SEEDS, simultaneous recordings of high-density EMG and finger joint angles during multiple hand movements

Abstract

We present the SurfacE Electromyographic with hanD kinematicS (SEEDS) database. It contains electromyographic (EMG) signals and hand kinematics recorded from the forearm muscles of 25 non-disabled subjects while performing 13 different movements at normal and slow-paced speeds. EMG signals were recorded with a high-density 126-channel array centered on the extrinsic flexors of the fingers and 8 further electrodes placed on the extrinsic extensor muscles. A data-glove was used to record 18 angles from the joints of the wrist and fingers. The correct synchronisation of the data-glove and the EMG was ascertained and the resulting data were further validated by implementing a simple classification of the movements. These data can be used to test experimental hypotheses regarding EMG and hand kinematics. Our database allows for the extraction of the neural drive as well as performing electrode selection from the high-density EMG signals. Moreover, the hand kinematic signals allow the development of proportional methods of control of the hand in addition to the more traditional movement classification approaches.

Muscle-Specific High-Density Electromyography Arrays for Hand Gesture Classification

“…Based on the performance of the proposed electrode arrays, it is anticipated that the use of these EMG electrodes on the hand could contribute to electrophysiological research, ranging from the complex innervation pathways necessary for synergic muscle coordination and plasticity [44], motor unit decomposition during hand motions [45], to augment or validate complex chemo-electromechanical skeletal muscle or EMG models [46]. Future studies will aim to incorporate hand kinematics recorded of subjects' hands while performing movements to validate HD-EMG data, as well as integrating extrinsic hand muscle recordings to increase the quantity and quality of decoded motions [47].…”

Hand-movement Prediction from EMG with LSTM-Recurrent Neural Networks

“…In this study, we use EMG data and joint angles from the first 7 participants in the SEEDS database [12]. Eight monopolar electrodes are used to record the signals of the extensor muscles of the forearm, while a set of 126 electrodes arranged in an high-density EMG (HD-sEMG) array of 9 rows by 14 columns is used for recording the signals from the flexor muscles.…”

“…-One with 17 inputs (which correspond to the VAR feature computed for each of the 17 electrodes used), an LSTM hidden layer with 9 neurons, and 18 outputs (one for each DoF); -One with 51 inputs (where we extracted MAV, VAR and ZC from each of the 17 electrodes), an LSTM hidden layer with 9 neurons, and 18 outputs (as above). We trained individual models for each of the first 7 participants in the SEEDS dataset [12], using the first two sessions for training (80%) and validation (20%), and the third one for testing. Each of the sessions includes 6 repetitions at different speeds of 13 different movements, performed in random order.…”

Toward Robust, Adaptiveand Reliable Upper-Limb Motion Estimation Using Machine Learning and Deep Learning–A Survey in Myoelectric Control

“…To this end, several benchmark datasets have been shared online, saving other researchers a considerable amount the time and providing effective platforms for model comparison. Representative works include Ninapro [146], CapgMyo [10], CSL-HDEMG [147], SEEDS [148], HITSIMCO [149], and Hyser [150]. Data were acquired from different experimental protocols, several types of electrodes, healthy participants and amputees, static and dynamic scenarios, and a variety of discrete gestures or continuous movements.…”