Neuroimaging

Freesurfer output

Distortion-corrected MPRAGE data is segmented into both Desikan-Killiany (gyral-based) and Destrieux (sulco-gyral-based) cortical and subcortical regions, and hippocampal subfields, using Freesurfer (Fischl et al., Cereb Cortex 2004;14:11-22) (surfer.nmr.mgh.harvard.edu ). All output is visually inspected, corrections are made when necessary and Freesurfer is rerun. Regional volumes, cortical thicknesses and surface areas are calculated and tabulated. Freesurfer output is not generated when the MPRAGE data is of low signal to noise ratio (SNR), low gray-white matter contrast to noise ratio (CNR), or is contaminated by motion artifacts. Partial Freesurfer output is generated when most of the brain allows reliable segmentation while some regions cannot be reliably segmented (e.g. due to low contrast or lesions). (For an example application of our Freesurfer output see Fleischman et al., Brain Struct Funct 2014; 219:2029-2049).

Total volumes

Distortion-corrected MPRAGE data is segmented into gray matter, white matter and cerebrospinal fluid (CSF) using the Computational Anatomy Toolbox (CAT) (www.neuro.uni-jena.de/cat/ ) SPM (Friston et al., Hum Brain Map 1995;3-165-189) (www.fil.ion.ucl.ac.uk/spm ). The total volumes of gray matter, white matter, and CSF are calculated. The intracranial volume is calculated as the sum of the gray matter, white matter and CSF volumes. Total volumes are not calculated when the MPRAGE data is of low signal to noise ratio (SNR), low gray-white matter contrast to noise ratio (CNR), or is contaminated by motion artifacts.

White matter hyperintensities

White matter lesions appearing hyperintense in T2-weighted images are segmented using BIANCA (Griffanti et al., Neuroimage 2016; 141:191-205) (fsl.fmrib.ox.ac.uk/fsl/fslwiki/ ) on FLAIR. BIANCA has been pre-trained on our data. A mask of white matter hyperintensities is generated and the total volume of hyperintensities is calculated. White matter hyperintensities are not segmented when the FLAIR data is of low signal to noise ratio (SNR) or is contaminated by motion artifacts.

T2 maps

Maps of T2 relaxation times are generated by fitting the multi-echo fast spin-echo signals with a mono-exponential decay model. T2 maps are not generated if the mean chi-square throughout the brain exceeds a certain threshold.

Diffusion tensor imaging (DTI)

Distortions in the DW volumes caused by eddy-currents, as well as bulk-motion, are corrected by affine registration to b=0 s/mm2 images using TORTOISE (Pierpaoli et al. ISMRM 2010; p.1597) (science.nichd.nih.gov/confluence/display/nihpd/TORTOISE ). Distortions due to field non-uniformities are corrected by non-linear registration to the MPRAGE data. The B-matrix is appropriately reoriented. Finally, the diffusion tensor is estimated in each voxel using non-linear tensor fitting, and maps of fractional anisotropy (FA), trace, axial and radial diffusivity are generated. DTI output is not generated if raw diffusion-weighted images are contaminated by motion (blind artifact across slices) or have low signal to noise ratio (SNR), and if the chi-square in tensor fitting or the number of voxels with outlier signals exceed certain thresholds. (For an example application of our DTI output see Han et al., Neuroimage 2016;130:223-229).

Functional connectivity MRI (functional connectome)

Gradient-echo echo-planar imaging data is first undistorted using either the field map and FEAT, or two spin-echo echo-planar imaging acquisitions with opposing polarities of the phase-encode blips and TOPUP (fsl.fmrib.ox.ac.uk/fsl/fslwiki/ ). The data is then corrected for slice acquisition timing and motion, co-registered with the MPRAGE data and normalized to MNI space using SPM (Friston et al., Hum Brain Map 1995;3-165-189) (www.fil.ion.ucl.ac.uk/spm/ ). Signal time courses are band-pass filtered (0.01-0.1 Hz). The estimated motion parameters, mean white matter and CSF signals are regressed out. The gray matter in reference space has been parcellated to a number of regions using our data on 689 participants and spatially constrained spectral clustering based on the temporal correlation between voxel time courses (Craddock et al., Hum Brain Map 2012;33:1914-1928) (ccraddock.github.io/cluster_roi/ ). Each one of these regions is used as a seed for functional connectivity and a whole brain functional connectivity matrix is generated for each participant. A whole brain functional connectivity matrix is not generated when head motion exceeds 1.9 degrees of rotation or 1.9mm displacement in any 12-second interval, or when thresholds are exceeded in terms of the signal to fluctuation noise ratio (SFNR), spikiness of signals per slice and per brain image volume, velocity of the mean signal throughout the brain, percentage of "outlier" voxels in each timepoint, smoothness of the data.

Microbleeds

Susceptibility-weighted images are reconstructed from the 3D GRE data and microbleeds are counted (currently under development).

Magnetic susceptibility

The multi-echo 3D gradient-echo data is used to generate and unwrap phase maps using Laplacian-based phase unwrapping in STI Suite (STI.Suite.MRI@gmail.com). The normalized background phase is removed using V-SHARP. Magnetic susceptibility maps are generated using STAR-QSM (Wei et al., NMR Biomed 2015;28:1294-1303). All processing steps are conducted using STI Suite. Magnetic susceptibility maps are not generated when the chi-square for the fit of the multi-echo 3D gradient-echo data to a monoexponential decay exceeds a certain threshold.

Infarcts

Lacunar infarcts larger than 3mm in at least one direction are segmented based on FLAIR, MPRAGE, and T2-weighted FSE images. The total number as well as the volume of the infarcts is calculated. Infarcts are not segmented when FLAIR, MPRAGE, and T2-weighted FSE images are of low signal to noise ratio (SNR) or are contaminated by motion artifacts.


Return to category