16
color-coded epoch time courses in acquisition order to form a color-coded 2-D BOLD image. Clicking on the Time text selection box and selecting the Component + BOLD item, a text entry window asking for the experiment epoch length and SOA will appear (as shown in the figure above). When these values are specified (followed by pressing OK), the BOLD image of the component time course will be displayed as the right lower panel below.
3.2 Visualize component maps on structural images
FMRLAB provides a function to overlay component ROA maps on the structural images. If you input structural images, you can also overlay the component ROAs on top of them by selecting Visualize > Component ROAs > On Structural Images from the FMRLAB menu. An example is shown in the figure below. The display features are the same as for mapping ROAs on the functional images (3.1 above). Although the user can also use this function to browse the component ROAs, the structural images must be available. Also, the structural maps appear much more slowly than maps on the functional images since the ROAs must be interpolated to fit the image dimensions of the structural scans. To save time, we recommend first mapping component ROAs on functional images, then
17
creating and saving ROA maps on structural images only for selected components of interest.
3.3 Find dominant components by maximum z value
Another FRMLAB component search method selects, for every voxel, the component having the maximum (z-value) weight at that voxel. The assigned component participates most strongly in generating the BOLD signal at the specific voxel. This component may also have the highest correlation coefficient between the back-projected component time course and the whole voxel time course, though this need not always be the case. Color-coding the dominant component for each voxel allows the user to graphically select voxel regions dominated by a single component. To bring up this search image, select Visualize > Dominant Component > by Max Z.
First, a ROAMAP Display window will pop up (as below) allowing you to input the lower-bound z value to use in the display. Entering a higher threshold will make the resulting figure simpler (with less “chickenpox” noise).
18
After the lower-bound threshold is set, a second ROAMAP Display window pops up to ask you if you want to show all components (option 1), or just the components selected as of interest in the component browser (option 2). You can also choose (option 3) to show all components except those on the “reject” list.
After the proper parameters are assigned, FMRLAB will calculate the z value of each voxel weights according to the activation (ui,j = Wi,k * xk,j, where i is the component number, j is the voxel index, and k is the time point of the fMRI time series) of each component, and will assign to every voxel a maximal z value (shown in the left panel of the figure below). To every voxel, the function will also assign a component having the maximum z value (as in the right panel below). We may call this component the “defining component” of the voxel.
Clicking on one of the thumbnail slice images of this map (on right, above) will pop up an enlarged
19
slice image in another window (as below), allowing the user to select a voxel of interest. Normally, we click on a voxel in a color-connected region, for example, as shown by the white arrow below) since most relevant hemodynamic processes for cognitive research may be those that affect the BOLD signals of geometrically connected voxel regions rather than of isolated voxels. After the voxel component of interest is selected, the pop-up window will be closed. On the command line FMRLAB will indicate the number of the component selected.
FMRLAB then computes the Region of Activity (ROA) of the component by searching across all voxels and finding those whose z-normalized component weights are higher than the default threshold. The function then determines the mean whole-BOLD-signal time course for the voxels in the ROA. We call this the ROA raw-mean time course. In the left panel of the figure below, the black trace shows the ROA raw-mean time course, and the red trace the back-projected time course of the defining component. The blue and green traces show the back-projected component time courses of the 2nd and 3rd components accounting most strongly for the ROA raw-mean time course variance.
The right panel shows the user the ROA of the specified defining component and the four components that account maximally for the variance of the ROA raw-mean time course. This panel also gives the pvaf for these four components. For example, below the pvaf of the defining component (IC16) is 66.7%. The second-largest pvaf (IC 47) is 22.0%. the third (IC8), 16.7% and the fourth (IC85), 10%. Note that the ROA raw-mean time course pvaf by the sum of these four independent components is 88.2% (shown under the ROA map). Note: Since spatially independent components need not have orthogonal time courses, the pvaf of the backprojected sum of two or more components is not usually the same as the sum of the individual pvaf’s of the individual component backprojections.
20
3.4 Find dominant components by PVAF
Another way to find dominant components is by ranking components by pvaf by assigning to every pixel the component maximally accounting for the variance of its raw time course. Select Visualize > Dominant Component > By PVAF from the menu. The system will first bring up a window allowing you to select the ICA components you want to consider in constructing the map. There are three options: Entering 1 will use all components in the analysis. If 2, only the browser-selected components will be considered. Entering 3 will cause the function to consider all components except those on the component-browser “Reject” list.
The two figures below show a maximal pvaf map (left) and a maximal z-value map (right). In the z-value map color shows the indices of the most highly weighted components. Here, dark red corresponds to the first (and largest) component, dark blue to the 100th (and smallest) component. These maps show some patches of connected voxels that are dominated by the same ICA component. These are often areas (and components) of functional interest.