26
Positive z values will be assigned to warm (red | orange | yellow) colors and the negative z values cold colors (green | blue | indigo). The third parameter (above) identifies the range of slice coordinate (in mm, in Talairach space) to display. It can be given by a 1-D vector (such as [2 3.5 5 …]), or in Matlab [start:gap:end] format. Example: Entering [–26:2:20] with axial orientation, tells FMRLAB to display slices from z-axis position –26 mm to +20 mm with a 2-mm gap. A typical 2-D slice-overlay display is shown below.
27
3.9 Produce a 3-D head-model rendered display
3-D rendering is probably the most popular and well-accepted format for displaying fMRI results. FMRLAB uses the 3-D template provided with SPM99 to overlay the region of activity (ROA) map onto the SPM’99 3-D template brain. The ROA maps must first be spatially normalized to standard Talairach coordinates using the spatial normalization function (see Section 3.6 above).
When the normalized ROA maps have been created, select Visualize > 3-D Display to start the 3-D rendering. First, the Pick a File Window will pop up (below, left), allowing you to specify the ROA map to display (normally a file named, again, something like “nroa_005.img”). When ready, click the Open button to close the file selection window and bring up the 3-D Rendering window (right below). The top parameter input is the lower-bound z-value threshold, which is used to ignore insignificant voxels in the ROA map. The second entry specifies the translucency of the color display. Translucency allows the viewer to “see through” the brain to activations within the outer brain surface. Typical values for translucency are 0.25, 0.5, 0.75, 1 or NaN. The lower the value, the more opaque the brain template. To display without translucency, use [NaN] (Matlab for “not a number”). The third option allows the user to specify which 3-D brain template to use. Possible values are 1 (SPM96 template), 2 (subject-average template) or 3 (single-subject template).
After these inputs are complete, press OK to begin the 3-D rendering, which will produce a figure like that below.
28
For other functions in the FMRLAB toolbox, see the FMRLAB manual. We hope that you will enjoy exploring the complexity of BOLD data sets using FMRLAB, and that in so doing, you may make exciting discoveries about what hemodynamics may tell us about how human brain dynamics support experience and behavior.
Jeng-Ren Duann
Scott Makeig
La Jolla 9/2002
29
Appendix – Function List of FMRLAB
A. 1 Main Files: fmrlab.m fmrlab.mat license.txt boldimage.m clear_fmri_global.m clear_workspace.m dilation.m erosion.m execute_ica.m export_result.m
main function of FMRLAB toolbox
MAT file to keep necessary parameters for FMRLAB toolbox GNU license
image the intertrial dynamics of BOLD signal
clear FMRI data structure from the working environment
clean up the workspace by closing all the opened windows by FMRLAB perform dilation on input image (used in extract_brain_ui()) perform erosion on input image (used in extract_brain_ui()) execute ICA with GUI for users to specify parameters
export region of activity (ROA) maps to ANALYZE format for further visualization
extract_brain_by_edit.m set threshold value for removing off-brain voxels by key in value in edit box extract_brain_ui.m GUI for user to remove the off-brain voxels extract_brain_ui.mat fmri_bpfilter.m get_status.m ica_linux jr_color.m jr_normalization.m jr_render.m load_dataset.m make_blobs.m map_on_fmri.m map_on_struc.m modify_param.m modify_struc_info.m progressbar.m pvafmap_ui.m
MAT file to keep the necessary fields for extract_brain_ui() perform ideal high/band/low-pass filter on fMRI time courses get current status of FMRI data structure main program of binary ICA
specify the colormap used to display the functional ROAs 3D normalize ROA map to standard brain template
3D rendering of ROA map on 3D standard brain template provided by SPM99
load FMRI data structure up to the working space
read spatially normalized ICA ROA map and convert it to the data structure used to in 3D rendering processes
component browser by overlaying ROA onto 2D slices of functional images with interactive graphic user interface
component browser by overlaying ROA onto 2D slices of structural images with interactive graphic user interface
modify necessary parameters for data analysis and visualization modify parameters of structural images
progress bar showing the progress of the running program
display percentage variance accounted for (pvaf) map with graphic user 30
interface
read_analyze_hdr.m read_structure.m remove_dummy.m reselect_fmri.m rm_slice.m roamap_ui.m roaproj_ui.m
read header file of images saved in ANALYZE format read structural images according to the specified parameter remove dummy scans from the fMRI time series data select new fMRI data set with the same parameters remove noisy slices from fMRI data display ROA maps with graphic user interface
ROA back-projection to fine the back_projected ICA time courses and mean time course of the ROA voxels and calculate the PVAF for a specified component
find the mean time course of the ROA voxels save FMRI data structure as .fmr file in disk
construct FMRI data structure as global variable in current workspace for further analysis
set_fmri_global() with interactive graphic user interface
select structural images into FMRI data structure and set the necessary parameters
call show_actslice() and display normalized ICA ROA maps onto normalized 2D structural image of individual subjects or 2D brain template in a slice-by-slice manner
display normalized ICA ROA maps onto the rendered 3D brain templates provided by SPM99
show_actslice.m
overlay the activation map onto the structural image. Both structural images and activation map should be normalized to the standard brain space (Talairach space) with SPM
display normalized ICA ROA maps on mip template provided by SPM99 show parameters of image acquisition and analysis in main window adjust image inhomogeneity due to different acquisition timing for each slice graphic user interface of slice_timing()
necessary information needed for slice_timing_ui()
spatially smooth image slices to remove the spiky noise due to slignal lose in image acquisition
temporally smooth fMRI time courses with 3 time-point averaging compact version of subplot()
roatc_ui.m save_dataset.m set_fmri_global.m set_fmri_global_ui.m set_struc_info.m show_2d.m
show_3d.m
show_mip.m
show_parameters.m slice_timing.m slice_timing_ui.m slice_timing_ui.mat spatial_smooth.m temporal_smooth.m tightsubplot.m