Improved Fascicle Length Estimates From Ultrasound Using a U-net-LSTM Framework

Overview

Journal IEEE Int Conf Rehabil Robot

Specialties Biotechnology
Rehabilitation Medicine

Date 2023 Nov 27

PMID 38010923

Authors

Letizia Gionfrida

Richard W Nuckols

Conor J Walsh

Robert D Howe

Affiliations

Soon will be listed here.

Abstract

Brightness-mode (B-mode) ultrasound has been used to measure in vivo muscle dynamics for assistive devices. Estimation of fascicle length from B-mode images has now transitioned from time-consuming manual processes to automatic methods, but these methods fail to reach pixel-wise accuracy across extended locomotion. In this work, we aim to address this challenge by combining a U-net architecture with proven segmentation abilities with an LSTM component that takes advantage of temporal information to improve validation accuracy in the prediction of fascicle lengths. Using 64,849 ultrasound frames of the medial gastrocnemius, we semi-manually generated ground-truth for training the proposed U-net-LSTM. Compared with a traditional U-net and a CNNLSTM configuration, the validation accuracy, mean square error (MSE), and mean absolute error (MAE) of the proposed U-net-LSTM show better performance (91.4%, MSE =0.1± 0.03 mm, MAE =0.2± 0.05 mm). The proposed framework could be used for real-time, closed-loop wearable control during real-world locomotion.

References

Zhao H, Zhang L . Automatic tracking of muscle fascicles in ultrasound images using localized Radon transform. IEEE Trans Biomed Eng. 2011; 58(7):2094-101. DOI: 10.1109/TBME.2011.2144593. View

Drazan J, Hullfish T, Baxter J . An automatic fascicle tracking algorithm quantifying gastrocnemius architecture during maximal effort contractions. PeerJ. 2019; 7:e7120. PMC: 6611451. DOI: 10.7717/peerj.7120. View

Zhou Y, Zheng Y . Estimation of muscle fiber orientation in ultrasound images using revoting hough transform (RVHT). Ultrasound Med Biol. 2008; 34(9):1474-81. DOI: 10.1016/j.ultrasmedbio.2008.02.009. View

Lai A, Lichtwark G, Schache A, Lin Y, Brown N, Pandy M . In vivo behavior of the human soleus muscle with increasing walking and running speeds. J Appl Physiol (1985). 2015; 118(10):1266-75. DOI: 10.1152/japplphysiol.00128.2015. View

Bohm S, Mersmann F, Santuz A, Arampatzis A . The force-length-velocity potential of the human soleus muscle is related to the energetic cost of running. Proc Biol Sci. 2019; 286(1917):20192560. PMC: 6939913. DOI: 10.1098/rspb.2019.2560. View

Muramatsu T, Muraoka T, Kawakami Y, Shibayama A, Fukunaga T . In vivo determination of fascicle curvature in contracting human skeletal muscles. J Appl Physiol (1985). 2001; 92(1):129-34. DOI: 10.1152/jappl.2002.92.1.129. View

Cunningham R, Loram I . Estimation of absolute states of human skeletal muscle via standard B-mode ultrasound imaging and deep convolutional neural networks. J R Soc Interface. 2020; 17(162):20190715. PMC: 7014797. DOI: 10.1098/rsif.2019.0715. View

Kurokawa S, Fukunaga T, Fukashiro S . Behavior of fascicles and tendinous structures of human gastrocnemius during vertical jumping. J Appl Physiol (1985). 2001; 90(4):1349-58. DOI: 10.1152/jappl.2001.90.4.1349. View

Nuckols R, Lee S, Swaminathan K, Orzel D, Howe R, Walsh C . Individualization of exosuit assistance based on measured muscle dynamics during versatile walking. Sci Robot. 2021; 6(60):eabj1362. PMC: 9052350. DOI: 10.1126/scirobotics.abj1362. View

10.

Loram I, Maganaris C, Lakie M . Use of ultrasound to make noninvasive in vivo measurement of continuous changes in human muscle contractile length. J Appl Physiol (1985). 2005; 100(4):1311-23. DOI: 10.1152/japplphysiol.01229.2005. View

11.

van der Zee T, Kuo A . TimTrack: A drift-free algorithm for estimating geometric muscle features from ultrasound images. PLoS One. 2022; 17(3):e0265752. PMC: 8947026. DOI: 10.1371/journal.pone.0265752. View

12.

Dick T, Arnold A, Wakeling J . Quantifying Achilles tendon force in vivo from ultrasound images. J Biomech. 2016; 49(14):3200-3207. PMC: 5074891. DOI: 10.1016/j.jbiomech.2016.07.036. View

13.

Van Hooren B, Teratsias P, Hodson-Tole E . Ultrasound imaging to assess skeletal muscle architecture during movements: a systematic review of methods, reliability, and challenges. J Appl Physiol (1985). 2020; 128(4):978-999. DOI: 10.1152/japplphysiol.00835.2019. View

14.

Farris D, Lichtwark G . UltraTrack: Software for semi-automated tracking of muscle fascicles in sequences of B-mode ultrasound images. Comput Methods Programs Biomed. 2016; 128:111-8. DOI: 10.1016/j.cmpb.2016.02.016. View

15.

Rosa L, Zia J, Inan O, Sawicki G . Machine learning to extract muscle fascicle length changes from dynamic ultrasound images in real-time. PLoS One. 2021; 16(5):e0246611. PMC: 8153491. DOI: 10.1371/journal.pone.0246611. View

16.

Hochreiter S, Schmidhuber J . Long short-term memory. Neural Comput. 1997; 9(8):1735-80. DOI: 10.1162/neco.1997.9.8.1735. View

17.

Zhou G, Chan P, Zheng Y . Automatic measurement of pennation angle and fascicle length of gastrocnemius muscles using real-time ultrasound imaging. Ultrasonics. 2014; 57:72-83. DOI: 10.1016/j.ultras.2014.10.020. View

18.

Donahue J, Hendricks L, Rohrbach M, Venugopalan S, Guadarrama S, Saenko K . Long-Term Recurrent Convolutional Networks for Visual Recognition and Description. IEEE Trans Pattern Anal Mach Intell. 2016; 39(4):677-691. DOI: 10.1109/TPAMI.2016.2599174. View

19.

Gionfrida L, Nuckols R, Walsh C, Howe R . Age-Related Reliability of B-Mode Analysis for Tailored Exosuit Assistance. Sensors (Basel). 2023; 23(3). PMC: 9921922. DOI: 10.3390/s23031670. View

20.

Ritsche P, Franchi M, Faude O, Finni T, Seynnes O, Cronin N . Fully Automated Analysis of Muscle Architecture from B-Mode Ultrasound Images with DL_Track_US. Ultrasound Med Biol. 2023; 50(2):258-267. DOI: 10.1016/j.ultrasmedbio.2023.10.011. View