Effect of frequency band expansion on muscle annotations

💬 Support & Discussions
1
  • MNE version: e.g. 0.24.1
  • operating system: Windows 11, Python 3.9.7

I am reproducing the tutorial at Annotate muscle artifact on my own data. In order to be more discriminating I have set the threshold at 5. And I wanted to know the effect of taking a wider frequency band on annotation detection, for example 80/140 Hz. I thought (perhaps wrongly) that this would have increased the number of muscle artefacts detected. In fact, the number of muscle artefacts detected was reduced and artefact detected on the default frequency band (110/140 Hz) were no longer detected (see the attached photos). Does anyone have an explanation for this phenomenon please?

Frequency band : 80/140 Hz - Threshold : 5

BP_80_140_Threshold_5
BP_80_140_Threshold_51920×1200 300 KB

Frequency band : 110/140 Hz - Threshold : 5

BP_110_140_Threshold_5
BP_110_140_Threshold_51920×1200 305 KB

I specify that the electrodes visualized are all the electrodes taken into account in the annotation function.

Thank you very much.

2

Hi! When you take a wider frequency range than effectively discriminates muscle artifact trials, you dilute the signal with noise, causing less trials to be rejected. Hope that helps!

3

Hi ! Thanks for you answer !

Yes, I totally agree with you, hence my question: Why with a 5 threshold when I take a wider frequency band (80-140 Hz) I have less detection than with a smaller frequency band (110-140 Hz)?

4

The threshold is effectively lowered because the frequencies in the larger range may not have muscle activity and bring down the average.