Real-Time Task Discrimination for Myoelectric Control Employing Task-Specific Muscle Synergies

Abstract

We present a novel formulation that employs task-specific muscle synergies and state-space representation of neural signals to tackle the challenging myoelectric control problem for lower arm prostheses. The proposed framework incorporates information about muscle configurations, e.g., muscles acting synergistically or in agonist/antagonist pairs, using the hypothesis of muscle synergies. The synergy activation coefficients are modeled as the latent system state and are estimated using a constrained Kalman filter. These task-dependent synergy activation coefficients are estimated in real-time from the electromyogram (EMG) data and are used to discriminate between various tasks. The task discrimination is helped by a post-processing algorithm that uses posterior probabilities. The proposed algorithm is robust as well as computationally efficient, yielding a decision with > 90% discrimination accuracy in approximately 3 ms . The real-time performance and controllability of the algorithm were evaluated using the targeted achievement control (TAC) test. The proposed algorithm outperformed common machine learning algorithms for single- as well as multi-degree-of-freedom (DOF) tasks in both off-line discrimination accuracy and real-time controllability (p < 0.01).

EMG-Controlled Transradial Prostheses - An Investigation into Machine Learning Techniques

“…The identified movement intent by the machine learning algorithm is passed on as the control information to the actuation mechanism of the prosthetic device. The performance of these prosthetic devices significantly depends on the classification accuracy of the employed machine learning algorithms [4], [16], [24], [27]. The accuracy of these machine learning algorithms, in turn, may significantly be affected by the choice of the type and the number of features, the dimensionality reduction technique, as well as, the type of the classification algorithm used.…”

“…EMG data from forearm muscles were recorded using wireless probes with an inbuilt preamplifier. The DTS Analog Module further transmitted analog output to an NI-USB 6009 (National Instrument Corporation, Austin, TX, USA) data acquisition card to acquire and digitize the EMG data at the rate of 2000 samples per second [24]. The BioPatRec software was modified to acquire and process the EMG data [20], [24].…”

“…The schematic layout of the experimental setup is given in Fig. 2A Experimental data collection and acquisition of the EMG data from the forearm muscles using TeleMyo DTS, DTS analog module, and a data acquisition module (NI-USB 6009) [24]. The data were recorded using the BioPatRec software at the sampling rate of 2000 Hz [20].…”

“…The EMG signals have been used by researcher and clinicians for decades to investigate muscles in healthy as well as pathological conditions [23], [26], [1]. Recently, owing to the advancements in machine learning and pattern classification techniques, the EMG signals are being used extensively to control prosthetic devices [6], [4], [27], [24], [2]. These prosthetic devices are externally powered and are actuated using the intended movement information produced by these machine learning algorithms [6], [25].…”

“…Mathematical definitions of these features are provided in Table 1. The quantity of the EMG data is generally increased for better training and to improve the classification accuracy of machine learning algorithms [24]. However, as the number of EMG channels, i.e.…”