Student's t Test

A students t-Test is a statistic for measuring the significance of a difference of means between two distributions[7]. In conventional fMRI, the intensity of each voxel in an image is recorded temporally, a plot of which is referred to as its time course. During the experiment there are control (cntl) and stimulation (stim) periods. It is the set of intensity values during these periods that make up the distributions for the t-Test. The t-Test parameter popup window is shown in Figure 20.

  
Figure 20: Student's t Test parameter popup window.

1. The first step in calculating a t-Test map is to populate the Time Course Periods list (20-6). This is accomplished by entering a Period Type (20-7), Start time (20-8), End time (20-9), and then pressing Add (20-10). The Change (20-22) and Delete (20-23) buttons can be used for editing the currently selected period from the list (20-6).
2. Then create a t-Test by pressing the Add button (20-2), which will make an entry in the T Tests list (20-1). Next an association must be made between particular periods to the currently selected t-Test. This is accomplished by selecting a period from list (20-6) and pressing the Add button (20-5). Or simply press the All button (20-24) to associate all the periods. The Delete button (20-4) can be used to remove an association. This procedure can be repeated if multiple t-tests are desired.
3. A graph of the time course periods can be displayed by pressing the Plot button (20-20). As shown in Figure 21, the lines at 1.0 show stimulation periods and the lines at -1.0 show control periods. Those periods which are associated to the currently selected t-Test have a second line at 1.5 and -1.5 for stimulation and control, respectively.
4. Next, one of two options can be specified to speed up the analysis by limiting which pixels are evaluated. If the Mask checkbox (20-11) is selected, the BckGnd Thres (20-12) is ignored and only those pixel locations with non-zero values in the currently loaded mask are evaluated. When the Mask option is not selected, then those pixels, whose cntl period mean value is below the BckGnd Thres value, are not evaluated.
5. Then specify the t-Test Confidence threshold (20-13). Only those pixels whose t-test result is above this threshold will pass and have non-zero map values. A value of zero is used in the map at pixel locations that are not evaluated or whose t-Test confidence value is less than the threshold. For multiple t-Tests, a pixels time course will be considered to have failed if it does not pass all of the specified t-Tests.
6. There are two Map Type (20-14) options which determine the non-zero values in the map. The Confidence option simply uses the t-Test confidence values. The Percent Change option calculates the ratio of the average intensity change between the control and stimulation periods as given in Equation 4.
    
   
7. The PZMV (20-15) option stands for ``Period Zero Mean Variance.'' and is useful when there exists a baseline drift in the time course and at least one distribution consists of more than one period. With the PZMV option, the calculation of the distribution variance is based on data where the period mean is subtracted from each point within its respective period.
8. Prior to calculating the t-Test map, the parameters must be saved to a file. Pressing the Save button (20-19) will popup the file browser (See Section 5). After the parameters are saved, the name of the file will appear in the lower left hand corner (20-18). When any parameter is changed the file name disappears since it is no longer consistent with the contents of the file. However, while a file name is displayed, the AutoSave option (20-16) can be selected. The AutoSave option will automatically save the parameters to the current file. The Load button (20-17) can be used to retrieve previously saved parameters.
9. Having all the parameters and options set, pressing the Do Map Calc button (20-21) invokes the calculation of the functional map. See Section 9.1 for the details on how to display the map.

  
Figure 21: Plot of t-Test periods.