DTI_Register

Register DTI to the fixed volume using DTIReg

Command:

dmriprep

Module version:

0.1

Module type:

prep

Attributes:

utility, indentity

Notes:

DTI_Register README

Source:

dtiplayground/dmri/preprocessing/modules/DTI_Register

Global variables

Values this module shares with the other modules of the pipeline (dmriprep run -g <name>=<value> sets one).

  • dti_path

  • displacement_field_path

  • inverse_displacement_field_path

  • registered_dti_path

  • initial_affine_path

  • registered_metric_paths

  • reference_dti_path

  • tensor_flip_applied

Command line variables

Values this module shares with the other modules of the pipeline (dmriprep run -g <name>=<value> sets one).

  • reference_dti

  • reference_normative_model

  • age

  • tensor_flip

Protocol options

method

Method to register DTI (ANTs)

Type:

list

Default:

ANTs

Choices:

  • ANTs – Use ANTs to register

referenceImage

Reference (Fixed Image)

Type:

string

Default:

not set

referenceNormativeModel

Normative model of the reference atlas (folder with manifest.json and <age bin>/DTI_mean.nrrd, written by ‘dmrifiberprofile qc-registration –build-normative’). The DTI is then registered to the mean tensor of the age bin of the subject instead of the reference image; without an age, or a matching bin, the reference image is used

Type:

dirpath-remote

Default:

not set

age

Age of the subject in the unit of the age bins of the normative model (months); taken from ageCSV, or read from the image path with ageRegex, if not set

Type:

float

Default:

not set

ageCSV

CSV/TSV with the age of each scan (as ‘dmrifiberprofile qc-registration –age-csv’), for cohorts whose session names don’t carry the age. Subject and session columns are detected (participant_id/session_id and variants); one row per subject applies to all of its sessions. The subject and session of the scan are read from its path (sub-<id>, ses-<id>)

Type:

filepath-remote

Default:

not set

ageColumn

Name of the ageCSV column holding the age (default - auto-detect - age, age_months, candidate_age, …)

Type:

string

Default:

not set

ageUnits

Units of the age column of ageCSV (converted to the months of the age bins)

Type:

list

Default:

months

Choices:

  • months – Ages of ageCSV are in months

  • years – Ages of ageCSV are in years

  • weeks – Ages of ageCSV are in weeks

  • days – Ages of ageCSV are in days

ageRegex

Regular expression for the age in the path of the input image (group 1), used when neither age nor ageCSV gives one

Type:

string

Default:

ses-(\d+)m

ANTsPath

ANTs installation directory (default is dtiplayground-tools/ANTs)

Type:

dirpath-remote

Default:

$ANTSDIR

ANTsMethod

ANTS method

Type:

string

Default:

useScalar-ANTS

registrationType

Registration Type

Type:

list

Default:

GreedyDiffeo

Choices:

  • GreedyDiffeo – Greedy Diffeo

similarityMetric

Similarity Metric

Type:

list

Default:

CC

Choices:

  • CC – Cross Correlation

similarityParameter

Similarity Parameter (radius of the CC metric)

Type:

number

Default:

2

ANTsIterations

Iteration parameter for ANTS

Type:

string

Default:

100x50x20

gaussianSigma

Gaussian Sigma

Type:

number

Default:

1

ANTsTransformationStep

Gradient step of the ANTS transformation

Type:

number

Default:

0.25

ANTsUseHistogramMatching

Match the histograms of the scalar images before the ANTS registration

Type:

boolean

Default:

true

scalarMeasurement

Scalar image of the tensors used for the registration (DTI-Reg and BRAINSFit)

Type:

list

Default:

FA

Choices:

  • FA – Register the FA images of the tensors

  • MD – Register the MD images of the tensors

tensorCorrection

Correction of tensors with negative eigenvalues when computing the scalar images and resampling

Type:

list

Default:

abs

Choices:

  • abs – Negative eigenvalues are replaced by their absolute value

  • zero – Negative eigenvalues are set to zero

  • nearest – Nearest positive definite tensor

  • none – No correction

initialAffine

Initial affine transform of the DTI to the reference, refined by the ANTS registration

Type:

list

Default:

BRAINSFit

Choices:

  • BRAINSFit – Affine registration of the scalar images with BRAINSFit (fixed = reference, moving = DTI)

  • file – Use the transform given in initialAffineFile

  • none – No initial affine transform

initialAffineFile

ITK transform file (fixed = reference, moving = DTI), used if initialAffine is ‘file’

Type:

string

Default:

not set

tensorFlip

Flip of the tensor frame applied to the DTI before the registration: none (also if empty and the global variable tensor_flip isn’t set), auto (detected: agreement of the principal directions with the reference after the initial affine, or their coherence along the tracts without an initial affine), the axes to flip (e.g. x or x,z), or voxel for components in the frame of the voxel axes (rotated into the space of the header; with flips e.g. voxel,x)

Type:

string

Default:

not set

tensorFlipFAThreshold

FA threshold of the white matter voxels used by the automatic flip detection

Type:

number

Default:

0.3

BRAINSFitTransforms

Comma delimited BRAINSFit transform stages (Rigid, ScaleVersor3D, ScaleSkewVersor3D, Affine), run in this order

Type:

string

Default:

Rigid,Affine

BRAINSFitInitializeTransformMode

Initialization of the BRAINSFit registration

Type:

list

Default:

useCenterOfHeadAlign

Choices:

  • useCenterOfHeadAlign – Align the centers of the heads

  • useMomentsAlign – Align the centers of mass and principal axes

  • useGeometryAlign – Align the image centers

  • Off – No initialization

BRAINSFitSamplingPercentage

Fraction (0-1) of the voxels sampled for the BRAINSFit metric. Default 0.5; smaller values are faster but less reproducible (BRAINSFit default 0.002 gave initial affines several mm apart between repeated runs)

Type:

number

Default:

0.5

registerMetrics

Apply the displacement field to the diffusion metrics in the folder of the input DTI that share its file name prefix (e.g. <scan>_dwi_QCed_FA.nii.gz for <scan>_dwi_QCed_tensor.nrrd). Tensor images (e.g. free-water corrected tensors) are resampled log-Euclidean with reorientation like the DTI, scalar images linearly. Written as registered_<name>

Type:

boolean

Default:

true

metricExclude

Comma delimited parts of file names that are not registered (e.g. brain masks); integer images and images that are neither scalar nor tensor are not registered either

Type:

string

Default:

mask