minScore and number of rows

[beY6ai7u]

Hi, I am going to use hdemg data and will be happy to start contributing here

The script does generate the picture [sample_data/sample1_MUs.jpg] when setting minScore = 0.1.
If I keep minScore = 0.9 (default in the script), I end up with only 12 rows.

What does this mean, do you know which electrodes they refer to?
Many thanks in advance!

[beY6ai7u]

oh and unrelated, but the upload of the file [sample1_ica_results.mat] doesn't seem to have worked (? problem with LFS)

[neuromechanist]

Thanks for raising both issues; regarding the second part, the LFS is for local use, I updated the README to currectly point to the results: de6ed02. I can upload the MAT file to a Google Drive and share that as well.

I'll check the first issue you raised shortly and get back.

[neuromechanist]

This is almost a year-old issue, sorry for the delay.
You are correct, using minScore_toPlot=0.9 with the sample decomposition will result in a 14-source decomposition. The sample data was created using minScore_toPlot=0.7.

Each decomposed motor unit is supposed to create similar spikes (and similar peak values) throughout its activation. The score here is representative of such likelihood but calculating the silhouette score for the spikes, essentially giving them a similarity score. Higher scores indicate that the spike peaks are more similar to each other within a decomposed source. You should try to make a judgment on what score is good for your specific research based on your preliminary results.

I just acquired a personal machine capable of working on this repository and its Python/PyTorch equivalent, so the codebase will be updated more frequently.