SPECT Reconstruction Audit

This audit was performed in 2001 and was led by Peter Jarritt. The purpose of the audit was to establish the accuracy of SPECT reconstruction software without any filtering. It used simulated test data from a software phantom without any attenuation or noise added. The test data were reconstructed by users without any additional filtering and the reconstructed transaxial sections analysed to assess quantitative accuracy.

The audit found that although the absolute value of reconstructed counts from a uniform section of the phantom varied between centres, when reconstructed counts were plotted against count density in the phantom, most systems gave a linear relationship. In some cases it was found that different versions of the same software could give different results. The data also included projection images from a real line source which were reconstructed to compare the resolution with the planar resolution of the same line. The results were published in 2002 (Jarritt PH, Whalley DR, Skrypniuk JV, Houston AS, Fleming JS and Cosgriff PS (2002). “UK Audit of single photon emission computed tomography reconstruction software using software generated phantoms”. Nucl. Med. Commun. 23, 483-491.)

The data used for this audit can be used to test any SPECT reconstruction software (either using FBP or iterative reconstruction) without any smoothing filter applied. Note that the NMSQG have performed a separate audit on the use of filters in SPECT. The data for the SPECT reconstruction audit are available in either Interfile or DICOM format together with full instructions for reconstruction and analysing the results.

Dataset download – interfile

Dataset download – DICOM

Instructions PDF

Leave a Reply