Recognition of the physiological actions of the triphasic EMG pattern by a dynamic recurrent neural network

Abstract

Triphasic electromyographic (EMG) patterns with a sequence of activity in agonist (AG1), antagonist (ANT) and again in agonist (AG2) muscles are characteristic of ballistic movements. They have been studied in terms of rectangular pulse-width or pulse-height modulation. In order to take into account the complexity of the EMG signal within the bursts, we used a dynamic recurrent neural network (DRNN) for the identification of this pattern in subjects performing fast elbow flexion movements. Biceps and triceps EMGs were fed to all 35 fully-connected hidden units of the DRNN for mapping onto elbow angular acceleration signals. DRNN training was supervised, involving learning rule adaptations of synaptic weights and time constants of each unit. We demonstrated that the DRNN is able to perfectly reproduce the acceleration profile of the ballistic movements. Then we tested the physiological plausibility of all the networks that reached an error level below 0.001 by selectively increasing the amplitude of each burst of the triphasic pattern and evaluating the effects on the simulated accelerating profile. Nineteen percent of these simulations reproduced the physiological action classically attributed to the 3 EMG bursts: AG1 increase showed an increase of the first accelerating pulse, ANT an increase of the braking pulse and AG2 an increase of the clamping pulse. These networks also recognized the physiological function of the time interval between AG1 and ANT, reproducing the linear relationship between time interval and movement amplitude. This task-dynamics recognition has implications for the development of DRNN as diagnostic tools and prosthetic controllers.

Toward an Integrative Dynamic Recurrent Neural Network for Sensorimotor Coordination Dynamics

“…We have also demonstrated that the DRNN is able to reproduce the major parameters of human limb kinematics such as: (1) fast drawing of complex figure by the upper limb , Draye et al 1997, (2) whole-body straightening (Draye et al 2002), (3) lower limb movement in human locomotion (Cheron et al 2003;Cheron et al 2006) and (4) pointing ballistic movement (Cheron et al 2007). In all of these experimental situations we have found that the attractor states reached through DRNN learning correspond to biologically interpretable solutions (Cheron et al 2007). It was recently demonstrated that neural network with continuous attractors might symbolically represent context-dependent retrieval of short-term memory from long-term memory in the brain (Tsuboshita and Okamoto, 2007).…”

“…This rhythmical organization could be viewed as resulting from a motor binding process (Sanes and Truccolo, 2003) supported by the synchronization of cortical neurons forming functional assemblies in the premotor and primary motor cortex (Jackson et al 2003;Hatsopoulos et al 2003Hatsopoulos et al , 2007Rubino et al 2006). Moreover, the timing of the antagonist EMG burst is pre-programmed by the cerebellum and our DRNN has been able to learn and reproduce such aspect of motor control (Cheron et al 2007). This will permit to use the antagonist EMG burst to stop adequately a prosthetic movement.…”

“…A major interest of the EMGs to kinematics mapping by a DRNN is that it may represent a new indirect way for a better understanding of motor organization elaborated by the central nervous system. After the learning phase and whatever the type of movement, the identification performed by the DRNN offers a dynamic memory which is able to recognize the preferential direction of the physiological action of the studied muscles (Cheron et al , 2003(Cheron et al , 2006(Cheron et al , 2007. In this chapter, we present the DRNN structure and the training procedure applied in case of noisy biological signals.…”

Biological Signals Identification by a Dynamic Recurrent Neural Network: from Oculomotor Neural Integrator to Complex Human Movements and Locomotion

“…We have to note that in this case the kinematics data are given by the position signals of the index finger (the outputs of the DRNN were the vertical and the horizontal position of the index). In a second set of studies (Cheron et al, 2007) the subjects were asked to perform 'as fast as possible' flexion movements of the elbow in the vertical plane. In this case the angular acceleration of the elbow was used as the output.…”

Physiological modules for generating discrete and rhythmic movements: action identification by a dynamic recurrent neural network

“…DRNNs are recognized as universal approximators of dynamical systems (Kuan and Hornik, 1991; Doya, 1996; Yi et al, 2006; Tani et al, 2008; Bicho et al, 2011; Laje and Buonomano, 2013) and the attractor states reached through DRNN learning of EMG-to-kinematic patterns correspond to biologically interpretable solutions (Cheron et al, 1996, 2003, 2006, 2007, 2011; Song and Tong, 2005; Liu and Buonomano, 2009). After the learning phase, the identification performed by the DRNN offers a dynamic memory which has been used, for example, to recognize the physiological preferred direction of action for the studied muscles (Cheron et al, 1996, 2003, 2006, 2007).…”