GSECARS Tomography Processing Software

This document describes the IDL software used for reconstructing and viewing tomography data at CARS. All of these IDL routines are available in tar and zip files. The reconstruction is normally done using the tomoRecon reconstruction code, which is written in C++. The idl_tomography.tar file contains the required tomoRecon DLLs (Windows) and shareable libraries (Linux, Mac) to run the IDL software without compiling any code. The tomoRecon source code is also available through the tomoRecon home page. tomoRecon is based on the Gridrec reconstruction algorithm, but with multi-threading to utilize multiple cores on modern workstations.

The normal user interface for this software is through the IDL GUI routine tomo_display. The following screen shot shows the tomo_display interface.

---

IDL tomo_display GUI

The following screen shot shows the screen that provides control for the preproccessing and reconstruction options.

---

IDL tomo_display Options screen

We have broken the tomography data processing down into 3 steps:

• Preprocessing
• Reconstruction
• Visualization

Preprocessing

The preprocessing routine read_tomo_data.pro performs the following steps:

• Reads CCD data files into memory. This is done with read_princeton.pro for data collected with the Princeton Instruments cameras, and with read_nd_netcdf.pro for the netCDF files data collected with the Point Grey and Prosilica cameras.
• Removes zingers (remove_tomo_artifacts.pro)
• Corrects for dark current
• Corrects for white (flat) field
• Writes the corrected data back to disk as a single 3-D file (X, Y, Theta) (write_tomo_volume.pro)

Reconstruction

The reconstruction routine reconstruct_volume.pro performs the following steps when using tomoRecon. (When using filtered backprojection similar steps are used, but all of the code is in IDL itself.)

• Reads the 3-D (X, Y, Theta) file from the preprocessing step into memory in chunks, typically 128 slices.
• Reconstruct each chunk using multiple cores. For each slice it does the following:
  • Builds a corrected sinogram: peforms logarithmization, pads the sinogram, normalizes to air on each side of the object, etc.
  • Removes ring artifacts
  • Reconstructs the slice with the Gridrec FFT-based algorithm
• Writes the reconstructed data back to disk as a single 3-D file (X, Y, Z) (write_tomo_volume.pro)

Visualization

The standard IDL procedures TV and TVSCL are very useful for displaying 2-D data. IDL's SLICER3 procedure is good for displaying 3-D data. In addition there is set of GSECARS-developed routines for visualizing 2-D and 3-D tomography (or other) data.

The reconstructed data are written as 16-bit integer netCDF files. These can be read into ImageJ using the ( ImageJ netCDF plugin jar file) and a small ImageJ script called GSETomo_.ijm that is also in the idl_tomography.tar file. Put both the netCDF_.jar and GSETomo_.ijm files in the ImageJ/Plugins folder. Plugins/GSETomo can then be used to read the reconstructed files.