Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2012 Oct;10(4):351-65.
doi: 10.1007/s12021-012-9152-3.

Development of PowerMap: A Software Package for Statistical Power Calculation in Neuroimaging Studies

Affiliations
Free PMC article

Development of PowerMap: A Software Package for Statistical Power Calculation in Neuroimaging Studies

Karen E Joyce et al. Neuroinformatics. .
Free PMC article

Abstract

Although there are a number of statistical software tools for voxel-based massively univariate analysis of neuroimaging data, such as fMRI (functional MRI), PET (positron emission tomography), and VBM (voxel-based morphometry), very few software tools exist for power and sample size calculation for neuroimaging studies. Unlike typical biomedical studies, outcomes from neuroimaging studies are 3D images of correlated voxels, requiring a correction for massive multiple comparisons. Thus, a specialized power calculation tool is needed for planning neuroimaging studies. To facilitate this process, we developed a software tool specifically designed for neuroimaging data. The software tool, called PowerMap, implements theoretical power calculation algorithms based on non-central random field theory. It can also calculate power for statistical analyses with FDR (false discovery rate) corrections. This GUI (graphical user interface)-based tool enables neuroimaging researchers without advanced knowledge in imaging statistics to calculate power and sample size in the form of 3D images. In this paper, we provide an overview of the statistical framework behind the PowerMap tool. Three worked examples are also provided, a regression analysis, an ANOVA (analysis of variance), and a two-sample T-test, in order to demonstrate the study planning process with PowerMap. We envision that PowerMap will be a great aide for future neuroimaging research.

Figures

Figure 1
Figure 1
An overview of power and sample size calculation by the PowerMap tool.
Figure 2
Figure 2
Extracting power or sample size information from a power surface. A power image may be generated by extracting a curve (pink plane) from the power surface at the desired df. Similarly, a sample size map may be generated by extracting this curve at the desired level of power (green plane).
Figure 3
Figure 3
Using the PowerMap tool to generate a power image from a linear regression analysis. The GUI input screens are shown in order (blue box), along with the resulting power surface and power image (green box).
Figure 4
Figure 4
Power images resulting from a linear regression analysis comparing age and fractional anisotropy with a desired sample size of 30 subjects.
Figure 5
Figure 5
Using PowerMap to generate a sample size image from a one-way analysis of variance (ANOVA). The GUI input screens are shown in order (blue box), along with the resulting power surface and sample size image (green box). Note that the group size input screen is shown for group 3 only. A total of 3 group size input screens are presented to the user, but the screens for groups 1 and 2 are omitted for space.
Figure 6
Figure 6
Sample size images resulting from an analysis of variance comparing voxel-based morphometry across 3 groups: younger, middle, older. The desired power was specified at 80%. Images were thresholded to show sample sizes in the range of 5 to 10 subjects per group.
Figure 7
Figure 7
Using PowerMap to generate a power image from a two-sample T statistic image using FDR correction. The GUI input screens are shown in order (blue box), along with the resulting power surface and power image (green box). Note that the group size input screen is shown for group 2 only. A total of 2 group size input screens are presented to the user, but the screen for group 1 is omitted for space.
Figure 8
Figure 8
Power images resulting from a two-sample T statistic image using FDR correction comparing changes in CBF in an attention training group to a control group. The sample size of the full scale study was specified to be 50 subjects per group.
Figure 9
Figure 9
Power images (a) and sample size images (b) based on two different types of image smoothness estimates. These images are generated on the same data set as Figure 3. Users can input the known image smoothness calculated from an image analysis software tool, or alternatively, PowerMap can employ the improvised method to calculate the estimated smoothness based on the statistics image as described in Appendix A. In each of the panels (a) and (b), the image on the left is based on manually input smoothness and the image on the right is based the smoothness estimated by PowerMap. Although there are subtle differences in power images (a), the sample size images appear very similar (b). For power images the desired sample size was assumed to be 30, and for sample size images the desired power was assumed to be 80%.
Figure 10
Figure 10
The effect of image smoothness that is specified by the user and image smoothness that is estimated by PowerMap on power images (a,c) and sample size images (b,d). Each scatter plot depicts the power (a) or sample size (b) using input smoothness versus estimated smoothness for any given brain matter voxel. The images depict the absolute value of the difference between the power (c) and sample size (d) images shown in Figure 9.

Similar articles

See all similar articles

Cited by 10 articles

See all "Cited by" articles

Publication types

LinkOut - more resources

Feedback