Contrast Masking and Differential fMRI Brain Responses Relative to a Rest Condition
L. Clark Johnson1, Todd L. Richards2, Andrew Poliakov3, Elizabeth H. Aylward2
1School of Nursing, University of Washington
2Department of Radiology, University of Washington
3Department of Biostructure, University of Washington

Objective: Analysis of an fMRI protocol often involves a comparison of the fMRI brain signal between task A (e.g., zero-back working memory task) and task B (e.g., 2-back working memory). When a third reference condition such as fixation is included then comparisons between the main tasks and this reference are also possible (e.g. 2-back versus fixation). In this case, the fMRI signal from each voxel’s response with regard to fixation can be classified into one of six possible response patterns (1), three for A > B (both A and B > fixation; both A and B < fixation; A > fixation > B) and three for B > A.
The objective of this project was to develop a method to separately map these six different fMRI brain response patterns in addition to the standard task A versus task B comparisons. Understanding the relationship between activation for each task in comparison to fixation is especially important when evaluating response to intervention in fMRI studies.

Methods: Analysis was carried out using FEAT (FMRI Expert Analysis Tool) Version 5.1, part of FSL (FMRIB's Software Library, www.fmrib.ox.ac.uk/fsl) (2) using a standard block design with a 2-back paradigm using pictures of sea creatures (the subject presses a button if the current picture is the same as the picture present two before) and a 0-back task using pictures of whales ( subject presses the button if the picture is a whale). The contrast masking feature of FEAT was used to generate images that mapped voxels with specific response patterns (e.g. six images corresponding to the six response patterns). Clusters with voxels reflecting significant between-task differences were also identified for each of the one-sided comparisons of 2-back versus 0-back (e.g. A > B and B > A). Software was prepared to code each significant cluster voxel with regard to the response pattern from which it was derived. The results of all six comparisons were rendered onto a 3-dimensional structural brain using a cross platform Java/Java3D software toolkit for processing and visualizing brain imaging data (3).
Results & Discussion: Figure 1 shows the fMRI overlay of brain activation which results from the six different contrast masking conditions. Each of the response patterns on the left margin of Figure 1 represent between-task comparisons that will be significant when A > B (e.g., back represents voxels A > B and both A and B > fixation), and response patterns on the right margin are significant when B > A. As can be seen in the Figure, there were contiguous voxels that have different response patterns even within the same cluster.
Conclusions: Contrast masking may be a useful method to characterize the fMRI signal response patterns that can exist in on-off block designs.

Contrast masking may be a useful method to characterize the fMRI signal response patterns that can exist in on-off block designs.
References & Acknowledgements: 1) Gusnard DA and Raichle ME, Nature Reviews, 2,685-694, 2001.
2) Smith SM et al, Neuroimage, 23, S208-S219, 2004.
3) Moore EB, Poliakov AV, Brinkley JF, 2004, 1761, 2004