SYNOPSIS

mia-2dmyoica-nonrigid2 -i <in-file> -o <out-file> [options]

DESCRIPTION

mia-2dmyoica-nonrigid2 This program runs the non-rigid registration of an perfusion image series.In each pass, first an ICA analysis is run to estimate and eliminate the periodic movement and create reference images with intensities similar to the corresponding original image. Then non-rigid registration is run using the an "ssd + divcurl" cost model. The B-spline c-rate and the divcurl cost weight are changed in each pass according to given parameters.In the first pass a bounding box around the LV myocardium may be extractedto speed up computation Special note to this implemnentation: the registration is always run from the original images to avoid the accumulation of interpolation errors.

OPTIONS

File-IO

-i --in-file=(input,required)

input perfusion data set

-o --out-file=(output,required)

output perfusion data set

-r --registered=reg

file name base for registered fiels

--save-cropped=

save cropped set to this file

--save-feature=

save segmentation feature images and initial ICA mixing matrix

ICA

-C --components=0

ICA components 0 = automatic estimation

--normalize

don't normalized ICs

--no-meanstrip

don't strip the mean from the mixing curves

-s --segscale=0

segment and scale the crop box around the LV (0=no segmentation)

-k --skip=0

skip images at the beginning of the series e.g. because as they are of other modalities

-m --max-ica-iter=400

maximum number of iterations in ICA

-E --segmethod=features

Segmentation method

delta-peak \(hy difference of the peak enhancement images

features \(hy feature images

delta-feature \(hy difference of the feature images

Registration

-O --optimizer=gsl:opt=gd,step=0.1

Optimizer used for minimization For supported plugins see PLUGINS:minimizer/singlecost

-a --start-c-rate=32

start coefficinet rate in spines, gets divided by --c-rate-divider with every pass

--c-rate-divider=4

cofficient rate divider for each pass

-d --start-divcurl=20

start divcurl weight, gets divided by --divcurl-divider with every pass

--divcurl-divider=4

divcurl weight scaling with each new pass

-w --imageweight=1

image cost weight

-p --interpolator=bspline:d=3

image interpolator kernel For supported plugins see PLUGINS:1d/splinekernel

-l --mg-levels=3

multi-resolution levels

-P --passes=3

registration passes

Help & Info

-V --verbose=warning

verbosity of output, print messages of given level and higher priorities. Supported priorities starting at lowest level are:

info \(hy Low level messages

trace \(hy Function call trace

fail \(hy Report test failures

warning \(hy Warnings

error \(hy Report errors

debug \(hy Debug output

message \(hy Normal messages

fatal \(hy Report only fatal errors

--copyright

print copyright information

-h --help

print this help

-? --usage

print a short help

--version

print the version number and exit

Processing

--threads=-1

Maxiumum number of threads to use for processing,This number should be lower or equal to the number of logical processor cores in the machine. (-1: automatic estimation).

PLUGINS: 1d/splinekernel

bspline

B-spline kernel creation , supported parameters are:

d = 3 (int)

Spline degree. in [0, 5]

omoms

OMoms-spline kernel creation, supported parameters are:

d = 3 (int)

Spline degree. in [3, 3]

PLUGINS: minimizer/singlecost

gdas

Gradient descent with automatic step size correction., supported parameters are:

ftolr = 0 (double)

Stop if the relative change of the criterion is below.. in [0, INF]

max-step = 2 (double)

Minimal absolute step size. in [1, INF]

maxiter = 200 (uint)

Stopping criterion: the maximum number of iterations. in [1, 2147483647]

min-step = 0.1 (double)

Maximal absolute step size. in [1e-10, INF]

xtola = 0.01 (double)

Stop if the inf-norm of the change applied to x is below this value.. in [0, INF]

gdsq

Gradient descent with quadratic step estimation, supported parameters are:

ftolr = 0 (double)

Stop if the relative change of the criterion is below.. in [0, INF]

gtola = 0 (double)

Stop if the inf-norm of the gradient is below this value.. in [0, INF]

maxiter = 100 (uint)

Stopping criterion: the maximum number of iterations. in [1, 2147483647]

scale = 2 (double)

Fallback fixed step size scaling. in [1, INF]

step = 0.1 (double)

Initial step size. in [0, INF]

xtola = 0 (double)

Stop if the inf-norm of x-update is below this value.. in [0, INF]

gsl

optimizer plugin based on the multimin optimizers ofthe GNU Scientific Library (GSL) https://www.gnu.org/software/gsl/, supported parameters are:

eps = 0.01 (double)

gradient based optimizers: stop when |grad| < eps, simplex: stop when simplex size < eps.. in [1e-10, 10]

iter = 100 (int)

maximum number of iterations. in [1, 2147483647]

opt = gd (dict)

Specific optimizer to be used.. Supported values are:

bfgs \(hy Broyden-Fletcher-Goldfarb-Shann

bfgs2 \(hy Broyden-Fletcher-Goldfarb-Shann (most efficient version)

cg-fr \(hy Flecher-Reeves conjugate gradient algorithm

gd \(hy Gradient descent.

simplex \(hy Simplex algorithm of Nelder and Mead

cg-pr \(hy Polak-Ribiere conjugate gradient algorithm

step = 0.001 (double)

initial step size. in [0, 10]

tol = 0.1 (double)

some tolerance parameter. in [0.001, 10]

nlopt

Minimizer algorithms using the NLOPT library, for a description of the optimizers please see 'http://ab-initio.mit.edu/wiki/index.php/NLopt_Algorithms', supported parameters are:

ftola = 0 (double)

Stopping criterion: the absolute change of the objective value is below this value. in [0, INF]

ftolr = 0 (double)

Stopping criterion: the relative change of the objective value is below this value. in [0, INF]

higher = inf (double)

Higher boundary (equal for all parameters). in [INF, INF]

local-opt = none (dict)

local minimization algorithm that may be required for the main minimization algorithm.. Supported values are:

gn-orig-direct-l \(hy Dividing Rectangles (original implementation, locally biased)

gn-direct-l-noscal \(hy Dividing Rectangles (unscaled, locally biased)

gn-isres \(hy Improved Stochastic Ranking Evolution Strategy

ld-tnewton \(hy Truncated Newton

gn-direct-l-rand \(hy Dividing Rectangles (locally biased, randomized)

ln-newuoa \(hy Derivative-free Unconstrained Optimization by Iteratively Constructed Quadratic Approximation

gn-direct-l-rand-noscale \(hy Dividing Rectangles (unscaled, locally biased, randomized)

gn-orig-direct \(hy Dividing Rectangles (original implementation)

ld-tnewton-precond \(hy Preconditioned Truncated Newton

ld-tnewton-restart \(hy Truncated Newton with steepest-descent restarting

gn-direct \(hy Dividing Rectangles

ln-neldermead \(hy Nelder-Mead simplex algorithm

ln-cobyla \(hy Constrained Optimization BY Linear Approximation

gn-crs2-lm \(hy Controlled Random Search with Local Mutation

ld-var2 \(hy Shifted Limited-Memory Variable-Metric, Rank 2

ld-var1 \(hy Shifted Limited-Memory Variable-Metric, Rank 1

ld-mma \(hy Method of Moving Asymptotes

ld-lbfgs-nocedal \(hy None

ld-lbfgs \(hy Low-storage BFGS

gn-direct-l \(hy Dividing Rectangles (locally biased)

none \(hy don't specify algorithm

ln-bobyqa \(hy Derivative-free Bound-constrained Optimization

ln-sbplx \(hy Subplex variant of Nelder-Mead

ln-newuoa-bound \(hy Derivative-free Bound-constrained Optimization by Iteratively Constructed Quadratic Approximation

ln-praxis \(hy Gradient-free Local Optimization via the Principal-Axis Method

gn-direct-noscal \(hy Dividing Rectangles (unscaled)

ld-tnewton-precond-restart \(hy Preconditioned Truncated Newton with steepest-descent restarting

lower = -inf (double)

Lower boundary (equal for all parameters). in [INF, INF]

maxiter = 100 (int)

Stopping criterion: the maximum number of iterations. in [1, 2147483647]

opt = ld-lbfgs (dict)

main minimization algorithm. Supported values are:

gn-orig-direct-l \(hy Dividing Rectangles (original implementation, locally biased)

g-mlsl-lds \(hy Multi-Level Single-Linkage (low-discrepancy-sequence, require local gradient based optimization and bounds)

gn-direct-l-noscal \(hy Dividing Rectangles (unscaled, locally biased)

gn-isres \(hy Improved Stochastic Ranking Evolution Strategy

ld-tnewton \(hy Truncated Newton

gn-direct-l-rand \(hy Dividing Rectangles (locally biased, randomized)

ln-newuoa \(hy Derivative-free Unconstrained Optimization by Iteratively Constructed Quadratic Approximation

gn-direct-l-rand-noscale \(hy Dividing Rectangles (unscaled, locally biased, randomized)

gn-orig-direct \(hy Dividing Rectangles (original implementation)

ld-tnewton-precond \(hy Preconditioned Truncated Newton

ld-tnewton-restart \(hy Truncated Newton with steepest-descent restarting

gn-direct \(hy Dividing Rectangles

auglag-eq \(hy Augmented Lagrangian algorithm with equality constraints only

ln-neldermead \(hy Nelder-Mead simplex algorithm

ln-cobyla \(hy Constrained Optimization BY Linear Approximation

gn-crs2-lm \(hy Controlled Random Search with Local Mutation

ld-var2 \(hy Shifted Limited-Memory Variable-Metric, Rank 2

ld-var1 \(hy Shifted Limited-Memory Variable-Metric, Rank 1

ld-mma \(hy Method of Moving Asymptotes

ld-lbfgs-nocedal \(hy None

g-mlsl \(hy Multi-Level Single-Linkage (require local optimization and bounds)

ld-lbfgs \(hy Low-storage BFGS

gn-direct-l \(hy Dividing Rectangles (locally biased)

ln-bobyqa \(hy Derivative-free Bound-constrained Optimization

ln-sbplx \(hy Subplex variant of Nelder-Mead

ln-newuoa-bound \(hy Derivative-free Bound-constrained Optimization by Iteratively Constructed Quadratic Approximation

auglag \(hy Augmented Lagrangian algorithm

ln-praxis \(hy Gradient-free Local Optimization via the Principal-Axis Method

gn-direct-noscal \(hy Dividing Rectangles (unscaled)

ld-tnewton-precond-restart \(hy Preconditioned Truncated Newton with steepest-descent restarting

ld-slsqp \(hy Sequential Least-Squares Quadratic Programming

step = 0 (double)

Initial step size for gradient free methods. in [0, INF]

stop = -inf (double)

Stopping criterion: function value falls below this value. in [INF, INF]

xtola = 0 (double)

Stopping criterion: the absolute change of all x-values is below this value. in [0, INF]

xtolr = 0 (double)

Stopping criterion: the relative change of all x-values is below this value. in [0, INF]

EXAMPLE

Register the perfusion series given in 'segment.set' by using automatic ICA estimation. Skip two images at the beginning and otherwiese use the default parameters. Store the result in 'registered.set'. mia-2dmyoica-nonrigid2 -i segment.set -o registered.set -k 2

AUTHOR(s)

Gert Wollny

COPYRIGHT

This software is Copyright (c) 1999\(hy2013 Leipzig, Germany and Madrid, Spain. It comes with ABSOLUTELY NO WARRANTY and you may redistribute it under the terms of the GNU GENERAL PUBLIC LICENSE Version 3 (or later). For more information run the program with the option '--copyright'.