CBF and T1 map Processing¶
This section describes the current processing stream for Baseline CBF data.
CBF preprocessing¶
The preprocessing module incorporates the following:
- motion correction (via FSL’s mcflirt)
- outlier rejection
- intensity normalization
Outlier rejection is performed using the series of tag-control images after the motion correction stage. This is criteria to determine outliers is based on the interquartile range of mean-squared-error in intensity differences as described in more detail in the functional pipeline section on outlier_detection.
Intensity normalization is performed to remove the non-uniform coil sensitivity from the BaselineCBF images prior to quantification.
In addition to the above processing, summary images and volumes corresponding to the mean percent signal change and average intensity normalized subtraction volume are produced.
pipeline module: cmind_baselineCBF_preprocess.py
T1 estimation preprocessing¶
This script takes a series of volumes acquired at different inversion times and
prepares them for later fitting to an inversion recovery model. The individual
files are sorted by TI value, coregistered and stored in a 4D file named
T1concat_mcf.nii.gz
. The TI values are stored in a corresponding TIvals.txt
file.
Note
by default an initial estimation of motion is performed using mcflirt. If no motion exceeding half a voxel is present, motion correction is not performed to avoid potential introduction of errors via misregistration of already aligned volumes of differing contrast. When substantial motion is detected the motion corrected timeseries is returned.
pipeline module: cmind_T1map_preprocess.py
T1 map fitting¶
This script takes the 4D NIFTI series and TI values from T1 map preprocessing and performs a fit to the following inversion recovery model on a voxelwise basis:

where ,
and
are the three parameters being
fit. The normalized root-mean-square error (NRMSE) between the data,
,
and the fit
is also evaluated and stored.
Note
To improve the robustness of the fits, an initial fit on a small subset of the voxels is performed. The level of error between the fit and each TI value is then evaluated to look for potential outliers. Up to three outlying TI values may be discarded in this manner and then the full fit is only performed on the surviving TI timepoints.
pipeline module: cmind_T1map_fit.py
Coregistration of T1map and CBF volumes¶
The T1 map is rigidly coregistered to the mean BaselineCBF subtraction volume.
Note
Both acquisitions utilized the same EPI readout parameters so any susceptibility related distortions are expected to be the same.
script: cmind_baselineCBF_coregister.py
Registration to Structural Space¶
The mean BaselineCBF subtraction volume was used for registration to the T1-weighted structural volume. For subjects without fieldmap data, this is an affine boundary-based registration [Greve2009] as implemented in FLIRT [Jenkinson2002]. For those subjects with fieldmap data, epi unwarping is also applied via FUGUE [Jenkinson2001].
pipeline module: cmind_fieldmap_regBBR.py
CBF quantification¶
CBF is quantified using the approach of [Wang2002], Eq. 1. Measured values for inversion efficiency and tissue T1 were used. The labeling duration was 2.0 seconds and the post-labeling delay was 1.4 seconds. The T1 decay rate of arterial blood was assumed to be 1.6 seconds, while the transit times to the arterial and tissue water compartments were assumed to be 1.3 and 1.5 seconds, respectively.
was estimated from the white matter signal intensity in the
BaselineCBF control images after correction for differences in proton density
and T2* decay. The TR of the BaselineCBF scan was 4 seconds, so the white matter
(where T1<1 s) can be assumed fully relaxed. The acquisition used a gradient echo
readout with TE of 11.5 ms. A small correction for the relative
decay of blood vs WM was applied, based on literature values for the relative
decay rates.
The CBF values output from this module are currently not corrected for partial volume effects.
pipeline module: cmind_baselineCBF_quantitative.py
Normalization to Standard Space¶
Quantitative CBF maps were registered to MNI space by concatenating the
low resolution CBF
->``high resolution structural`` transform with the
transform from anatomical to MNI space. The anatomical->MNI transformation
was described in the section on structural processing.
pipeline module: cmind_apply_normalization.py