Ultrasonic full-waveform inversion for dynamically monitoring bone micro-structure deterioration in osteoporosis progression
-
摘要:
研究了基于全波反演(FWI)的骨骼超声层析成像方法, 用于动态监测骨质疏松性微结构退化的进程。采用雌性小鼠注射药物建立骨质疏松症模型, 在第0, 2, 4, 6周通过Micro-CT扫描活体小鼠, 重建获得小鼠股骨骨骼结构。以第0周骨骼结构为基准输入模型, 通过不同的超声收发模式(透射、反射及透射−反射双模式), 仿真分析了FWI监测不同骨质疏松进程的骨微结构退化的效果。结果表明, 初始模型为均匀介质(纯水)时, FWI反演失效, 不能准确重建骨骼结构。初始模型中考虑了基准骨骼结构(第0周)时, FWI能准确反演骨骼组织声速(均方根误差(RMSE) < 17 m/s, 平均相对误差(MRE) < 7.2%), 精确重建骨骼结构(结构相关系数(CC) > 0.85)), 因此可以准确监测不同骨质疏松进程(第2, 4, 6周)的骨微结构退化情况。对比不同超声收发模式, 透射−反射双模式FWI监测骨微结构退化的性能优于单一透射或反射FWI监测方法。考虑了基准骨骼模型的FWI可用于动态监测骨微结构退化, 对评估骨质疏松进展具有一定意义。
Abstract:The aim of this study is to investigate the method of ultrasound tomography of bone based on ultrasonic full-waveform inversion (FWI), and to dynamically monitor the progression of osteoporotic micro-structural deterioration. Female mice are injected with 4-vinylcyclohexene diepoxide (VCD) to induce premature ovarian failure and establish the mice osteoporosis model. The mice are scanned by Micro-CT in vivo at different osteoporosis progression (the 0, 2, 4 and 6 weeks), and bone micro-structure in the femurs of mice are reconstructed. Taking the initial bone micro-structure of mice (at the 0 week) as the baseline input model, the effects of FWI on monitoring bone micro-structure deterioration in osteoporosis progression are numerically analyzed, with different ultrasound transmitting and receiving modes (transmission, reflection and transmission-reflection dual mode). The results show that FWI failed to reconstruct bone micro-structure when using the uniform medium (pure water) as the initial model. Compared with the initial model with uniform medium (pure water), the FWI using the baseline bone (at the 0 week) model could accurately monitor the bone micro-structure deterioration in different osteoporotic processes (at the 2, 4, 6 week), the ultrasound velocity inversion is with a root-mean-square error (RMSE) < 17 m/s, mean relative error (MRE) < 7.2%, and structural correlation coefficient (CC) > 0.85. Compared with different ultrasound transmitting and receiving modes, the combination of transmission-reflection dual-mode in FWI is significantly superior to the single transmission or reflection FWI in monitoring bone micro-structural deterioration. This study demonstrate that FWI using the baseline bone model can be used to dynamically monitor bone micro-structural deterioration and might have some significance in assessing the progression of osteoporosis.
-
表 1 骨骼组织声速反演结果的定量指标
超声收发模式指标 初始模型为均匀介质(纯水) 初始模型为基准骨骼结构 第2周 第4周 第6周 第2周 第4周 第6周 透射模式 RMSE (m/s) 49.1 48.8 54.4 16.4 13.9 12.2 MRE (%) 23.1 21.7 20.9 7.1 6.1 5.3 CC −0.02 0.01 0.00 0.85 0.88 0.91 反射模式 RMSE (m/s) 33.9 32.1 35.0 16.0 14.4 12.1 MRE (%) 15.0 15.0 14.0 7.0 6.4 5.3 CC 0.17 0.28 0.21 0.86 0.88 0.91 透射 − 反射双模式 RMSE (m/s) 43.9 40.7 33.1 15.4 13.3 12.0 MRE (%) 19.2 18.3 15.0 5.9 1.6 1.5 CC 0.29 0.23 0.20 0.91 0.92 0.94 -
[1] Compston J E, McClung M R, Leslie W D. Osteoporosis. Lancet, 2019; 393: 364—376 doi: 10.1016/S0140-6736(18)32112-3 [2] 夏维波. 骨质疏松症防治中的难点和应对. 国内分泌代谢杂志, 2022; 42(3): 165—169 doi: 10.3760/cma.j.cn121383-20220406-04011 [3] 宋小军, 他得安, Petro Moilanen. 采用超声导波特征参量的骨质疏松识别方法. 声学学报, 2017; 42(5): 517—576 doi: 10.15949/j.cnki.0371-0025.2017.05.007 [4] 第五强强, 李博艺, 李颖, 等. 有监督学习的超声背散射方法在骨质评价中的应用. 声学学报, 2019; 44(5): 818—825 doi: 10.15949/j.cnki.0371-0025.2019.05.002 [5] 刘洋, 郭霞生, 章东, 等. 基于时间反转的骨裂纹超声成像模拟研究. 声学学报, 2011; 36(2): 179—184 doi: 10.15949/j.cnki.0371-0025.2011.02.021 [6] 王文斌, 郭佳林, 他得安, 等. 超声表观积分背散射系数评价新生儿骨状况的可行性研究. 声学学报, 2012; 37(2): 177—180 doi: 10.15949/j.cnki.0371-0025.2012.02.007 [7] Langton C, Palmer S, Porter R. The Measurement of broadband ultrasonic attenuation in cancellous bone. Eng. Med., 1984; 13: 89—91 doi: 10.1243/EMED_JOUR_1984_013_022_02 [8] Mami M. Bone ultrasound. Jpn. J. Appl. Phys., 2019; 58: SG0802 doi: 10.7567/1347-4065/ab0dfa [9] Laugier P, Haïat G. Bone quantitative ultrasound. Springer, 2011: 576 [10] Sandhu G Y, Li C, Roy O, et al. Frequency domain ultrasound waveform tomography: breast imaging using a ring transducer. Phys. Med. Biol., 2015; 60(14): 5381 doi: 10.1088/0031-9155/60/14/5381 [11] Bernard S, Monteiller V, Komatitsch D, et al. Ultrasonic computed tomography based on full-waveform inversion for bone quantitative imaging. Phys. Med. Biol., 2017; 62(17): 7011 doi: 10.1088/1361-6560/aa7e5a [12] Wiskin j, Malik B, Borup D, et al. Full wave 3D inverse scattering transmission ultrasound tomography in the presence of high contrast. Sci. Rep., 2020; 10(1): 1—14 doi: 10.1038/s41598-019-56847-4 [13] Li Y, Shi Q, Li Y, et al. High-resolution bone microstructure imaging based on ultrasonic frequency-domain full-waveform inversion. Chin. Phys. B, 2021; 30(1): 014302 doi: 10.1088/1674-1056/abc7aa [14] Guasch L, Calderón Agudo O, Tang M X, et al. Full-waveform inversion imaging of the human brain. NPJ Digit. Med., 2020; 3(1): 1—12 doi: 10.1038/s41746-020-0240-8 [15] 张晓毓. 超声全波形反演在骨定量测量中的应用研究. 博士学位论文, 武汉: 武汉大学, 2018 [16] Suo M, Zhang D, Yang Y. Application of an improved ultrasound full-waveform inversion in bone quantitative measurement. Symmetry, 2021; 13(2): 260 doi: 10.3390/sym13020260 [17] 徐晶晶, 沈力行, 赵改平, 等. 基于数字图像处理技术测量松质骨孔隙率的方法. 中国组织工程研究与临床康复, 2010; 14(17): 3062—3064 doi: 10.3969/j.issn.1673-8225.2010.17.006 [18] Xu W, Yuan M, Xuan W, et al. Quantitative inspection of complex-shaped parts based on ice-coupled ultrasonic full waveform inversion technology. Appl. Sci., 2021; 11(10): 4433 doi: 10.3390/app11104433 [19] Hustedt B, Operto S, Virieux J. Mixed-grid and staggered-grid finite-difference methods for frequency-domain acoustic wave modelling. Geophys. J. Int., 2004; 157(3): 1269—1296 doi: 10.1111/j.1365-246X.2004.02289.x [20] Byrd R H, Nocedal J, Schnabel R B, et al. Representations of quasi-Newton matrices and their use in limited memory methods. Math. Program., 1994; 63: 129—156 doi: 10.1007/BF01582063 [21] Sirgue L, Pratt R G. Efficient waveform inversion and imaging: A strategy for selecting temporal frequencies. Geophysics, 2004; 69(1): 231—248 doi: 10.1190/1.1649391 [22] Erlangga Y A. Advances in iterative methods and preconditioners for the Helmholtz equation. Arch. Comput. Methods Eng., 2008; 15: 37—66 doi: 10.1007/s11831-007-9013-7 [23] 孙文瑜, 徐成贤, 朱德通. 最优化方法. 北京: 高等教育出版社, 2004 [24] Virieux J, Operto S. An overview of full-waveform inversion in exploration geophysics. Geophysics, 2009; 74(6): 26 doi: 10.1190/1.3238367 [25] 张麒, 汪源源, 王威琪, 等. 血管内超声图像的仿真. 声学学报, 2008; 33(6): 512—519 doi: 10.3321/j.issn:0371-0025.2008.06.005 [26] 王连坤, 方伍宝, 段心标, 等. 全波形反演初始模型建立策略研究综述. 地球物理学进展, 2016; 31(4): 1678—1687 doi: 10.6038/pg20160436 [27] Lucka F, Pérez-liva M, Treeby B E, et al. High resolution 3D ultrasonic breast imaging by time-domain full waveform inversion. Inverse Probl., 2021; 38(2): 025008 doi: 10.1088/1361-6420/ac3b64 -