= Plots and adjusted data in SPM99 =
== Getting adjusted data from SPM99 ==
1. Click on Results. Select a contrast that will show you the voxel you are interested in
1. After you have the SPM in the graphics window, select the voxel you are interested in, either by clicking on the voxel, or typing the coordinate in the box in the lower left window
1. Click on Plot, in the lower left window.
1. Select "contrast of parameter estimates" from the menu offered
1. In the contrast manager, select an F contrast that has a row specifying each of the columns that are of interest. For example, let us say that you have done a simple PET experiment with four conditions, and no covariates of interest. In that case, the default F contrast - "effects of interest" uses only the condition columns of the design matrix, and selecting this contrast here will give you data adjusted for the other effects, such as global effects, and subject effects.
1. There is now a variable "y" in the matlab workspace. This contains the data adjusted for the effects of no interest.
1. If you want to save this for further use, you can type
{{{
save y y
}}}
in the matlab window, to save the y variable as a matlab .mat file called y.mat, or
{{{
save y.dat y -ascii
}}}
to save as a text file. The text file can be loaded into other programs for plotting, statistics etc, or back into matlab with
{{{
load y.dat
}}}
(which will recreate the variable "y" in the workspace), or
{{{
y = spm_load('y.dat')
}}}
== Doing nice SPM96-like plots by condition in SPM99 ==
Often, in a PET experiment, we have, say, four conditions. In that case, the plots PET people like, in SPM96, are bar graphs of the signal level for each condition, with the scatter plotted for each condition. These are not so straighforward to get with SPM99. Here's what you do: 1. Click on Results. Select a contrast that will show you the voxel you are interested in
1. After you have the SPM in the graphics window, select the voxel you are interested in, either by clicking on the voxel, or typing the coordinate in the box in the lower left window
1. Click on Plot, in the lower left window.
1. Select "contrast of parameter estimates" from the menu offered
1. In the contrast manager, select an F contrast that has one row for each of the conditions. If you have got no covariates of interest, then the default F contrast - "effects of interest" is exactly what you want. If not, then you will have to make a new F contrast, with one row for each condition. Don't forget, the numbers in the row have to sum to zero to make it a valid F contrast, so, if you had four conditions, a valid F contrast for the four conditions would be:
{{{
3 -1 -1 -1
-1 3 -1 -1
-1 -1 3 -1
-1 -1 -1 3
}}}
If you select - or make - such a contrast, you will get a plot, with one bar for each row in the contrast - in this case one bar for each condition, and a line on top of each bar showing the standard error for each condition.
Matthew Brett 18/11/99