EMG-Based Feature Extraction and Classification for Prosthetic Hand Control

Azhiri, Reza Bagherian; Esmaeili, Mohammad Ali; Nourani, Mehrdad

doi:10.48550/arxiv.2107.00733

Cited by 4 publications

(11 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is challenging to recognize the six sEMG signals when directly inputting them into a pattern recognition algorithm. 59 Therefore, we first extracted 10 eigenvalues, including the time domain and frequency domain to characterize the integral properties of the detected signals. To demonstrate the effectiveness of the feature extraction process, we used the principal component analysis (PCA) algorithm to perform a dimensionality reduction visual analysis on the extracted 10 eigenvalues.…”

Section: Resultsmentioning

confidence: 99%

A crack compensation strategy for highly stretchable conductors based on liquid metal inclusions

Li¹,

Ma²,

Xu³

et al. 2022

iScience

View full text Add to dashboard Cite

Section: Resultsmentioning

confidence: 99%

A crack compensation strategy for highly stretchable conductors based on liquid metal inclusions

Li¹,

Ma²,

Xu³

et al. 2022

iScience

View full text Add to dashboard Cite

“…(1) The Integrated Absolute of Second Derivative: This mathematical approach calculates the rate of change of a signal, enabling the detection of properties such as signal concavity, inflection points, and changes in curvature. It is particularly useful for identifying features that are less affected by signal noise (Azhiri et al, 2021).…”

Section: Time Domain Featuresmentioning

confidence: 99%

Machine Learning Techniques for Distinguishing Hand Gestures from Forearm Muscle Activity

2023

Proceedings of the XXXVth Annual International Occupational Ergonomics and Safety Conference

View full text Add to dashboard Cite

This research aimed to explore the feasibility of using forearm electromyography (EMG) data for identifying different hand gestures. The study involved 10 participants (7 males and 3 females) who performed eight different hand gestures while their flexor digitorum superficialis (FDS) and extensor digitorum communis (EDC) muscles were recorded. The recorded data was split into 80% for training and 20% for testing. The study compared the effectiveness of Deep Neural Network and Random Forest algorithms based on five different levels of moving window sizes (200, 400, 600, 800, and 1000 milliseconds). A wavelet approach was used to reconstruct the data and remove noise that does not represent the EMG signals accurately. Eighteen features were then extracted from the time and frequency domains of the analysis to characterize the complex signals. The study found that larger moving window sizes had higher accuracy rates than small moving window sizes due to the larger temporal resolution it provides. The Deep Neural Network showed a higher performance with a 1000millisecond stream, achieving an accuracy rate of 97.13%, while the Random Forest algorithm achieved the highest accuracy rate of 96.65% with the same stream duration. In conclusion, the study found that a Deep Neural Network based on a 1000millisecond stream was the most accurate, with an accuracy rate of 97.13%, while a Random Forest based on a 200millisecond stream was the most efficient, with an accuracy rate of 84.77%. Future research could expand the sample size, include more hand gestures, and use different feature extraction methods and modeling algorithms to further improve the accuracy and efficiency of the system.

show abstract

“…Willison Amplitude [27] WAMP [13] IATD [13] IEAV [13] IALV compares the test accuracy of proposed RNN architectures at different signal lengths for the first dataset. The method which is used for the postprocessing is majority voting.…”

Section: Statistical Metric Abbreviation Formulamentioning

confidence: 99%

“…Various optimization methods such as Particle Swarm Optimization (PSO) and Genetic algorithm (GA) [2], [3] can be employed to select a set of features with significant importance. Also, a plethora number of classifiers have been reported by researchers including decision trees [4], random forest [5], K nearest neighbor (KNN) [6], [7], naïve Bayes classifier [8], multilayer perceptron (MLP) [9], gradient boosting (GB) [10], support vector machine (SVM) [11], [12] and extreme value machine (EVM) [13].…”

Section: Introductionmentioning

confidence: 99%

Real-Time EMG Signal Classification via Recurrent Neural Networks

Azhiri¹,

Esmaeili²,

Nourani³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Real-time classification of Electromyography signals is the most challenging part of controlling a prosthetic hand. Achieving a high classification accuracy of EMG signals in a short delay time is still challenging. Recurrent neural networks (RNNs) are artificial neural network architectures that are appropriate for sequential data such as EMG. In this paper, after extracting features from a hybrid time-frequency domain (discrete Wavelet transform), we utilize a set of recurrent neural network-based architectures to increase the classification accuracy and reduce the prediction delay time. The performances of these architectures are compared and in general outperform other state-of-the-art methods by achieving 96% classification accuracy in 600 msec.

show abstract

EMG-Based Feature Extraction and Classification for Prosthetic Hand Control

Cited by 4 publications

References 22 publications

A crack compensation strategy for highly stretchable conductors based on liquid metal inclusions

A crack compensation strategy for highly stretchable conductors based on liquid metal inclusions

Machine Learning Techniques for Distinguishing Hand Gestures from Forearm Muscle Activity

Real-Time EMG Signal Classification via Recurrent Neural Networks

Contact Info

Product

Resources

About