How it works

Packages tomopyui relies on

As mentioned on the home page, tomopyui is built on many open-source python packages including:

tomopy

• Documentation

• Automatic center of rotation finding using tomopy.recon.rotation.find_center and tomopy.recon.rotation.find_center_vo

• Manual center of rotation finding using tomopy.recon.rotation.write_center

• Iterative alignment algorithm

• CPU-based reconstruction algorithms (TK)

• Wrapper for ASTRA algorithms

astra-toolbox

• Documentation

• Tomographic reconstruction toolbox with many GPU-accelerated algorithms

ipywidgets

• Documentation

• The user interface is created inside of JupyterLab using ipywidgets.

cupy

• Documentation

• Sends numpy arrays to the GPU for calculations

bqplot

• Documentation

bqplot-image-gl

• pip

tomocupy

• Documentation TK

• Utilizes GPU-acceleration to speed up automatic alignment

• Built from tomopy, but sends data to the GPU to do some of the calculations.

• Only a few functions available now, but could be expanded significantly

• Included as a module in tomopyui (for now)

• Helps run a lot of the backend of tomopyui

Structure

The code is divided into the frontend (tomopyui.widgets) and the backend (tomopyui.backend and tomopyui.tomocupy).

When the user calls tomopyui.widgets.main.create_dashboard, they are creating several different objects:

1. Import tab:

   • The type of Import tab is dictated by the user’s choice in string on the call from the Jupyter notebook (e.g., “SSRL_62C” or “ALS_832”)

   • Holds both raw and prenormalized Uploaders, which hold a specific type of RawProjections (depending on the choice above) and a general Prenormalized_Projections object, respectively.

   • Uploaders upload the data (hold buttons, viewers, etc.), and Projections hold the data.

   • Projection objects also hold specific types of metadata. Storing metadata along the processing pipeline is critical.

2. Center tab:

   • Used to interactively find the center of rotation

   • Contains two Accordion widgets:

     1. tomopy.recon.rotation.find_center and tomopy.recon.rotation.find_center_vo for automatic center finding.

     2. Reconstruction of a given slice of the data at various centers. The user can then maneuver the slider to find the best center by eye.

3. Align tab:

   • Used to align jittery data using the automatic alignment algorithms built into tomopy.

   • Align only a part of a dataset by selecting an ROI in the “Imported Projections” viewer.

4. Recon tab:

   • Used to reconstruct data using one of many algorithms available from TomoPy and Astra Toolbox.

5. Data Explorer tab:

   • Used to look at prior alignments.

All of these objects talk to view.py, which holds the image viewers.