Cross-camera comparison of SPECT measurements of a 3-D anthropomorphic basal ganglia phantom

SPECT examinations of neurotransmitter systems in the brain have to be comparable between centres to generate a comprehensive data pool, e.g. for multicentre studies. Equipment-specific effects on quantitative evaluations and corresponding methods for compensation, however, have been insufficiently examined. Previous studies have shown that quantitative results may vary significantly according to the imaging equipment used, thereby affecting clinical interpretation of the data. The aim of this study was to determine correction factors for common camera/collimator combinations based on standardised measurements of an anthropomorphic 3D basal ganglia phantom to compensate for the effects of different SPECT camera/collimator equipment. The latter may serve as a model for human studies of the dopaminergic system.

The striatum and background chambers of a commercially available phantom (RSD Alderson) were filled with various( 123)I concentrations encompassing specific striatum/background ratios from 0.6 to 16.1. This setup was imaged with the following four camera/collimator combinations: Siemens Multispect 3 fitted with LEHR and( 123)I parallel-hole collimators, Siemens ECAM with LEHR parallel-hole collimators and Philips Prism 3,000 fitted with LEHR fanbeam collimators, using standardised protocols for acquisition and reconstruction. All scans were automatically co-registered to a SPECT template of the phantom and quantified using a 3D volume of interest (VOI) map based on a CT scan of the phantom. All striatal/background ratios calculated by SPECT were compared with the true ratios calculated from the measurements in a well counter. Regression analyses were performed and recovery correction factors between measured and true ratios determined.

The relation between true and measured ratios could be sufficiently described by a linear regression for each camera/collimator combination without relevant improvement when using second-order polynomial regression models. The recovery correction factors and standard errors were 2.04+/-0.04 for the Philips Prism 3,000, 2.67+/-0.03 for the Siemens Multispect 3/LEHR parallel-hole collimators, 2.15+/-0.03 for the Siemens Multispect 3/(123)I collimators and 2.81+/-0.03 for the Siemens ECAM. Percentage recovery ranged from 36% to 49%.

Measurements of a 3D basal ganglia phantom with various imaging devices revealed linear correlations between measured and true striatal/background ratios. Based on these findings, adjustment of quantitative results between different equipment seems possible, provided that acquisition, reconstruction and evaluation are adequately standardised. The use of identical evaluation methods in phantom and patient studies (comparable shape, size and location of the VOIs) might allow transfer of the calculated correction factors from phantom to studies of the dopaminergic system in patients.