Investigation of Motor Units Activity: Comparison of Single Channel Surface EMG Deconvolution and Blind Source Separation of Multichannel Data
The firing instants of single motor units (MUs) can be identified by decomposing electromyograms (EMG) detected with intramuscular or grids of surface electrodes. The latter is sometimes preferred due to its larger detection volume and non-invasiveness. When the interest is in firing instants and no...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10477412/ |
| _version_ | 1797238831130345472 |
|---|---|
| author | Luca Mesin Emiliano Robert Gennaro Boccia Taian Vieira |
| author_facet | Luca Mesin Emiliano Robert Gennaro Boccia Taian Vieira |
| author_sort | Luca Mesin |
| collection | DOAJ |
| description | The firing instants of single motor units (MUs) can be identified by decomposing electromyograms (EMG) detected with intramuscular or grids of surface electrodes. The latter is sometimes preferred due to its larger detection volume and non-invasiveness. When the interest is in firing instants and not in investigating the activity of specific MUs, in-silico studies have shown that deconvolution of a single surface EMG is a low cost method providing reliable information. In this study, we explored this possibility by testing the experimental validity of deconvolution by comparison with decomposition of multichannel surface EMG. A single kernel deconvolution method is proposed to estimate the cumulative firings of the MUs from bipolar surface EMGs collected from the biceps brachii of 10 healthy subjects, recorded during contractions of different force levels and an endurance test. Different parameters were tested: force levels, inter-electrode distance, electrode size and location. Validity was assessed by correlating the cumulative firings (after 5–45 Hz band-pass filtering) between the proposed, deconvolution approach and the already validated, EMG decomposition. For all conditions tested, decomposition and deconvolution provided correlation coefficients of about 40%. When considering experimental signals reconstructed with the firings of decomposed MUs, markedly higher correlation values were obtained (median correlations of 90%). High correlation (about 80%) was obtained even when a signal with large interference was built by adding about 90 MU action potential trains, decomposed from different EMGs of our dataset with same contraction levels. Analysis of residual root mean squared error (median across tests of about 40% and 15% for decomposition and deconvolution, respectively) together with the good estimation on reconstructed signals with high interference suggest that deconvolution may identify additional contributions that are not explained by decomposition. This additional information provided by deconvolution may justify in part the discrepancy when comparing the outputs of the two methods applied to the original signals. The cumulative firing instants associated with action potentials can be accurately estimated with the deconvolution of a single bipolar surface EMG. |
| first_indexed | 2024-04-24T17:41:53Z |
| format | Article |
| id | doaj.art-839257f66d1d4bd8be12c31cd28829b8 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-24T17:41:53Z |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-839257f66d1d4bd8be12c31cd28829b82024-03-27T23:00:25ZengIEEEIEEE Access2169-35362024-01-0112431264313810.1109/ACCESS.2024.338000510477412Investigation of Motor Units Activity: Comparison of Single Channel Surface EMG Deconvolution and Blind Source Separation of Multichannel DataLuca Mesin0https://orcid.org/0000-0002-8239-2348Emiliano Robert1Gennaro Boccia2https://orcid.org/0000-0001-8706-4098Taian Vieira3https://orcid.org/0000-0002-6239-7301Department of Electronics and Telecommunications, Mathematical Biology and Physiology Research Group, Politecnico di Torino, Turin, ItalyDepartment of Electronics and Telecommunications, Mathematical Biology and Physiology Research Group, Politecnico di Torino, Turin, ItalyDepartment of Clinical and Biological Sciences, NeuroMuscularFunction Research Group, University of Turin, Turin, ItalyDepartment of Electronics and Telecommunications, Politecnico di Torino, Turin, LISiN, ItalyThe firing instants of single motor units (MUs) can be identified by decomposing electromyograms (EMG) detected with intramuscular or grids of surface electrodes. The latter is sometimes preferred due to its larger detection volume and non-invasiveness. When the interest is in firing instants and not in investigating the activity of specific MUs, in-silico studies have shown that deconvolution of a single surface EMG is a low cost method providing reliable information. In this study, we explored this possibility by testing the experimental validity of deconvolution by comparison with decomposition of multichannel surface EMG. A single kernel deconvolution method is proposed to estimate the cumulative firings of the MUs from bipolar surface EMGs collected from the biceps brachii of 10 healthy subjects, recorded during contractions of different force levels and an endurance test. Different parameters were tested: force levels, inter-electrode distance, electrode size and location. Validity was assessed by correlating the cumulative firings (after 5–45 Hz band-pass filtering) between the proposed, deconvolution approach and the already validated, EMG decomposition. For all conditions tested, decomposition and deconvolution provided correlation coefficients of about 40%. When considering experimental signals reconstructed with the firings of decomposed MUs, markedly higher correlation values were obtained (median correlations of 90%). High correlation (about 80%) was obtained even when a signal with large interference was built by adding about 90 MU action potential trains, decomposed from different EMGs of our dataset with same contraction levels. Analysis of residual root mean squared error (median across tests of about 40% and 15% for decomposition and deconvolution, respectively) together with the good estimation on reconstructed signals with high interference suggest that deconvolution may identify additional contributions that are not explained by decomposition. This additional information provided by deconvolution may justify in part the discrepancy when comparing the outputs of the two methods applied to the original signals. The cumulative firing instants associated with action potentials can be accurately estimated with the deconvolution of a single bipolar surface EMG.https://ieeexplore.ieee.org/document/10477412/Surface EMGdecompositiondeconvolution |
| spellingShingle | Luca Mesin Emiliano Robert Gennaro Boccia Taian Vieira Investigation of Motor Units Activity: Comparison of Single Channel Surface EMG Deconvolution and Blind Source Separation of Multichannel Data IEEE Access Surface EMG decomposition deconvolution |
| title | Investigation of Motor Units Activity: Comparison of Single Channel Surface EMG Deconvolution and Blind Source Separation of Multichannel Data |
| title_full | Investigation of Motor Units Activity: Comparison of Single Channel Surface EMG Deconvolution and Blind Source Separation of Multichannel Data |
| title_fullStr | Investigation of Motor Units Activity: Comparison of Single Channel Surface EMG Deconvolution and Blind Source Separation of Multichannel Data |
| title_full_unstemmed | Investigation of Motor Units Activity: Comparison of Single Channel Surface EMG Deconvolution and Blind Source Separation of Multichannel Data |
| title_short | Investigation of Motor Units Activity: Comparison of Single Channel Surface EMG Deconvolution and Blind Source Separation of Multichannel Data |
| title_sort | investigation of motor units activity comparison of single channel surface emg deconvolution and blind source separation of multichannel data |
| topic | Surface EMG decomposition deconvolution |
| url | https://ieeexplore.ieee.org/document/10477412/ |
| work_keys_str_mv | AT lucamesin investigationofmotorunitsactivitycomparisonofsinglechannelsurfaceemgdeconvolutionandblindsourceseparationofmultichanneldata AT emilianorobert investigationofmotorunitsactivitycomparisonofsinglechannelsurfaceemgdeconvolutionandblindsourceseparationofmultichanneldata AT gennaroboccia investigationofmotorunitsactivitycomparisonofsinglechannelsurfaceemgdeconvolutionandblindsourceseparationofmultichanneldata AT taianvieira investigationofmotorunitsactivitycomparisonofsinglechannelsurfaceemgdeconvolutionandblindsourceseparationofmultichanneldata |