What's new - daducci/AMICO GitHub Wiki

New features and improvements from AMICO v2

Multithreading

AMICO now uses multithreading insted of multiprocessing. This allows to further speed up the fitting process. See the Multithreading section for more information.

Principal diffusion direction

The principal diffusion direction is now computed by default using the Ordinary Least Squares (OLS) method instead of the Weighted Least Squares (WLS) method. You can still use the WLS method by setting the DTI_fit_method parameter to WLS. See the Global configuration section for more information.

Configuration parameters

study_path and subject Default study_path and subject parameters in the Evaluation class are now set to '.' instead of None. This means that if you don't set these parameters, AMICO will look for the data in the current directory.

ndirs Default ndirs parameter in the generate_kernels() method is now set to 500 instead of 32761.

DTI_fit_method New parameter to set the method used to fit the diffusion tensor model to compute the principal diffusion directions.

nthreads Now the parameter to set the number of threads used by AMICO during the model fitting is nthreads instead of parallel_jobs.

BLAS_nthreads New parameter to set the number of threads used by BLAS libraries.

doComputeRMSE New parameter to compute the Root Mean Square Error (RMSE) between the predicted and the measured signal.

doSaveModulatedMaps New parameter (specific for the NODDI model) to compute the modulated NDI and ODI maps for Tissue-weighted mean analysis.

NODDI maps names

The NODDI maps are called with names in accordance with the original paper:

  • NDI instead of ICVF
  • ODI instead of OD
  • FWF instead of FISO

License

AMICO is now released under a proprietary license which allows free use for non-commercial purposes. If you are interested in using AMICO for commercial purposes, please contact Alessandro Daducci at [email protected]. You can read the full license text here.