[2/3] Fastgraph: Remove SPES artifacts using EMD

[chinmaychinara91]

This PR is the 2nd part required for Fastgraph. It involves cleaning the data for any artifacts.

Prerequisite: Follow steps in PR #911

Remove SPES artifacts

Linked to process_remove_spes_artifacts and it involves removing SPES artifacts and slow drifts using the Empirical Mode Decomposition (EMD) based filtering.

The GUI

Inputs

• Stimulation event: The stimulation event around which to remove the artifacts (here, STIM)
• EMD cutoff frequency (in Hz): Self explanatory
• Time to account for artifact (in ms): The duration of the actual stimulation artifact itself, starting at the stimulation event. Used to set the main artifact span inside the interpolation window (here, 5 ms).
• Time on either side of the artifact to spline with (in ms): How much clean signal to take from before and after that artifact window so the spline has anchor points to interpolate across the bad segment (here, 3 ms).

Steps to use

1. Follow steps as in PR [1/3] Fastgraph: Customize SPES (Nihon Kohden) #911 .
2. Drag the loaded stimulation blocks into the Process1 tab. For e.g. SB A13-A14 4 (#1) .
3. Click on Run > Stimulation > Remove SPES artifacts.
4. Fill in the values as seen in the GUI figure above. Click Run.

It creates new stimulation block e.g. SB A13-A14 4 (#1)_emd. The results can be seen as under.

Associated PRs

1. [1/3] Fastgraph: Customize SPES (Nihon Kohden) #911
2. [3/3] Fastgraph: Plot Fastgraphs #916

@jcmosher @Nastaranlotfi @yashvakilna

[rcassani]

This PR presents the process process_remove_spes_artifacts to remove artifacts from stimulation, and to remove drift.

There is reason for artifact-removal operations to be in the same process. It would be more versatile to have them in two different process, so Brainstorm users can use them separately.

[rcassani]

You can use bst_closest

[chinmaychinara91]

Done 0f12ef0

[chinmaychinara91]

There is reason for artifact-removal operations to be in the same process.

The reason we do stimulation artifact removal before drift correction is to avoid the large transient artifacts that may bias the drift estimate. Hence, we decided to club them in one process here.

It would be more versatile to have them in two different processes, so Brainstorm users can use them separately.

I agree that the two operations are conceptually independent, and separating them into two independent processes would definitely make the workflow more flexible for users. Thank you for the recommendation.

@jcmosher, any more inputs on this ?

Should I go ahead and create two separate processes? While documenting in the tutorial, we can suggest the recommended order for the users when both steps are applied.

[rcassani]

Should I go ahead and create two separate processes? While documenting in the tutorial, we can suggest the recommended order for the users when both steps are applied.

Drift removal with EMD can be applied to any other data, that is the reason to have it separated.
Yes, create two processes.

In the tutorial explain what each process do and that will lead the reasoning about the order in which the processes must be used for that tutorial, not a suggestion. As swapping the order is incorrect way to do it for the data of that tutorial.