Ctrl+K

Quick Start

Contents

Quick Start#

Demos#

The best way to start is to:

  • Install MBIRJAX in a conda environment and activate this environment using the instructions provided on the Installation Page.

  • Run Demo using the Python program entitled demo_3D_shepp_logan.py located in [mbirjax/demo].

You can then modify the Python script to suit your needs.

Quick Start#

Below are simple instructions on how to do your first reconstruction:

  • Get your data:

    • Import your sinogram data as a 3D numpy array organized by (views, detector rows, detector columns).

    • Create a 1D numpy array called angles that contains the rotation angle in radians of each view.

    • (Optional) Convert arrays both arrays to JAX format using the commands array = jnp.array(array).

    Note that each row of sinogram data is assumed to be perpendicular to the rotation axis and each view is assumed to be in conventional raster order (i.e., left-to-right, top-to-bottom) looking through the object from the source to the detector.

  • Initialize a model:

    • Run model = mbirjax.parallel_beam.ParallelBeamModel(angles, sinogram.shape) to initialize a parallel beam model.

    You will then use the model object to perform the various reconstruction functions.

  • Reconstruct:

    • Run recon = model.recon(sinogram) to reconstruct using MBIR.

    • Then run recon_3d = parallel_model.reshape_recon(recon) to reshape your reconstruction into (rows, columns, slices) format.

    Even the default parameter setting will usually produce a good quality reconstruction.

Contents