Corticomuscular coherence

Discussions
1

Hello, we started a new project on corticomuscular coherence, using EEG (Biosemi) and EMG (Cometa). However, when we apply coherence analysis, the results appear incorrect (see image).

We have tried different methods, such as calculating coherence using (coherence AxB) between these two files, and also integrating the EMG data directly into the EEG data file before running coherence analysis (coherence 1xN), but results obtained are the same.

Thanks for your help,

CMC
CMC684×727 33.2 KB

2

Hi @dex, the fact that the MSC is 1 for all the frequencies suggest that the signals are the same of have a purely linear relationship, which should not be the case if the signals come form different modalities.

Have you taken a look this tutorial on Corticomuscular coherence between MEG and EMG?
https://neuroimage.usc.edu/brainstorm/Tutorials/CorticomuscularCoherence

1 Like
3

Hello,

Yes, we have looked at the tutorials on corticomuscular coherence. We have tried different methods to obtain corticomuscular coherence. Our methods consisted of importing EEG and EMG files separately. EEG positions worked on EEG files (μV) and EMG (mV) was labeled as EMG sensors. We then measured corticomuscular coherence using the Coherence AxB, where file A was the EMG data and file B was the EEG data. We obtained different results. First, when we used Hilet transform in the Time-Frequency decomposition or Morvelet wavelets it worked (see Image), but when we used Fourier Transform it did not work. We believe that the Fourier transform requires more samples than the 385 available in these files.

Thanks for your help,

image
image1366×732 141 KB

4

This is expected, as those timefrequency representations have different formulations.

  • Did you get any error message, either a pop window or in the Matlab command line?

  • What are the parameters that you are using for those transformations?

You may want to revise your approach, in the shared screenshot, coherence is computed for one segment of 750ms. However, it is necessary to have multiple windows to estimate coherence if FFT (either multiple files or longer files), so the coherence estimate will be computed with a low number of windows, and they will be short, giving you a poor frequency resolution.

5

Hello,

Following your comments, we analyzed and calculated a single file, which turned out to be short. By incorporating additional files, we were able to obtain the FFT.

Thank you very much for your help.