SNARK09 - A programming system for the reconstruction of 2D images from 1D projections

What is SNARK09?

The reconstruction problem has arisen in a large number of scientific fields (including computed tomography, electron microscopy, radiology, radio astronomy and holography). Many different methods (algorithms) have been suggested for its solution.

SNARK09 is a programming system for the reconstruction of 2D images from 1D projections. It is designed to help researchers interested in developing and evaluating reconstruction algorithms.

In the area of image reconstruction, researchers often desire to compare two or more reconstruction techniques and assess their relative merits. SNARK09 provides a uniform framework in which to implement algorithms and evaluate their performance. SNARK09 has been designed to treat both parallel and divergent projection geometries and can create test data for use by reconstruction algorithms. A number of frequently used reconstruction algorithms are incorporated.

New in SNARK09

SNARK09 is an updated version of SNARK05. The following are the major advances that are incorporated into the SNARK09 package:

• capability of generating mathematically described phantoms that include inhomogeneity;
• capability of applying beam hardening correction to polychromatic X-ray projection data;
• the imagewise-region-of-interest figure of merit;
• up to 10 user implemented algorithms, at most five using blobs;
• up to five user defined figures of merit.

History

SNARK09 is a descendant of earlier releases of SNARK, the first one of which was written by Richard Gordon in 1970 in FORTRAN.

SNARK77 and SNARK89 were specifically designed to help with the problem of reconstructing cross-sections of the X-ray absorption coefficient distribution inside the body from X-ray projections. SNARK93 extended this capability to include positron emission tomography, PET.

The SNARK93 programming system was implemented in FORTRAN77. It was designed to

• be capable of dealing with many modes of data collection (different geometrical arrangements of X-ray source and detectors, different X-ray spectra, etc.);
• contain many of the published reconstruction algorithms;
• be capable of generating mathematically described phantoms that realistically represent various crosssections of the human body, together with mathematically simulated projection data of these crosssections reflecting the characteristics (including noise) of various possible tomography devices;
• contain subroutines to carry out work which appears to be common to many reconstruction algorithms, so as to ease the incorporation of additional (user-defined) algorithms;
• be capable of a variety of display modes;
• contain routines for the statistical evaluation of reconstruction algorithms;
• provide a methodology for testing for statistically significant differences between reconstruction algorithms.

The most recent previous version, SNARK05, was implemented in C++. The following were the major advances that were incorporated into the SNARK05 package:

• SNARK05 was implemented in C++;
• the file structures for holding projection data, phantoms and reconstructions had been redesigned to match the capabilities of the typical computer environment at the beginning of the third millennium (specifically, XML headers are used in data files);
• all iterative algorithms were capable of handling image representations that use blobs, as well as those that use pixels;
• the efficient data access ordering was a standard feature;
• all artificial restrictions on sizes were removed - the only remaining restrictions were those imposed by the hardware, compiler and operating system.