Nonparametric image and shape (2D or 3D) regression by acceleration controlled diffeomorphisms. If you use this software for research, please cite the following papers.
- Fishbaugh, J. and Gerig, G. Acceleration Controlled Diffeomorphisms For Nonparametric Image Regression. IEEE ISBI. 2019
- Fishbaugh, J., Durrleman, S., Gerig, G. Estimation of smooth growth trajectories with controlled acceleration from time series shape data. MICCAI. 2011.
Requirements
This software is built from the source code of Deformetrica. It is recommended you install deformetrica following the directions at http://www.deformetrica.org, which will install necessary dependencies.
Running the application
The application is called with the command:
acceleration_diffeos.py estimate model.xml data_set.xml --p optimization_parameters.xml
where
model.xml
contains information about the template (baseline) shape as well as hyper-parameters for the deformation model.data_set.xml
contains the paths to the input objects which are the observed data.optimization_parameters.xml
contains optional details about optimization.
Try an example
Navigate to the directory examples/2D/bullseye
and run the command:
../../../acceleration_diffeos.py estimate model.xml data_set.xml --p optimization_parameters.xml
which will output the current model every 50 iterations into examples/2D/bullseye/output/
.