I need to segment a sequence of sEMG signals, dividing it into windows which correspond to single movements of the subject. What is the best method to define the threshold value over the rectified and smoothed electromyographic signal? Is there any metric to evaluate this process?
there are a multitude of different methods to determine sEMG onset threshold. Clark (2011, Muscle performance and physical function are associated with voluntary rate of neuromuscular activation in older adults) uses amplitude exceeds 3x resting baseline. I believe Aagaard used a different method. Yang (2017, Accurate EMG onset detection in pathological, weak and noisy myoelectric signals) used a TKE calculation for calculating onset in a disordered/diseased population. But overall, the best method is visual inspection.
There are a variety of methods: with my experience, the best one is mean + 3x SD from the baseline values. Thus I agree with the proposed assessment of Micah.
Another good method (it depends on if you look at the rectified and averaged EMG signal or not) is to use the gradient or slope tangentially. In case the signal is very stable, this can be a good solution as wee.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series