Scout plot in "scout tab" doesn't match with process "extract > scout time series"

Hi,
I am having trouble visualizing the data waveforms for sources. It doesn't seem to match the extracted values. I have extracted the whole time series waveform for certain scouts with extract values function (using the same source file I use to plot data in Brainstorm). I used the scout function: max, compute absolute values and concatenate signals (dimension 1). But the values don't match the visualization with scout timeseries using the visualization tools.

I extracted the values and exported them to excel to view the whole timeseries waveform and compared it to the scout waveform, but they are different. Also the values are in different scale. Furthermore, for my first participant the values and waveform shape matched with the extracted values and the brainstorm illustration but not with other subjects, even though I think I have done identical steps. I extract the values from the same source file which I then use to plot the values for the same scout (I have selected absolute values, for the time series options). Where could the error come from? I have tried to trouble shoot but I don't see why this is correct for only one participants even if I do the same exact steps for the other participants.

I have checked many times that I have selected the same file and scout for plotting. I have also done the extracting multiple times.

Thanks for your help,
-Elisa

Potential sources of difference can be:

  • The use of absolute values do not match between the waveforms from the Scout tab and the waveforms from the Extract > Scout time series process.

  • The order between using Absolute values and Scout function. The Absolute values in the Scout tab means that the Scout is computed with the original time series, and then absolute values are obtained for the scout time series.

  • Is there any flattening applied on the sources?

  • For constrained sources, when the scout function in the Scout tab is computed using Flip the sign of sources with opposite directions. As such this needs to be used as well with the extract process.

Can you provide the following?

  • Are your sources constrained, loose, unconstrained?
  • Screenshot of the waveforms obtained with the Scout tab, as well as a screenshot of the tab
  • Screenshot of the waveforms obtained with the Extract > Scout time series process, and a screenshot of the process GUI

Thanks for the reply,
I have used unsconstrained sources. I can add here a screenshot of the source processing. I compared the source processing for the two data (id1 and id5) and I noticed that they are the same but for some reason for the id1 the selection for noice-covariance was diagonal instead of regular. But I processed the file id 1 again with selection regular and still with this participant the scout tab waveform and the extract values waveform match, but not for other subjects.

I try to include here screencaptures but I am not sure if they are shown, since I can't see them here.

capture for source processing (id 1 and id 5):

Id 1 scout tab and extract value plots:

Id 5 scout tab and extract values plots

How are you obtaining the scout time series?

For the scout tab I just click the waveform icon when the scout has been selected. For the extracted values I export them by right clicking the extracted file (the one that I got after extracting values) and selecting export to file (getting an excel-file). And then plotting the waveform in excel.

I now tried plotting the extracted values in Brainstorm by right clicking the extracted file and selecting time series. Here are on the left the extracted values plot and on the right the same source file plotted with scout tab. It is still different.

I checked that the name of the file is the same for both plots, so it should be from the same values. Something must be then different when plotting with the scout tab. But I don't understand why it works for one participant and these others.

What are the parameters that you are using to extract the Scout time series?

Here is screencapture of the selections for extracting.

Hi @Elma, this configuration of the process Extract values matches the steps in the Scout tab:

  1. Scout function is applied in each of the 3 dimensions, leading to 3 time series per Scout (if unconstrained sources)
  2. The norm is computed for these 3 Scout time-series

I have tried this, and it works as it should.

I noticed that in your first screenshots the Scout tab shows Mean as Scout function, but you mentioned that Max is used, and Max is set in the Extract values GUI.
Could this account for the observed difference?

Hi,
Thanks for the help. I think it might be related to the max-selection. I now checked that if I extract the mean values for the time-series, then it matches the scout tab. From the scout tab, I am not aware how to select max values instead of mean.
However, for that one participant the max and mean values selection produces the same results. That's why the figures matched for that participant even with the max-value selection. It is a bit weird. But I will then use the mean-values if I then can get the figures to match.

I was originally looking the max-values because I wanted to extract the peak max-value from a time window. Now I noticed that the max-function doesn't take the peak maximum value from the selected window, but instead a max value from each timepoint. I guess I can first extract the mean values from a scout, with the extract values function and then use the "find maximum in time" function. I noticed that this function brings out two values (one for the start and one for the end point of the time series). But I think it is still the maximum value within the time window, since the two values are the same.

Here is the selection:

And the result:

Find_maximum_in_time_table

You can read more about how the different scout functions and how to set them in here:
https://neuroimage.usc.edu/brainstorm/Tutorials/Scouts#Scout_function

This is right, it is the same value that is repeated. This is a side effect as the information of the time window where it was found is also two time samples (the beginning and the end of the window).