/ThreePoint4DFlow

Code for Machine Learning prediction of velocities from complex data

Primary LanguagePython

ThreePoint4DFlow

Code for Machine Learning prediction of velocities from complex data. From the paper:

Kim D, Jen M-L,Eisenmenger LB, Johnson KM. Accelerated 4D-flow MRI with 3-point encoding enabledby machine learning.Magn Reson Med.2023;89:800-811. doi: 10.1002/mrm.29469 ( https://onlinelibrary.wiley.com/doi/epdf/10.1002/mrm.29469 )

Data Preparation

Data is preared in convert_datasets.py where the data is in HDF5 format with the structure:

  1. '/IMAGE'

    • Size: 6x320x320x320
    • Datatype: H5T_IEEE_F32LE (single)
    • Description: Raw 3 point flow data from complex images stores as 6 channels (real encode 0, imag coil 0, real encode 1, ...)

  2. '/VELOCITY'

    • Size: 3x320x320x320
    • Datatype: H5T_IEEE_F32LE (single)
    • Description: Three channel velocity groundtruth

  3. '/WEIGHT'

    • Size: 320x320x320
    • Datatype: H5T_IEEE_F32LE (single)
    • Description: Magnitude image, weighting is handled in load_4Dflow threepoint_io.py

Data Training

Data is trained in blocks with threepoint_train.py

Inference

Inference is in threepoint_check.py. It expects data in the same format as the training data.