of sound speed and attenuation distributions in musculoskeletal tissues, with a total inversion time of 16.53 min. The structural similarity indices of the reconstructed sound speed and attenuation images reach 0.954 0 and 0.902 3, respectively. The results indicate that the proposed method achieves a favorable balance between imaging accuracy and computational efficiency, providing an efficient and robust technical solution for high-resolution quantitative ultrasound imaging of musculoskeletal tissues.
查看全文  查看/发表评论  下载PDF阅读器