Abstract
Involuntary contractions are common in muscles paralyzed by spinal cord injury (SCI). These contractions may impact joint movements and upset daily tasks. Existing rule-based algorithms for counting the number of muscle contractions from electromyographic (EMG) signals work on one channel at a time. However, to understand activation of muscles during involuntary contractions, it is important to develop algorithms that can process signals from multiple muscles simultaneously. To characterize these contractions, EMG signals recorded from paralyzed muscles were analyzed. First, existing singlechannel signal processing techniques were applied to each EMG recording. Then an Eigenvalue decomposition technique on a block of signals from all muscles was used to extract features of the signal-block. An extended version of the KL-distance measure (a method to compute the distance between two probability mass distributions) was used to find the distance between two adjacent sets of Eigenvalues. The regions with significant distances were marked as potential areas for identification of muscle contractions. We further developed algorithms to identify co-Activation of muscles and the muscle that was activated first, since this muscle may be targeted in interventions that aim to dampen these contractions. These algorithms were tested on five hours of EMG data (2:00 am to 7:00 am) recorded from five paralyzed (no voluntary control) leg muscles of a person with SCI. Most of the myoclonus spasms that involved contractions containing clonic-like EMG were identified accurately. Although the soleus muscle was often co-Active, on average 74.31% of the time, it initiated the contraction in only 60.91% of the cases. In contrast, the tibialis anterior muscle was co-Activated 37.92% of the time, on average, but was the first muscle to respond 85.48% of the time. While the proposed approach has shown great potential for concurrent analysis of multi-muscle EMG recordings, we want to continue testing on larger data sets and for other spasm types.
| Original language | English (US) |
|---|---|
| Title of host publication | 2017 IEEE Signal Processing in Medicine and Biology Symposium, SPMB 2017 - Proceedings |
| Publisher | Institute of Electrical and Electronics Engineers Inc. |
| Pages | 1-6 |
| Number of pages | 6 |
| Volume | 2018-January |
| ISBN (Electronic) | 9781538648735 |
| DOIs | |
| State | Published - Jan 12 2018 |
| Event | 2017 IEEE Signal Processing in Medicine and Biology Symposium, SPMB 2017 - Philadelphia, United States Duration: Dec 2 2017 → … |
Other
| Other | 2017 IEEE Signal Processing in Medicine and Biology Symposium, SPMB 2017 |
|---|---|
| Country | United States |
| City | Philadelphia |
| Period | 12/2/17 → … |
Fingerprint
ASJC Scopus subject areas
- Health Informatics
- Clinical Neurology
- Signal Processing
- Cardiology and Cardiovascular Medicine
Cite this
Method for concurrent processing of EMG signals from multiple muscles for identification of spasms. / Sikder, F.; Sarkar, Dilip; Schwartz, Odelia; Thomas, Christine K.
2017 IEEE Signal Processing in Medicine and Biology Symposium, SPMB 2017 - Proceedings. Vol. 2018-January Institute of Electrical and Electronics Engineers Inc., 2018. p. 1-6.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Method for concurrent processing of EMG signals from multiple muscles for identification of spasms
AU - Sikder, F.
AU - Sarkar, Dilip
AU - Schwartz, Odelia
AU - Thomas, Christine K
PY - 2018/1/12
Y1 - 2018/1/12
N2 - Involuntary contractions are common in muscles paralyzed by spinal cord injury (SCI). These contractions may impact joint movements and upset daily tasks. Existing rule-based algorithms for counting the number of muscle contractions from electromyographic (EMG) signals work on one channel at a time. However, to understand activation of muscles during involuntary contractions, it is important to develop algorithms that can process signals from multiple muscles simultaneously. To characterize these contractions, EMG signals recorded from paralyzed muscles were analyzed. First, existing singlechannel signal processing techniques were applied to each EMG recording. Then an Eigenvalue decomposition technique on a block of signals from all muscles was used to extract features of the signal-block. An extended version of the KL-distance measure (a method to compute the distance between two probability mass distributions) was used to find the distance between two adjacent sets of Eigenvalues. The regions with significant distances were marked as potential areas for identification of muscle contractions. We further developed algorithms to identify co-Activation of muscles and the muscle that was activated first, since this muscle may be targeted in interventions that aim to dampen these contractions. These algorithms were tested on five hours of EMG data (2:00 am to 7:00 am) recorded from five paralyzed (no voluntary control) leg muscles of a person with SCI. Most of the myoclonus spasms that involved contractions containing clonic-like EMG were identified accurately. Although the soleus muscle was often co-Active, on average 74.31% of the time, it initiated the contraction in only 60.91% of the cases. In contrast, the tibialis anterior muscle was co-Activated 37.92% of the time, on average, but was the first muscle to respond 85.48% of the time. While the proposed approach has shown great potential for concurrent analysis of multi-muscle EMG recordings, we want to continue testing on larger data sets and for other spasm types.
AB - Involuntary contractions are common in muscles paralyzed by spinal cord injury (SCI). These contractions may impact joint movements and upset daily tasks. Existing rule-based algorithms for counting the number of muscle contractions from electromyographic (EMG) signals work on one channel at a time. However, to understand activation of muscles during involuntary contractions, it is important to develop algorithms that can process signals from multiple muscles simultaneously. To characterize these contractions, EMG signals recorded from paralyzed muscles were analyzed. First, existing singlechannel signal processing techniques were applied to each EMG recording. Then an Eigenvalue decomposition technique on a block of signals from all muscles was used to extract features of the signal-block. An extended version of the KL-distance measure (a method to compute the distance between two probability mass distributions) was used to find the distance between two adjacent sets of Eigenvalues. The regions with significant distances were marked as potential areas for identification of muscle contractions. We further developed algorithms to identify co-Activation of muscles and the muscle that was activated first, since this muscle may be targeted in interventions that aim to dampen these contractions. These algorithms were tested on five hours of EMG data (2:00 am to 7:00 am) recorded from five paralyzed (no voluntary control) leg muscles of a person with SCI. Most of the myoclonus spasms that involved contractions containing clonic-like EMG were identified accurately. Although the soleus muscle was often co-Active, on average 74.31% of the time, it initiated the contraction in only 60.91% of the cases. In contrast, the tibialis anterior muscle was co-Activated 37.92% of the time, on average, but was the first muscle to respond 85.48% of the time. While the proposed approach has shown great potential for concurrent analysis of multi-muscle EMG recordings, we want to continue testing on larger data sets and for other spasm types.
UR - http://www.scopus.com/inward/record.url?scp=85050662356&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85050662356&partnerID=8YFLogxK
U2 - 10.1109/SPMB.2017.8257023
DO - 10.1109/SPMB.2017.8257023
M3 - Conference contribution
VL - 2018-January
SP - 1
EP - 6
BT - 2017 IEEE Signal Processing in Medicine and Biology Symposium, SPMB 2017 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
ER -