This page describes how to perform a Parametric Modulation fMRI analysis in FSFAST for FreeSurfer version 5.0 and higher.

Parametric Modulation refers to the modulation of the amplitude of a hemodynamic response based on some parameter. For example, a task might manipulate the memory load, pain intensity, or risk. One might be interested in measuring the relationship between the parameter and the hemodynamic response. One could imaging plotting the amplitude of the response against the parameter, then fitting a line to this scatter plot. This linear model would have two parameters: an intercept/offset and a slope. Exploring the parametric modulation is the same as asking questions about the slope (is it different than 0? more than 0? less than 0?). The following describes how this is accomplished in FSFAST.

Example: a subject is stimulated 50 times with a painful stimulus. The painfulness of the stimulus can be any of 5 different levels.

The "usual" way to analyze this would be to set up a paradigm file (see FsFastParadigmFile) with 5 event types. For parametric modulation, you would create a different paradigm file. In this file, there will be only two even types but they will represent different aspects of the same event. Each presentation is represented by two entries, both the same except for the condition ID and the weight. For condition ID 1, set the weight to 1. For condition ID 2, set the weight to the shock value.

For example, if the following shock values were presented at the following times each for 1.5 sec:

Time Shock
2.0   4.0
5.0   2.0
9.0   3.0
11.0  1.0
15.0  5.0

The "usual" paradigm file would look like this:

2.0   4  1.5  1.0  Shock4
5.0   2  1.5  1.0  Shock2
9.0   3  1.5  1.0  Shock3
11.0  1  1.5  1.0  Shock1
15.0  5  1.5  1.0  Shock5

Each shock level gets its own condition and the weight (4th column) is always 1.0.

The Parametric Modulation paradigm file would look like this:

2.0   1  1.5  1.0  ShockOffset
2.0   2  1.5  4.0  ShockSlope

5.0   1  1.5  1.0  ShockOffset
5.0   2  1.5  2.0  ShockSlope

9.0   1  1.5  1.0  ShockOffset
9.0   2  1.5  3.0  ShockSlope

11.0  1  1.5  1.0  ShockOffset
11.0  2  1.5  1.0  ShockSlope

15.0  1  1.5  1.0  ShockOffset
15.0  2  1.5  5.0  ShockSlope

Each presentation is represented by two "conditions", one to code the offset/intercept (always a weight of 1.0), the other to code the slope (always a weight equal to the shock intensity).

When creating the analysis with mkanalysis-sess, specify that there are two conditions (-nconditions 2). You can create two contrasts with mkcontrast-sess. One to test the ShockOffset and one to test the ShockSlope

mkcontrast-sess -contrast ShockOffset -a 1 -c 0
mkcontrast-sess -contrast ShockSlope  -a 2 -c 0

In this case, the ShockOffset will be the hemodynamic response when the shock intensity equals 0 and should be similar to a contrast in the "usual" analysis in which all the shock values are compared against baseline.

As with any other contrast, the maps significances of these contrasts can be viewed at an individual level or passed up to the next higher level with isxconcat-sess.

The parametric modulation does not have to be integer values.

FsFastParametricModulation (last edited 2011-03-24 11:54:01 by DougGreve)