The purpose of this tutorial is to get you acquainted with the concepts need to perform multi-modal integration in FreeSurfer using fMRI and DTI analysis. You will not learn how to perform fMRI or DTI analysis here; that knowledge is already assumed. The fMRI makes use of data from the Functional Biomedical Informatics Research Network (fBIRN, www.nbirn.net).
In fMRI, stimuli are presented to a subject, which creates a BOLD hemodynamic response function (HRF) in certain areas of the brain. The analysis is performed by first performing motion correction, then correlating each voxel's time course with the stimulus schedule convolved with an assumed HRF shape. The result is an estimate of the HRF amplitude for each condition at each voxel, contrasts of the HRF amplitudes of various conditions, the variance of this contrast, and some measure of the signficance (eg, p, t, F, or z) map. All these maps are aligned with the motion correction template, which should be used as the registration tempate.
This Data Set
All the commands in this section should be run from this directory
These are 5 subjects from the fBIRN Phase I acquisition. They are fbirn-10?, where "?" is 1, 3, 4, 5, 6 (note that #2 is missing). Each has a FreeSurfer reconstruction by the name fbirn-anat-10?.v4.
The data are the results from a sesorimotor paradigm (flashing checkerboard, audible tone, and finger tapping). The raw fMRI data were motion corrected but not smoothed. Each subject has four volumes:
template.nii - motion correction template ces.nii - contrast effect size cesvar.nii - variance of contrast effect size sig.nii - signed signifiance of contrast (-log10(p))
The contrast is the contrast between the ON and the OFF (ie, a comparison against baseline). The sig.nii volume has signed -log10(p) values. So, if the p-value = .01, -log10(p) = 2. If the contrast was positive, then the value would be +2, if negative (ie, ON<OFF), then the value would be -2.
Here we are going to look at the results on a single subject.
First (and always), check the registration (see the Registration Tutorial for more information).
tkregister2 --mov template.nii \ --reg bb.register.dat --surf
This registration should already be good, so there is no need to make any modifications. In a real analysis, you should check the registrations for all subjects, but that is not necessary here.
View sig map on anatomical volume
tkmedit fbirn-anat-101.v4 orig.mgz -aux brain -seg aparc+aseg.mgz \ -overlay sig.nii -reg bb.register.dat
ON>OFF is red/yellow
ON<OFF is glue/cyan
- Notice activation in motor, auditory, and visual regions.
View sig map on left hemisphere
Before you can view the fMRI data on the surface, you must resample the data onto the surface:
mri_vol2surf --mov sig.nii \ --reg bb.register.dat \ --projfrac 0.5 --interp nearest \ --hemi lh --o lh.sig.mgh
- The "moveable" is the signficance map (which is in line with the template.nii used for registration).
- "--projfrac 0.5" indicates that the significance should be sampled half way between the white and pial surfaces.
- "--interp nearest" means use nearest neighbor interpolation (good for sig).
The output is lh.sig.mgh the significance sampled onto the left hemisphere. It has the same size as any other surface overlay for this subject, eg, lh.thickness. To see it's dimensions, run:
You will see "dimensions: 164121 x 1 x 1", indicating that there are 164121 "columns" (ie, vertices), 1 "row", and 1 "slice".
tksurfer fbirn-anat-101.v4 lh inflated -annot aparc \ -overlay lh.sig.nii
ROI Analysis without a Functional Constraint
In this section, we will compute the average HRF contrast (ces) in some ROIs. First, we need to resample the ces.nii volume into the individual's anatomical space:
mri_vol2vol --mov ces.nii \ --reg bb.register.dat \ --fstarg --interp nearest \ --o ces.anat.bb.mgh
Notes: 1. "--fstarg" means to sample the data into the anatomical space 1. Nearest-neighbor (--interp nearest) because we will average within each ROI so we do not want averaging from interpolation.
The output is ces.anat.bb.mgh. Look at it's dimensions:
Note that it is 256x256x256 with each voxel being 1mm isotropic. This is the same size as the FreeSurfer anatomical.
Now run the segmentation statistics:
mri_segstats \ --seg $SUBJECTS_DIR/fbirn-anat-101.v4/mri/aparc+aseg.mgz \ --ctab $FREESURFER_HOME/FreeSurferColorLUT.txt \ --id 1021 --id 1022 --id 1030 --id 17 \ --i ces.anat.bb.mgh --sum ces.bb.stats
Notes: 1. The subject's own aparc+aseg.mgz is used 1. Segmentation names come from the LUT 1. Report on only 4 segmentations are requested (1021=ctx-lh-pericalcarine, 1022=ctx-lh-postcentral, 1030=ctx-lh-superiortemporal, and 17=Left-Hippocampus). The first three are related to the paradigm. 1. The input is ces.anat.bb.mgh
Look at the output ces.bb.stats in a text viewer (or click here). It has a format very similar to the aseg.stats file in each subject's stats directory. The "Mean" here is the average HRF amplitude in the given ROI in raw MR units. Eg, the average HRF amplitude in ctx-lh-pericalcarine is 112.9913. Note that the volume is 2586.0 mm3; this is the volume of the ctx-lh-pericalcarine segmentation.
ROI Analysis with an Unsigned Functional Constraint
In this section, we will compute the average HRF contrast (ces) in anatomical ROIs constrained by functional activation. First, we need to resample the sig.nii volume into the individual's anatomical space:
mri_vol2vol --mov sig.nii \ --reg bb.register.dat \ --fstarg --interp nearest \ --o sig.anat.bb.mgh
Notes: 1. "--fstarg" means to sample the data into the anatomical space 1. Nearest-neighbor (--interp nearest) because this is a significance volume.
Now run the segmentation statistics with the functional constraint:
mri_segstats \ --seg $SUBJECTS_DIR/fbirn-anat-101.v4/mri/aparc+aseg.mgz \ --ctab $FREESURFER_HOME/FreeSurferColorLUT.txt \ --id 1021 --id 1022 --id 1030 --id 17 \ --i ces.anat.bb.mgh --sum ces.abs-masked.bb.stats \ --mask sig.anat.bb.mgh --maskthresh 2 --masksign abs
Notes: 1. The mask is the signficance map 1. The threshold is 2 (p<.01) 1. "--masksign abs" means to use any voxel that exceeds threshold regardless of its sign.
Look at the output ces.abs-masked.bb.stats in a text viewer (or click here). The "Mean" for ctx-lh-pericalcarine is now 181.3009, an increase of 60%. The increase is because all of those voxels that did not have any signal were excluded from the ROI average. The volume is now 1512.0 mm3, meaning that 58% (1512.0/2586.0) of the ROI is above thershold.
ROI Analysis with a Positive Functional Constraint
In this section, we will compute the average HRF contrast (ces) in anatomical ROIs constrained by positive functional activation only. Run the segmentation statistics with the functional constraint:
mri_segstats \ --seg $SUBJECTS_DIR/fbirn-anat-101.v4/mri/aparc+aseg.mgz \ --ctab $FREESURFER_HOME/FreeSurferColorLUT.txt \ --id 1021 --id 1022 --id 1030 --id 17 \ --i ces.anat.bb.mgh --sum ces.pos-masked.bb.stats \ --mask sig.anat.bb.mgh --maskthresh 2 --masksign pos
Notes: 1. "--masksign pos" means to use any voxel that exceeds threshold and has a positive sign.
Look at the output ces.pos-masked.bb.stats in a text viewer (or click here). The "Mean" for ctx-lh-pericalcarine is now 198.2415, an increase of 10% over the unsigned case. The increase is because some of the voxels in the unsigned case are negative and so reduce the ROI average. Note that the volume has dropped to 1408.0 mm3 because the negative voxels have been removed. Note: in this case the Mean will always be positive because we have constrained it that way!