Author: Georgios Rovas
This software generates a standardized 2D map of the left atrium by unfolding a 3D mesh and its variables. The code can be used for any variable, such as those derived from imaging (CT, CMR, MRI, LGE-CMR), electrophysiological variables derived from catheterization, and hemodynamical or mechanical variables derived from CFD or FEM analyses.
This software was used for the following publication:
Left atrial wall shear stress distribution correlates with atrial endocardial electrogram voltage and fibrosis in patients with atrial fibrillation. Adamopoulos, D., Rovas, G., Johner, N., Müller, H., Deux, J.-F., Crowe, L.A., Vallée, J.-P., Mach, F., Stergiopulos, N., Shah, D. 2024. medRxiv 2024.07.11.24310174. https://doi.org/10.1101/2024.07.11.24310174
Please cite this reference if you are using the software.
The software has been tested on Windows and Linux systems. Usage in macOS should also be possible by installing/compiling the necessary dependencies.
The scripts in this repository were successfully run with:
Clone the repository and cd into it:
git clone https://github.com/g-rov/lausm
cd lausmSet paths according to your system in constants.py if the dependencies are already installed.
MATLAB_BIN_PATH
CURRENTS_BUILD_PATH
MESHLABSERVER_PATHThe default parameters of the algorithm can be modified in main.py inside the get_parameters() method.
Download MeshLab 1.3.3 and install it. Update the installation location in constants.py.
Download and install MATLAB. Install the MinGW-w64 Compiler from the MATLAB add-ons menu. Update the installation location in constants.py.
The easiest way to install everything you need is by using conda. It includes Python, NumPy, VMTK and VTK.
First, download and install conda, anaconda or miniconda package:
Create an environment and activate it:
conda create --name vmtk python=3.6.4
conda activate vmtkInstall vmtk, vtk, itk and NumPy:
conda install -c vmtk vmtk=1.4.0 itk vtk
Anytime you wish to use vmtk (e.g. to run the code in this repository) you will need to activate the environment: On Windows:
conda activate vmtk
or on Linux:
source activate vmtk
From the MATLAB command prompt execute the following starting from the parent directory:
cd currents_build
mex -I./ convol.c -llibfftw3-3 -L./
mex -I./ gridOptim.c -llibfftw3-3 -L./
mex -I./ projConvol.cpp -llibfftw3-3 -L./
To test the installation run in MATLAB:
match2vtks('./test/source.vtk','./test/target.vtk','./test/output','1','0.0001')
It should take 5-15 mins to run. Then, check in VMTK, Paraview or equivalent software that the 'output.vtk' and 'expected_output.vtk' look the same. For example, you can use the following commands in VMTK's PypePad:
vmtksurfaceviewer -ifile ./currents_build/test/output.vtk
If you encounter errors during this process please read currents_build/README for additional instructions.
If compliled with the instructions above, there is no need to change CURRENTS_BUILD_PATH in constants.py.
Run the following command:
python ./main.py --meshfile ./data/test/test.vtk --datatype tavf
Compare the generate images in the data/test folder with those in data/expected_output; they should be identical. The 3D model used for the test is a mock atrial mesh on which the following mock scalars have been added: TAWSS, Age, BV, Fibrosis. The mock 3D model colored by those scalars can be viewed in the data/test/test_*.png.
The pipeline is split in four parts.
- run_standardization: standardizes meshes from a raw mesh. Depends on VMTK, VTK and MeshLab
- run_currents: registers mesh to an atlas using currents registration. Depends on VTK, MATLAB and currents_build
- run_sum: computes standardized unfold map. Depends on VMTK and VTK
- run_quantification: computes regional quantification (extent or mean value). Depends on VMTK and VTK
main.py [-h] --meshfile MESHFILE --datatype DATATYPE
[--pvcliptype PVCLIPTYPE]
[--paired_unfold_disk PAIRED_UNFOLD_DISK]
[--mitral_clip_type MITRAL_CLIP_TYPE] [--use_seed_selector]
[--use_laa_seed] [--use_similarity] [--use_glyphs]
[--visualize] [--skip_standardization] [--skip_currents]
[--skip_sum] [--skip_quantification]
-h, --help show this help message and exit
--meshfile MESHFILE Full path to mesh file in VTK format
--datatype DATATYPE Data type to process: tavf | scalar | force | lge | lat
TAVF: (T)ime-averaged wall shear stress, (A)ge of blood,
bipolar (V)oltage, and (F)ibrosis
Scalar: Generic type used for a single scalar field
Force|LGE|LAT: Older datatypes for catheter contact force,
LGE-CMR and local activation time.
--pvcliptype PVCLIPTYPE How to clip the pulmonary veins: short | long
--paired_unfold_disk PAIRED_UNFOLD_DISK
Full path to a paired unfold disk. The edges of the
current unfold will be overlaid on the paired unfold
disk
--mitral_clip_type MITRAL_CLIP_TYPE
The algorithm will compute a mitral plane based on the
body and PVs centroids. If the meshfile already
contains a mitral plane clip, use the 'manual' option
to preserve it
--use_seed_selector Seed selection will always run for a new case. To
select new seeds, activate this flag
--use_laa_seed Activate seed for appendage
--use_similarity The method uses an affine transform to initialize mesh
registration. Activate this flag to use a similarity
transform
--use_glyphs Activate glyph on visualization. Recommended for spare
measurements (e.g. some lat meshes)
--visualize Activate visualization
--skip_standardization Skip run_standardization step
--skip_currents Skip run_currents step
--skip_sum Skip run_sum step
--skip_quantification Skip run_quantification step
The .vtk file used as MESHFILE should have a specific structure outlined below:
# vtk DataFile Version 3.0
vtk output
ASCII
POINT_DATA N
SCALARS tawss float 1
LOOKUP_TABLE default
{ N x 3 } table
FIELD FieldData 3
age 1 N float
{ N x 1 } scalar
bv 1 N float
{ N x 1 } scalar
fibr 1 N float
{ N x 1 } scalar
Where N is the number of points.
You may use data/test/test.vtk as a template. Depending on the software used to generate the .vtk mesh, some subsequent editing of the file in a text editing software might be required to reach the desired structure.
- The
--visualizeflag is helpful while learning to use the software and for troubleshooting. - Press
qto advance the visualizations. - If registration (currents) fails, try switching the flipping flag in
processing.pyline 227. - If mitral clipping is too high or too low or if the appendage is clipped, adjust the added radius in
SUM_utils.pyline 831.
The following usage of the code is maintained for compatibility purposes:
python ./main.py --meshfile ./data/outdated/force/mock_force.vtk --datatype force
python ./main.py --meshfile ./data/outdated/lge/mock_lge.vtk --datatype lge --paired_unfold_disk ./data/outdated/force/mock_force_FTI_disk_uniform.vtp
python ./main.py --meshfile ./data/outdated/lat/mock_lat.vtk --datatype lat
python ./main.py --meshfile ./data/outdated/lat/mock_lat.vtk --datatype lat --use_glyph --skip_standardization --skip_currents
It can be used to generate overlayed contour+line plots ("paired") and scatterplots ("glyph").
v2. https://github.com/g-rov/lausm
Left atrial wall shear stress distribution correlates with atrial endocardial electrogram voltage and fibrosis in patients with atrial fibrillation. Adamopoulos, D., Rovas, G., Johner, N., Müller, H., Deux, J.-F., Crowe, L.A., Vallée, J.-P., Mach, F., Stergiopulos, N., Shah, D. 2024. medRxiv 2024.07.11.24310174. https://doi.org/10.1101/2024.07.11.24310174
Changelog:
- Added Windows support.
- The code was updated to Python 3.
- The libraries and dependencies were updated to recent versions.
- Added multiple scalar field support and export options.
- Fixed consistensy of plots.
v1. https://github.com/catactg/sum
Standardized unfold mapping: a technique to permit left atrial regional data display and analysis. Williams SE, Tobon-Gomez C, Zuluaga MA, Chubb H, Butakoff C, Karim R, Ahmed E, Camara O, Rhode KS. J Interv Card Electrophysiol. 2017 Sep 7. doi: 10.1007/s10840-017-0281-3.
Please cite these references when using this code.
BSD 3-Clause