Introduction
Fmripower was introduced in the 2007 OHBM Poster. New versions will incorporate all of the techniques described in the paper Power Calculations for Group fMRI, which has recently been accepted by NeuroImage.
Fmripower is a matlab based power calculation tool that has been developed for the calculation of power for future group fMRI studies utilizing the analysis results from previous fMRI studies. It was developed by Jeanette Mumford with input from her graduate advisor Tom Nichols and postdoc mentor Russ Poldrack.
Highlights
- Matlab based application (requires SPM5).
- Uses previous anlyses carried out in FSL (any version of FSL).
- Power analysis carried out in an ROI approach. Interpretation of power is for an average voxel of that ROI.
- Reports mean in standard deviation units, which is easy to transfer to a grant application.
- Flexibility to change first level design matrix. Allows user to see how the length of the study paradigm and the number of subjects interact.
- Add ability to enter SPM analyses as well.
- Will include all the details outlined in the OHBM 2006 methods poster and the recently accepted NeuroImage paper, Power Calculations for Group fMRI
Who can use fmripower?
The gui should work on all computer platforms, but it currently has only been tested on various Linux systems. Matlab and SPM5 are required. Windows users will need cygwn, but this is a requirement for FSL, so if you're using FSL you already have it.How can I get fmripower?
You can download the beta version.How can I get help with fmripower?
If you find any bugs, need help, or have suggestions for future versions email help@fmripower.org.How do I use fmripower?
First download and unzip the fmripower_gui.tgz file.- Make sure that the directory for fmripower_gui
files as well as SPM5 are in your matlab path.
- Type fmripower to open the gui.
- The first step is to choose a .gfeat directory and set
the .gfeat settings.
- Choose .gfeat directory by either clicking button and
choosing the directory or typing the directory path into the field.
- Select lower level cope of interest This refers
to the cope.feat directory within the .gfeat
directory containing the analysis that you are interested in.
- Select top level cope of interest This is the
cope that was estimated at the top level (group model) that you are
interesting in calculating power for and
is located in the .gfeat/cope#.feat/stats directory.
- An error message will appear if the .gfeat directory
refers to an analysis with multiple variance groups. The
power calculation method used assumes the same variance
across subjects.
- Choose .gfeat directory by either clicking button and
choosing the directory or typing the directory path into the field.
- Choose group design matrix Load a design matrix from
a file. The design matrix must be saved in ascii format. Soon
this will be extended to include binary .mat files.
- Tip:Use unique and informative design matrix
filenames Power results are saved in a matlab structure
and are distinguishable by the name of the design matrix.
- A window will pop up so you can check that you defined
the columns of your design matrix so they properly coincide
with the contrast.
- Tip:Use unique and informative design matrix
filenames Power results are saved in a matlab structure
and are distinguishable by the name of the design matrix.
- Specify the ROI mask You can use the default
automated anatomical labeling (aal) roi mask, (Tzourio-Mazoyer,
et al., NI, 2002) or your own ROI mask. The regions must be
numbered by positive integers in the mask.
- Specify the type I error rate Typically 0.005 or
0.0001.
- Wait for results After a few moments a new window
will pop up that displays 4 brain images of power, the mean,
standard deviation and mean in standard deviation units. The
mean and standard deviation are probably not in specific units.
The number by each orthogonal view is the value in the voxel
marked by the crosshairs. The region number is located in the
Crosshair Position table at the bottom of the figure.
- Run another analysis Close graphics window, enter a
new design and/or new alpha value and push calculate
- Done? Click exit. This will zip up any nii.gz files
that were unzipped for the calculation.
- What files were created? In the cope.feat directory
is a new directory called Power within this are
directories that share the titles of the ROI masks used in the
power calculations and then a directory for the cope of
interest. Finally there are 4 image files and one matlab
structure stored in this directory
- mn.nii The mean image
- sd.nii The sd image
- mn_sd.nii The mean in sd units
- pow_tmp.nii The power image from the most recent
power calculation
- pow_results.mat Contains all the power calculation information and can be used to construct power curves (power as a function of sample size) in matlab. The roi_num field is a list of the region numbers, des_name is a list of the design names, mn, sd, and mn_sd_units, are the measurements for each ROI (do not change with alpha or design) and power is the power calculation for each roi and each design.
- mn.nii The mean image