Enhanced Dynamic Surface EMG Decomposition Using the Non-Negative Matrix Factorization and Three-Dimensional Motor Unit Localization

Objective: Surface electromyography (sEMG) signal decomposition is of great importance in examining neuromuscular diseases and neuromuscular research, especially dynamic sEMG decomposition is even more technically challenging. Methods: A novel two-step sEMG decomposition approach was developed. The...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE transactions on biomedical engineering Ročník 71; číslo 2; s. 1 - 12
Hlavní autoři: He, Jinbao, Liu, Yang, Li, Sheng, Zhou, Ping, Zhang, Yingchun
Médium: Journal Article
Jazyk:angličtina
Vydáno: United States IEEE 01.02.2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Témata:
3D motor unit localization
Accuracy
Action Potentials
Algorithms
Computer Simulation
Decomposition
dynamic muscle contraction
Eigenvectors
Electrodes
Electromyography
Electromyography - methods
Factorization
Firing
Heuristic algorithms
Localization
Matrix decomposition
Motor units
Muscle, Skeletal
Neuromuscular diseases
non-negative matrix factorization
Reproducibility of Results
Signal-To-Noise Ratio
Surface waves
Three-dimensional displays
ISSN:0018-9294, 1558-2531, 1558-2531
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Objective: Surface electromyography (sEMG) signal decomposition is of great importance in examining neuromuscular diseases and neuromuscular research, especially dynamic sEMG decomposition is even more technically challenging. Methods: A novel two-step sEMG decomposition approach was developed. The linear minimum mean square error estimation was first employed to extract estimated firing trains (EFTs) from the eigenvector matrices constructed using the non-negative matrix factorization (NMF). The firing instants of each EFT were then classified into motor units (MUs) according to their specific three-dimensional (3D) space position. The performance of the proposed approach was evaluated using simulated and experimentally recorded sEMG. Results: The simulation results demonstrated that the proposed approach can reconstruct MUAPTs with true positive rates of 89.12 ± 2.71%, 94.34 ± 1.85% and 95.45 ± 2.11% at signal-to-noise ratios of 10, 20 and 30 dB, respectively. The experimental results also demonstrated a high decomposition accuracy of 90.13 ± 1.31% in the two-source evaluation, and a high accuracy of 91.86 ± 1.14% in decompose-synthesize-decompose- compare evaluation. Conclusions: The adoption of NMF reduces the dimension of random pattern under the restriction of non-negativity, as well as keeps the information unchanged as much as possible. The 3D space information of MUs enhances the classification accuracy by tackling the issue of relative movements between MUs and electrodes during dynamic contractions. The accuracy achieved in this study demonstrates the good performance and reliability of the proposed decomposition algorithm in dynamic surface EMG decomposition. Significance: The spatiotemporal information is applied to the dynamic surface EMG decomposition.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ISSN:0018-9294
1558-2531
1558-2531
DOI:10.1109/TBME.2023.3309969