Greetings experts - 

We are attempting to build a mask of clusters of shared activation (derived from a subset of our sample) that can then be used to extract data about changes in activation within those areas from the rest of the sample. After standard preprocessing, individual analysis, and group analysis for the subset, we used MarsBaR to extract cluster coordinates for all clusters of interest. In PickAtlas, we initially attempted to use the .mat files generated by MarsBaR for each cluster to build the mask but received an error message re: not a valid design. We found that we could use the Shape Definition feature to input each set of coordinates, generate a sphere, dilate the sphere, and add it to the working mask. The DILEMMA here is that the MarsBaR cluster data only shows the coordinates of the peak voxel, not voxel size or shape. So how can we create accurate shapes in the PickAtlas working mask?

Thanks in advance for any help you can offer. Our lab is quite new to SPM compared to other neuroimaging analysis tools and are learning as we go.