mri_em_register - creates a tranform in lta format


mri_em_register [options] <in brain volume> <template gca> <output transform name>


Positional Arguments

<in brain volume>

input volume

<template gca>

template gca

<output transform name>

output transform

Optional Flagged Arguments

-dist distance



use FLASH forward model to predict intensity values

-mask <volname>

use volname as a mask


align to atlas containing skull (uns=5)

-uns nbrspacing

align to atlas containing skull setting unknown_nbr_spacing = nbrspacing

-diag diagfile

open diagfile for writing

-debug_voxel x y z

debug voxel (x, y, z)

-debug_label label

debug label (label)

-tr TR

use TR msec

-te TE

use TE msec

-alpha alpha

use alpha degrees

-example T1 seg

use T1 and seg as example T1 and segmentations respectively

-samples fname

write control points to fname

-fsamples fname

write transformed control points to fname

-nsamples fname

write transformed normalization control points to fname


use contrast to find labels

-flash_parms parameterfile

use FLASH forward model and tissue parms in parameterfile to predict intensity values


only compute translation parameters

-write_mean fname

write gca means to fname

-prior min_prior

use prior threshold min_prior

-spacing <int>

use max GCA spacing

-scales <int>

find optimal linear transform over int scales


do not use variance estimates

-dt dt

-tol tol


use GCA centroid as origin of transform


disable scaling


disable intensity scaling

-num num_xforms

find a total of num_xforms linear transforms

-area area

-nlarea nlarea

-levels levels

-intensity intensity

-reduce nreductions

reduce input images nreductions times before aligning

-nsamples nsamples

using n samples of GCA

-norm fname

normalize intensity and write to fname

-trans max_trans

setting max translation search range to be max_trans

-steps max_angles

taking max_angles angular steps

-l xform long_reg

Longitudinal: read previously computed atlas xform and apply registration long_reg

-f cpfile

read manually defined control points from cpfile

-d tx ty tz

-r rx ry rz

-t xform

using previously computed transform xform

-b blur_sigma

blurring input image with sigma=blur_sigma

-v diagno

-s max_angles

-max_angle max_angle

max_angle for rotational search in radians (def=15 deg)

-n niters

niterations = niters

-w write_iters

write iterations = write_iters

-p ctl_point_pct

use top pct percent wm points as control points

-m momentum

set momentum



transform in lta format



Example 1

mri_em_register -mask subject1/mri/brain subject1/mri/nu single_one.gca subject1/mri/transforms/talairach_one.lta


none known

See Also


FreeSurfer, FsFast

Methods Description

Fischl et al., 2002 - See "Optimal Linear Transform" section



Reporting Bugs

Report bugs to <>


mri_em_register (last edited 2018-01-26 14:40:26 by RicardoValle)