---

---

FMRI data analysis displays a visual overlay of the analyzed brain voxels, colour-coded depending on activation level, on top of the volumetric image scans. Fuzzy clustering results in a fuzzy partition and cluster centroids. With respect to fMRI, the centroid is the activation time-course for the cluster, and the fuzzy partition represents the activation levels of the voxels, that is, how similar a voxel's time-course is to the cluster's centroid (time-course). Figure 45 shows the proponent window for the Image Clusters module, from the IBD Analysis / Display package, that shows the activation maps and centroid time-course for 3 of the 5 clusters.

For each cluster, the activation map with membership values (voxel's time-course similarity to the centroid) of 0.5 or higher is displayed in a red scale. The cluster centroid is plotted over the activation map. The user can adjust the number of clusters to display, the current volumetric slice to display (this dataset has only one slice), and the membership value threshold, with valid values ranging from 0.0 to 1.0.

Figure 46 shows the final fMRI data analysis map with the results viewer module, Image Clusters added and properly connected to all the required modules. The results map needs the volumetric image dataset to display and so its input slot image_data is connected to the Mean Cube Image module's output slot mean_cube_image. It also needs to know which voxels, the ROI (x,y,z) coordinates, were used in the analysis. Thus, the image_roi output slot of the module Binary Cube ROI is connected to the input slot image_roi. Next, the fuzzy clustering results have to connected to the Image Clusters module. The Fuzzy C_Means module's output slot partitions is connected to the input slot cluster_mem_maps (these are the activation map values) and the output slot centres is connected to the input slot cluster_centroids (these are the centroid time-course plots). See the User's Guide for further documentation on the input slots for the Image Clusters module.

Before the map is run and the results displayed, the input slots for the Image Clusters results viewer should be made sticky as shown in Figure 47. Once the Input Slot Properties have been set, double click on the Scale Quad module. When it finishes, it will fire the Mean Cube Image module and so on until the Fuzzy C_Means fires and all the required data is available to the Image Clusters module. Middle-clicking the Image Clusters module will display the Properties window, and the fMRI activation maps will be shown, one activation map for each cluster.