Index

Name

mri_ca_label

Description

This program is used to label subcortical structures based in GCA model.

Synopsis

mri_ca_label [<options>] invol1 [invol2 ...] xform gcafile outvol

Arguments

Positional Arguments

invol1 [invol2 ...]

input volume(s)

xform

transform file

gcafile

Gaussian classifier atlas file

outvol

output volume

Required Flagged Arguments

None

Optional Flagged Arguments

-cross-sequence

label a volume acquired with sequence different than atlas

-nogibbs

disable gibbs priors

-wm path

use wm segmentation

-conform

interpolate volume to be isotropic 1mm^3

-topo_dist_thresh dist

do not relabel ventricle segments that are disconnected from the main body of the ventricle, are less than dist mm from the largest segment, and have a volume greater than topo_volume_thresh1

-topo_volume_thresh1 volume

do not relabel ventricle segments that are disconnected from the main body of the ventricle, are less than dist mm from the largest segment, and have a volume greater than volume.

-topo_volume_thresh2 volume

do not relabel ventricle segments that are disconnected from the main body of the ventricle and have a volume greater than volume

-t1 gca_t1

use file to label thin temporal lobe

-normpd

normalize PD image to GCA means

-debug_voxel x y z

debug voxel

-debug_node x y z

debug node

-debug_label <int n>

debug label

-tr TR

set TR in msec

-te TE

set TE in msec

-alpha ALPHA

set alpha in radians

-example mri_vol segmentation

use T1 (mri_vol) and segmentation as example

-pthresh thresh

use p threshold n for adaptive renormalization (default=.7)

-niter n

apply max likelihood for n iterations (default=2)

-novar

do not use variance in classification

-regularize reg

regularize variance to be sigma+nC(noise)

-nohippo

do not auto-edit hippocampus

-fwm mri_vol

use fixed white matter segmentation from wm

-mri mri_vol

write most likely MR volume to mri_vol

-heq mri_vol

use histogram equalization from mri_vol

-renorm mri_vol

renormalize using predicted intensity values in mri_vol

-flash

use FLASH forward model to predict intensity values

-flash_params filename

use FLASH forward model and tissue params in filename to predict

-renormalize wsize iter

reenorm class means [iter] times after initial label with window of [wsize]

-r mri_vol

set input volume

-h

use GCA to histogram normalize input image

-a n

mean filter n time to conditional densities

-w n filename

write snapshots of gibbs process every n times to filename

-m mri_vol

use mri_vol to mask final labeling

-e n

expand

-n n

set max iterations to n (default=200)

-f f thresh

filter labeled volume with threshold thresh (default=.5) mode filter f (default=0)times

-nowmsa

disables WMSA labels (hypo/hyper-intensities), selects second most probable label for each WMSA labelled voxel

-write_probs <char *filename>

Write label probabilities to filename.

-L <mri_vol> <LTA>

Longitudinal processing mri_vol is label from tp1, LTA is registration from tp1 to current data

-RELABEL_UNLIKELY <1/0> <wsize> <sigma> <thresh>

Reclassify voxels at least <thresh> standard devs from the mean using a <wsize> Gaussian window (with <sigma> standard devs) to recompute priors and likelihoods.

Outputs

outvol

output volume

Bugs

None

See Also

mri_cc

Links

FreeSurfer, FsFast

Reporting Bugs

Report bugs to <analysis-bugs@nmr.mgh.harvard.edu>

Author/s

BruceFischl