A Deep Learning Pipeline for Identification of Motor Units in Musculoskeletal Ultrasound

Ali, Hazrat; Umander, Johannes; Rohlén, Robin; Grönlund, Christer

doi:10.1109/access.2020.3023495

Cited by 18 publications

(16 citation statements)

References 42 publications

(58 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, had we used iEMG we would detect a smaller number of individual units within close range of the needle tip (Lowery, Weir and Kuiken, 2006), and the needle (for the case of needle recordings) would cause mechanical coupling disrupting the results. In principle this can be solved by directly decomposing the US in MU discharges (Ali, Umander and Rohlén, 2020; Rohlén et al ., 2020; Rohlén, Stålberg and Grönlund, 2020), however further validation on the basic assumptions underlying the mechanical components and coupling of MU twitches is necessary.…”

Section: Discussionmentioning

confidence: 99%

Kinematics of individual muscle units in natural contractions measured in vivo using ultrafast ultrasound

Lubel

Sgambato

Barsakcioglu

et al. 2022

Preprint

View full text Add to dashboard Cite

The study of human neuromechanical control at the motor unit (MU) level has predominantly focussed on electrical activity and force generation, whilst the link between these, the muscle deformation, has not been widely studied. Here, we describe a methodology utilising ultrafast ultrasound (US), allowing imaging of up to tens of thousands of frames per second, to measure the deformation of the muscular tissue due to individual MU twitches for a population of active MUs during voluntary contractions. We used the spiking activity of MUs decomposed from high-density surface electromyography recordings of the tibialis anterior muscle to guide the analysis of simultaneously recorded ultrafast US. With a novel analysis on the US images we identified, with high spatio-temporal precision, the velocity maps associated with single MU movements. From the individual MU profiles obtained from the velocity maps, the region of movement, the duration of the mechanical twitch, the total and active contraction times, and the activation time were computed. The latter features, the temporal features, showed high repeatability across different force levels. The former feature, the spatial feature, showed high consistency across force levels, however the complicated dynamics of the muscle motion resulted in morphing and translation of these regions. Furthermore, the experimental measures provided the first evidence of muscle unit twisting during voluntary contractions. The proposed approach allows, for the first time, non-invasive recordings of muscle deformation due to individual MU activations during voluntary contractions.Key pointsWe identified the activity of single motor units (MUs) from high-density surface electromyography (HDsEMG) and used this information in combination with ultrafast ultrasound to extract local muscle motion due to the contraction of individual muscle unitsMultiple MUs, including those with fibres overlapping in space, can be simultaneously and individually detected using this techniqueThe proposed method allows us to measure both the spiking activity of motor units and their movement within the muscles concomitantlyThe technique allows for populations of MUs to be tracked and monitored in the electrical and mechanical domains simultaneously and non-invasively during natural contractions, thus achieving a high spatio-temporal resolution in the characterization of MU behaviour

show abstract

Section: Discussionmentioning

confidence: 99%

Kinematics of individual muscle units in natural contractions measured in vivo using ultrafast ultrasound

Lubel

Sgambato

Barsakcioglu

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…The recent advent of imaging techniques for motor unit identification and quantification [ 3 – 9 ] utilize dynamics of intra-muscular contraction patterns, and the model may assist in further development of imaging methods. For example, the simulation model could be used for data augmentation and training of deep learning models for MU identification such as in Ali et al [ 47 ] when large amounts of training data are required. The initial simplified simulation model provides the labelling of the data and the domain transfer adapts the data to authentic experimental spatio-temporal features.…”

Section: Discussionmentioning

confidence: 99%

Modelling intra-muscular contraction dynamics using in silico to in vivo domain translation

et al. 2022

Self Cite

View full text Add to dashboard Cite

Background Advances in sports medicine, rehabilitation applications and diagnostics of neuromuscular disorders are based on the analysis of skeletal muscle contractions. Recently, medical imaging techniques have transformed the study of muscle contractions, by allowing identification of individual motor units’ activity, within the whole studied muscle. However, appropriate image-based simulation models, which would assist the continued development of these new imaging methods are missing. This is mainly due to a lack of models that describe the complex interaction between tissues within a muscle and its surroundings, e.g., muscle fibres, fascia, vasculature, bone, skin, and subcutaneous fat. Herein, we propose a new approach to overcome this limitation. Methods In this work, we propose to use deep learning to model the authentic intra-muscular skeletal muscle contraction pattern using domain-to-domain translation between in silico (simulated) and in vivo (experimental) image sequences of skeletal muscle contraction dynamics. For this purpose, the 3D cycle generative adversarial network (cycleGAN) models were evaluated on several hyperparameter settings and modifications. The results show that there were large differences between the spatial features of in silico and in vivo data, and that a model could be trained to generate authentic spatio-temporal features similar to those obtained from in vivo experimental data. In addition, we used difference maps between input and output of the trained model generator to study the translated characteristics of in vivo data. Results This work provides a model to generate authentic intra-muscular skeletal muscle contraction dynamics that could be used to gain further and much needed physiological and pathological insights and assess and overcome limitations within the newly developed research field of neuromuscular imaging.

show abstract

“…Previous studies involving the MU-UUS analysis have provided estimates of unfused tetanic signals of single MUs (Ali et al, 2020;Rohlén et al, 2022Rohlén et al, , 2020bRohlén et al, , 2020a. These unfused tetanic estimates have further been used to estimate their spike trains (neural discharges).…”

Section: Introductionmentioning

confidence: 99%

“…The challenge for this spike estimation is that the unfused tetani of voluntary contractions consist of variable successive twitches (Burke et al, 1976;Raikova et al, 2008Raikova et al, , 2007. For the MU-UUS analysis, a narrow band-pass filter has been used before applying peak detection to identify the time instants of the local maxima (Ali et al, 2020;Rohlén et al, 2020bRohlén et al, , 2020a. Although an optimized parameter set may have improved the performance, these studies did not focus on optimizing the parameters (Rohlén et al, 2020b).…”

Section: Introductionmentioning

confidence: 99%

Optimization and comparison of two methods for spike train estimation in an unfused tetanic contraction of low threshold motor units

Rohlén

Antfolk

Grönlund

2022

Preprint

View full text Add to dashboard Cite

Background: Human movement is generated by activating motor units (MUs), i.e., the smallest structures that can be voluntarily controlled. Recent findings have shown imaging of voluntarily activated MUs using ultrafast ultrasound based on displacement velocity images and a decomposition algorithm. Given this, estimates of trains of twitches (unfused tetanic signals) evoked by the neural discharges (spikes) of spinal motor neurons are provided. Based on these signals, a band-pass filter method (BPM) has been used to estimate its spike train. In addition, an improved spike estimation method consisting of a continuous Haar wavelet transform method (HWM) has been suggested. However, the parameters of the two methods have not been optimized, and their performance has not been compared rigorously. Method: HWM and BPM were optimized using simulations. Their performance was evaluated based on simulations and two experimental datasets with 21 unfused tetanic contractions considering their rate of agreement, spike offset, and spike offset variability with respect to the simulated or experimental spikes. Results: A range of parameter sets that resulted in the highest possible agreement with simulated spikes was provided. Both methods highly agreed with simulated and experimental spikes, but HWM was a better spike estimation method than BPM because it had a higher agreement, less bias, and less variation (p < 0.001). Conclusions: The optimized HWM will be an important contributor to further developing the identification and analysis of MUs using imaging, providing indirect access to the neural drive of the spinal cord to the muscle by the unfused tetanic signals.

show abstract

A Deep Learning Pipeline for Identification of Motor Units in Musculoskeletal Ultrasound

Cited by 18 publications

References 42 publications

Kinematics of individual muscle units in natural contractions measured in vivo using ultrafast ultrasound

Kinematics of individual muscle units in natural contractions measured in vivo using ultrafast ultrasound

Modelling intra-muscular contraction dynamics using in silico to in vivo domain translation

Optimization and comparison of two methods for spike train estimation in an unfused tetanic contraction of low threshold motor units

Contact Info

Product

Resources

About