Paper Submission & Registration
8th Dutch Bio-Medical Engineering Conference
13:40   Neuro-muscular
Chair: Hans van Dijk
15 mins
Alternate finger tapping: Repeatability, target distance and cueing effects in healthy volunteers
Soma Makai-Bölöni, Eva Thijssen, Geert Jan Groeneveld, Robert-Jan Doll
Abstract: Parkinson‘s disease (PD) is a progressive neurodegenerative disease that affects almost 2% of the population above the age of 65. The current standard treatments remain symptomatic and novel treatments are continuously being developed. To better quantify the effects of medications, fast and objective methods are needed. For example, a (touchscreen-based) alternate finger tapping task is a simple and effective tool for quantifying PD-related motor performance and medication effects. However, there are many variations in the implementation of the AFTT, and no study was found that explicitly compared different configurations. The present study compares various AFTT task configurations in healthy volunteers and the results can subsequently be used for designing clinical trials with PD patients. In this study, four task configurations of the AFTT were tested. The effects of target distance 2.5 cm (using the index and middle fingers) vs 20 cm (using the index finger) and the presence vs absence of visual cues were tested in 14 healthy volunteers. Various clinically relevant parameters were extracted and compared between the configurations. Linear mixed-effects models were used to study the effects of target distance and visual cueing, the within-day and between-day repeatability, and the (potential) sensitivity of the features. We found that visual cueing interferes with the tapping speed, rhythm, and accuracy. Additionally, increasing the inter-tap distance from 2.5 cm to 20 cm, resulted in subjects tapping significantly less frequently, with increased tapping errors, and improved rhythm. Additionally, several interaction effects were present between distance and cueing. Of all features, the rhythm, the total number of taps, and tapping accuracy were found to have the highest repeatability and sensitivity. The findings suggest that cueing tapping tasks can be undesirable for future clinical studies involving PD patients. Previous literature suggests that tapping speed and rhythm are clinically relevant predictors of dopaminergic medication effects. Hence, allowing these parameters to vary as freely as possible is crucial to efficiently capture medication effects while minimizing experimental noise. Lastly, as the within-day repeatability is good to excellent, there is no apparent need to extensively train patients before performing the tasks.
15 mins
Towards closed-loop HD-EMG-driven trans-spinal electrical stimulation of motor circuitries
Antonio Gogeascoechea, Alexander Kuck, Edwin van Asseldonk, Francesco Negro, Jan Buitenweg, Utku Yavuz, Massimo Sartori
Abstract: BACKGROUND: Trans-spinal direct current stimulation (tsDCS) is a non-invasive technique to restore neuromuscular function after neurological injuries. Current modelling and experimental evidence suggest global neuro-modulatory effects of tsDCS on spinal excitability. However, it remains unclear how individual motor-neurons (MNs) respond to electrical fields and how tsDCS-induced neural modulation affects the musculoskeletal system. To bridge this gap, we propose a comprehensive neuro-musculoskeletal framework based on: a signal-based strategy for interfacing with spinal cord alpha-MNs undergoing tsDCS1 and MN-driven musculoskeletal modelling2. METHODS: Four healthy individuals and four incomplete spinal cord injured patients undergoing tsDCS performed isometric dorsi-plantar flexion sub-maximal contractions. We recorded high-density electromyography from lower leg muscles and employed convolutive blind source separation to decompose underlying alpha-MNs discharges. We first assessed the quality of the decomposed MN spike trains to automatically remove those with poor quality. Second, we computed interspike train coherence to estimate the strength of common synaptic input (CSI) whether this modulated in response to tsDCS. To link the decoded spike trains with the mechanical output, we extended a current activation dynamics representation from a single neural transformation (to muscle) into multiple MN-specific formulations. For this, we discriminated MN types based on their individual properties, such as discharge rate and recruitment threshold, and mapped them into force-generation processes in the musculoskeletal model. RESULTS: The quality selection improved the correlation between neural drive and force (from z=0.58 to z=0.72). The coherence areas and peaks showed a consistent decrease after tsDCS with respect to the pre-stimulation condition (p<0.05 and p<0.1). Preliminary results of the MN-specific formulation showed continuous distributions of discharge rates and recruitment thresholds. The data projection on the linear combination of both features depicted two (overlapping) MN populations. CONCLUSIONS: We proposed a novel signal-based methodology to infer tsDCS effects at the spinal level. The improved correlation after quality selection indicates that no relevant neuro-mechanical information was lost. Coherence analyses suggest that the strength of CSI is modulated in response to electrical stimulation. The MN populations found in the combined component may help build causal relations between the neural and mechanical levels of human movement under electrical stimuli. REFERENCES: 1. Gogeascoechea A, Kuck A, van Asseldonk E, et al. Interfacing With Alpha Motor Neurons in Spinal Cord Injury Patients Receiving Trans-spinal Electrical Stimulation. Front Neurol. 2020;11:493. doi:10.3389/fneur.2020.00493 2. Sartori M, Yavuz U, Farina D. In Vivo Neuromechanics: Decoding Causal Motor Neuron Behavior with Resulting Musculoskeletal Function. Sci Rep. 2017;7(1):1-14. doi:10.1038/s41598-017-13766-6
15 mins
Subject-specific modelling of neuromechanical response to transcutaneous spinal cord stimulation
Rafael Ornelas Kobayashi, Antonio Gogeascoechea Hernandez, Massimo Sartori
Abstract: After neurological impairment, changes of the central nervous system lead to abnormal function in the neuro-musculoskeletal system. Transcutaneous spinal cord stimulation (tSCS) aims to restore its normal function by delivering non-invasive electrical stimulation into the nervous system. However, tSCS currently is performed based on empirical explorations and manual adjustments or electrodes and stimulation parameters, which often results in sub-optimal motor recovery. To maximize the efficacy of tSCS, we propose a subject-specific computational framework able to determine the optimal stimulation pattern for each individual. The proposed framework combines (1) subject-specific biophysical models of the spinal circuitry, (2) High-density electromyogram (HD-EMG) decomposition of in-vivo α-motoneurons (MNs) and (3) finite element method (FEM) models of the human body describing tSCS modulation of MNs behavior. The biophysical model of the spinal circuitry controlling the activity of Tibialis Anterior (TA) muscle was implemented as described by Cisis and Kohn (Cisi and Kohn, 2008). We recorded HD-EMG from TA of a subject performing a motor task where contraction force changed from 0 to 30 and then to 20% of the maximal voluntary contraction, and decoded the firing behavior of in-vivo MNs (Sartori et al., 2017). The smoothed summation of all MNs output (i.e. neural drive) was used as the common input driving the simulated MN pool. We then implemented a subject-specific calibration framework taking as input the difference between in-vivo MNs, decoded from HD-EMG, and simulated MNs, to determine an unique set of model parameters describing each subject’s anatomy. Preliminary results shown that simulated MNs reproduce the firing behavior of in-vivo MNs with high cross-correlation (=0.96) and coherence coefficients (>0.87 in 0 to 10Hz frequency bandwidth). By interfacing the biophysical model of the spine with a FEM model of the human body based on a pre-segmented model from the virtual population library (Christ et al. 2010), our framework demonstrated that tSCS raises the excitability of the MN pool. Although a larger number of subjects is required for validation, preliminary findings suggest that the proposed framework can predict the behavior of in-vivo MNs and their response to tSCS, opening a window towards model-based optimal tSCS therapies.
15 mins
Monitoring heart rate in Parkinson’s disease to distinguish voluntary from involuntary gait arrests
Helena Cockx, Ying Wang, Bas Bloem, Ian Cameron, Richard van Wezel
Abstract: Imagine that you are chopping some carrots in the kitchen and suddenly the doorbell rings. You turn around to go open the door, but your feet are not moving. This symptom, called freezing of gait (FOG), is debilitating in three out of five persons with Parkinson’s disease and is characterized by an alternating tremor of the legs (trembling), very short, shuffling steps (shuffling), or a total absence of limb movements (akinesia). Previous attempts to predict the occurance of FOG using wearable motion sensors on the limbs have not reached their full potential yet, given that akinetic freezes are difficult to distinguish from voluntary stopping. Here, we expand on an interesting observation that heart rate may increase prior to freezing (Maidan et al., 2010) (Mazilu et al., 2015), and confirm that heart rate increases during freezing, including for the akinetic type. In contrast, heart rate decreased during voluntary stop movements. Sixteen persons with Parkinson’s disease completed a trajectory containing narrow passages and turns, both known to trigger freezing of gait episodes. The trajectory was performed with and without a dual cognitive (auditory Stroop task) or motor (carrying a tray) task. Heart rate was monitored with a 3-lead electrocardiogram. In total, 845 freezing episodes were annotated by two independent trained raters based on video recordings (Cohen’s kappa agreement of 0.82). Furthermore, all episodes were annotated for FOG type (trembling, shuffling, or akinesia), trigger (turn or doorway), and dual-task condition (cognitive, motor, or no dual-task). Our results show a similar increase in heart rate as was shown in previous studies, independently of the type, trigger or dual-task condition, including the akinetic type of freezing. In contrast, a clear and steep drop in heart rate could be appreciated immediately after a normal stop movement. The clear contrast in heart rate between voluntary and involuntary (freezing) halt of gait shown in this study, might be promising to detect freezing episodes there where the existing motion-based algorithms are shortcoming, e.g. during akinetic freezing. Future research should clarify whether a simple heart rate monitor, like a smartwatch, can indeed improve the performance of FOG detection algorithms.

end %-->