Technical Note: Automated quantitative analysis of planar scintigraphic resolution with the ACR SPECT phantom.

Abstract

Evaluating gamma camera image quality is a routine task required for annual physics surveys and laboratory accreditation. A common method is visual assessment of a planar view of the American College of Radiology (ACR) phantom, for which a satisfactory result is to resolve 7.9 mm diameter cold rods with high contrast. The objective was to investigate an automated quantitative measure of planar image quality for more precise evaluation of gamma camera performance. Planar images were automatically co-registered to a template of the phantom's cold rod pattern. Contrast-to-noise ratio (CNR) was measured for each cold rods sector by automatic placement of regions of interests on each rod and on each gap between rods. A quadratic fit to CNR vs. rod diameter yielded the Minimum Detectable Rod Diameter (MDRD) corresponding to a CNR threshold of 4. This methodology was applied to planar images acquired on a variety of gamma cameras according to ACR guidelines and also to planar images generated by Monte Carlo simulation. The automated MDRD analysis had reproducibility of 0.1 mm in repeat phantom scans with random repositioning between scans. The MDRD was 5.8 mm for modern fully digital gamma cameras and 6.2 mm for an older analog gamma camera, whereas visual assessment was identical for all cameras. Simulated Monte Carlo images had MDRD of 5.6 mm using the following parameters: 3.5 mm FWHM intrinsic detector resolution, LEHR collimators, 9.0% FWHM energy resolution, and 15% energy window. Further simulations revealed that collimator design and detector resolution have a substantial impact on MDRD but that energy window width and energy resolution have a minor or negligible impact. The small difference between the Monte Carlo simulations and the modern fully digital gamma cameras was attributed to nonlinearities associated with event positioning and depth-of-interaction effects, which were not modeled in the simulations. Automated MDRD analysis of planar phantom images is a robust technique that provides more precise and meaningful evaluation of camera performance than visual assessment. MDRD is a suitable measure for quality assurance and for annual physics surveys of gamma cameras.