BrainVoyager version: 22.4.4
Dataset used: Getting Started Guide data
In BrainVoyager’s “Event-Related Averaging” dialog, there are three options available to compute the baseline for the calculation of the percent signal change:
- Epoch based
- File based
- Condition-per-file based
Each of these three options uses a different method to calculate the baseline for the percent signal change of activation within the time window defined by the user. In the following, BrainVoyager's Gettting Started Guide dataset, available via https://www.brainvoyager.com/bv/sampledata/index.html, will be used to show the differences between these options. The example will focus on three of the six protocol conditions, i.e. Houses in LVF, Houses in RVF and Houses in CVF.
First, a “Single Study GLM” is calculated and a “region of interest” is selected.
In the next step the Event-related Averaging calculation will be specified, with the three different baseline computation options. The resulting AVG filename will represent the options chosen by the user:
NFs-1_NCs-3_Pre-2-Post-16_PSC-1-BL-From--2-To-0_Bars-SEM_Data-VTC.avg
NFs-: number of functional runs
NCs-: number of protocol conditions selected for event-related averaging
Pre-: number of included datapoints before stimulus onset
Post-: number of included datapoints after stimulus onset
PSC-: percent-signal-change calculation
-BL: baseline computation option (1: file based, 2: condition-per-file based, 3: file based)
-From-: first datapoint included for baseline computation relative to stimulus onset 0
-To-: last datapoint included for baseline computation relative to stimulus onset 0
Bars: variability bars represented by standard errors of the mean (SEM) or by standard deviation (SDEV)
Data-: kind of functional datasets used for event-related averageing, FMR or VTC
1. Epoch based baseline computation
This option is comparable to the averaging performed in an Event Related Potential (ERP) analysis. The specific time point or time points (depending on the number of datapoints selected for the “Pre-period") before each different epoch (event in the timecourse) is used for the
calculation of the percent signal change. In the picture below, only one epoch (second block of the “yellow” condition) is taken into account.
The percent signal change is calculated as:
(value – specific_baseline_for_epoch ) / specific_baseline_for_epoch
First, this formula is used for all specific epochs (here three blocks per condition) within a run. After this, the percent signal change values for the same condition are averaged over the time course. This is the resulting event related averaging plot for six PRT conditions (Houses in LVF, Houses, in RVF, Houses in CVF):
2. File based baseline computation
This method is the one most comparable to the calculation performed in the general linear model. If the option “File based” is chosen, first every value of each pre-period over the whole time course is averaged. After doing so, the calculation of the percent signal change is performed with this average value.
The percent signal change is calculated as:
(value – average_baseline_for_run) / average_baseline_for_run
This is the resulting event related averaging plot:
3. Condition-per-file based baseline computation
Using this option, the pre-period averages are computed specifically for each protocol condition within the run. First, the average baseline of each condition is calculated (here, only the “yellow” condition is shown).
The percent signal change is calculated as:
(value – average_baseline_for_condition) / average_baseline_for_condition
This is the resulting event related averaging plot: