Creatis- Univesity of Lyon || NTNU || Technical University of Denmark


Evaluation framework
Matlab toolbox

The matlab toolbox has been realized and is maintained through a git repository. You can access to the public repository through the following link. You can also directly access to the download page through the following link.

After downloaded, add the folder (without subfolders) to MATLAB's path.

The global architecture of the toolbax is the following:
  • data % folder where to put the data to download
  • examples % folder that contains some examples
  • results % folder that stores the reconstructed images
  • +beamformer % folder that stores the beamformer codes
  • +tools % folder where several useful functions are stored
  • @linear_scan % Object class designed to deal with scanning region
  • @us_dataset % Object class designed to deal with rawdata
  • @us_image % Object class designed to deal with ultrasound image
  • @us_phantom % Object class designed to deal numerical phantom
  • @us_picmus_metrics % Object class designed to deal the different metrics involved in the PICMUS framework


The dataset is composed by one numerical phantom and three in vitro data (both rf and iq version) having the following name:
  • dataset_rf_numerical_transmission_X_nbPW_Y.hdf5
  • dataset_iq_numerical_transmission_X_nbPW_Y.hdf5
  • dataset_rf_in_vitro_type1_transmission_X_nbPW_Y.hdf5
  • dataset_iq_in_vitro_type1_transmission_X_nbPW_Y.hdf5
  • dataset_rf_in_vitro_type2_transmission_X_nbPW_Y.hdf5
  • dataset_iq_in_vitro_type2_transmission_X_nbPW_Y.hdf5
  • dataset_rf_in_vitro_type3_transmission_X_nbPW_Y.hdf5
  • dataset_iq_in_vitro_type3_transmission_X_nbPW_Y.hdf5
X corresponds to the transmission scheme used to generate the dataset. 1 stands for "regular transmission" while 2 stands for "dichotomous scheme". More information on the differences between the two transmission schemes can be found in the Procedure sub-menu.

Y corresponds to the number of steered plane-waves used in transmission to reconstruct the image. Y should be an odd value between 1 and 75.

After downloaded, the participants have to store the dataset inisde the folder named "data" from the matlab toolbox. This can be easily achieved by launching the script provided in the examples folder and named "script_download_full_dataset.m"

The full database is quite big. It is thus also possible to download each dataset file separatly by launching the script provided in the examples folder and named "script_download_specific_dataset.m".