User's Guide‎ > ‎Analysis‎ > ‎

8 - Fitting / back-projection to single subjects

Use the “landscape” obtained by the segmentation to determine periods of interest. These periods of interest can either be periods where the conditions are not explained by the identically numbered map or by onset, offset and/or duration shifts between similarly numbered maps. You may now use these time periods to check if the modulations in response strengths AND latencies you observed at group level are statistically significant at single subject level.

TIPS: 

-The time periods where topographic modulations occur also should appear significant in the TANOVA where you calculated the between-condition dissimilarity.

- The "constrained fitting" procedure allows for a full data-driven analysis as it does not require the experimenter to select the period of interest. The later will be determined automatically as the maps cartool will fit and the period at which they will be be fitted will depend on when they appear in the segmentation. Just drag and drop your .seg file in the fitting window and choose the dependant variable you want and that's it!


Step 1: Calculation of timing and appearance differences between the maps for each condition 

-          go to “Tools” >>> “Fitting of EEG Files”

-          choose as Template File the *.ep (equivalent to the *.seg) you defined as comprising appropriate no. of maps in the segmentation process

-          here: 2 groups of files are compared

-          “Add new group” and choose all the singl_avr.eph from the folder with your grand_avr from one condition (since they are baseline corrected and in the appropriate order…) >>> C1S1, C1S2, C1S3….

-          “Add new group” and choose all the singl_avr.eph from another condition >>> C2S1, C2S2, C2S3….

-          Add epoch (based on the GFP differences you observed during the preceding segmentation on grand_avr data) and name the maps included in the epoch (e.g. 3 4)

-          Choose labelling parameters (or leave it)

-          Chose variables (e.g. first onset, last offset, no. of TF)

-          “Send results to Clipboard” creates you an Excel file with the timing values to conduct your statistics on and create the graphs


Now, if you’re lucky the analyses of your data so far indicated to you

a)       GFP modulations (first in the t-statistics, then over period determined with the segmentation AND/OR

b)      Topographic modulations (first in the TANOVA, then in the segmentation and confirmed by the fitting procedure) >>> different maps for the conditions with or without differing onsets and/or durations).

 

Regardless of whether you have only one or both modulations, you should submit the time periods corresponding to the modulations the inverse solution algorithm. By this, the potential intracranial sources of your extracranially recorded brain responses can be back-tracked.

 

Step 2: Calculation of response strength differences between conditions

Periods of stable microstates (as determined by the segmentation) over which the statistics revealed significant GFP differences HAVE to be used (like in the 1. step).

For statistically testing GFP differences

a)       Export all the GFP`s  of each single subject average per  condition and paste it in an excel file (remember: click on “Tools” >>> “Export Tracks” after you loaded the singl_avr`s into CarTool

b)      calculate GFP area (i.e. averaging of instantaneous gfp measured at all TF over the period of interest) over the period of stable microstate for each subject and condition

c)     with the result of b), calculate with t-tests or ANOVA if GFP modulation also appear significant over the period of stable microstate. This output can also be used for correlation analysis (with behavior -RT, % correct,...) as inter-subkect variability can be extracted).

 

Be careful: GFP of the average is not equal to the average of the individual subject’s GFPs!! Always use the average of each subject and condition for your statistics (in particular for the statistics you’ll do from GFP areas).


Comments