Simul@trophy

Simulating Realistic Longitudinal Images with Atrophy

Requirements

1. Cmake > 2.8
2. PETSc version 3.6.3.1 Use --with-clanguage=cxx option when configuring PETSc.
3. ITK version 4.9

Installation

First make sure that:

• You have right versions of PETSc and ITK.
• That you configured PETSc with --with-clanguage=cxx option.
• Set PETSC_DIR, PETSC_ARCH and ITK_DIR. For example, I have my .bashrc with:

ITK_DIR=/home/bkhanal/Documents/softwares/ITK-build
export ITK_DIR
PETSC_DIR=/home/bkhanal/Documents/softwares/petsc
export PETSC_DIR
PETSC_ARCH=arch-linux2-cxx-debug
export PETSC_ARCH

Now, build simul@trophy with the following steps. I assume that you are inside a curr_dir, replace curr_dir with a directory of your choice.

curr_dir$ git clone https://github.com/Inria-Asclepios/simul-atrophy.git
curr_dir$ cd simul-atrophy
simul-atrophy$ mkdir build && cd build

Use cmake to build the files.

build$ cmake ..
build$ make

If you have ITK installed with c++11, then use the following options with cmake:

build$ cmake -DUSE_CXX11:bool=true ..
build$ make

This should install the software to your machine.

Quickstart

The main simulator executable is build/src/simul_atrophy. Run with -h option to see in detail how to use it.

A basic example

One very basic example of how to use simul@trophy is shown below on a relative small image so that this can be run quickly on a normal laptop/desktop. Note that if your image sizes are too large, usually greater than 120^3 you must use cluster computing for running the model due to memory limitations on normal desktops. Here I have used a dummy object instead of brain, but you can create your own test set by downsampling brain images too. Any file format readable by ITK will work; the example here uses .mha format.

The input files and the results of this basic example can be found at the directory basicExample.

There are three main input files:

1. Input atrophy map: test1Atrophy1.mha

2. Input segmentation file: bMask.mha

3. Input image file: bTest1.mha

The input parameters $\mu$ and $\lambda$ are set to 1 for both tissue and CSF regions.

The example is run for two steps.

$build/src/simul_atrophy -parameters 1,1,1,1 -boundary_condition dirichlet_at_skull --relax_ic_in_csf -atrophyFile basicExample/test1Atrophy1.mha -maskFile basicExample/bMask1.mha -imageFile basicExample/bTest1.mha --invert_field_to_warp -numOfTimeSteps 2 -resPath ./basicExample/ -resultsFilenamesPrefix basicTest1_

If everything is all right you should be getting the following output in your display terminal:

Model will be run for 2 time steps
dmda of size: (21,41,21)
grid with the spacings hx, hy, hz: (1.000000,1.000000,1.000000)
dirichlet_at_skull=>Skull velocity = 0.
Incompressibility constraint relaxed where brain mask has label 1 
Incompressibility constraint relaxation coefficient: 1/lambda.
solver uses discretization for constant viscosity case!
 muBrain=1.000000, muCsf=1.000000, lambdaBrain=1.000000, lambdaCsf=1.000000
First Lame Parameter (lambda) is a scalar.

 RHS will be taken as (mu + lambda)grad(a), i.e. with Lame parameters.

 computing the operator for linear solve with 8 point stencil for divergence

 Relax IC with div(u) + kp = 0 on relax cells

 Displacement field inversion: tolerance not reached in 0 voxels 


 Displacement field inversion: tolerance not reached in 0 voxels 

The output files created will be placed in the folder basicExample. The repository already contains the results of same example with prefix basicTest_. Now, after running this example you should get new files with the prefix basicTest1_.

Instead of using the simul_atrophy executable directly, you can also a run the model by using the python script at 'simul-atrophy/scripts/simul_atrophy.py'. Please see -h option to see the details of the script.