Application of support vector machines in detecting hand grasp gestures using a commercially off the shelf wireless myoelectric armband

The propose of this study was to assess the feasibility of using support vector machines in analysing myoelectric signals acquired using an off the shelf device, the Myo armband from Thalmic Lab, when performing hand grasp gestures. Participants (n = 26) took part in the study wearing the armband an...

Full description

Saved in:
Bibliographic Details
Main Authors: Amirabdollahian, Farshid, Walters, Michael L.
Format: Conference Proceeding
Language:English
Subjects:
Adult
Arm - physiology
Electrodes
Electromyography
Electromyography - methods
Feasibility Studies
Female
Gestures
Hand Strength - physiology
Humans
Kernel
Male
Middle Aged
Pattern Recognition, Automated
Performance evaluation
Signal Processing, Computer-Assisted
Support Vector Machine
Support vector machines
Training
Wrist
Young Adult
Citations: Items that cite this one
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The propose of this study was to assess the feasibility of using support vector machines in analysing myoelectric signals acquired using an off the shelf device, the Myo armband from Thalmic Lab, when performing hand grasp gestures. Participants (n = 26) took part in the study wearing the armband and producing a series of required gestures. Support vector machines were used to train a model using participant training values, and to classify gestures produced by the same participants. Different Kernel functions and electrode combinations were studied. Also we contrasted different lengths of training values versus different lengths for the classification samples. The overall accuracy was 94.9% with data from 8 electrodes, and 72% where only four of the electrodes were used. The linear kernel outperformed the polynomial, and radial basis function. Exploring the number of training samples versus the achieved classification accuracy, results identified acceptable accuracies (> 90%) for training around 2.5s, and recognising grasp with 0.2s of acquired data. The best recognised grasp was the hand closed (97.6%), followed by cylindrical grasp (96.8%), the lateral grasp (93.2%) and tripod (92%). These results allows us to progress to the next stage of work where the Myo armband is used in the context of robot-mediated stroke rehabilitation and also involves more dynamic interactions as well as gross upper arm movements.
ISSN:1945-7901
DOI:10.1109/ICORR.2017.8009231