Assessment of errors caused by X-ray scatter and use of contrast medium when using CT-based attenuation correction in PET.

PURPOSE: Quantitative image reconstruction in positron emission tomography (PET) requires an accurate attenuation map of the object under study for the purpose of attenuation correction. Current dual-modality PET/CT systems offer significant advantages over stand-alone PET, including decreased overall scanning time and increased accuracy in lesion localisation and detectability. However, the contamination of CT data with scattered radiation and misclassification of contrast medium with high-density bone in CT-based attenuation correction (CTAC) are known to generate artefacts in the attenuation map and thus the resulting PET images. The purpose of this work was to quantitatively measure the impact of scattered radiation and contrast medium on the accuracy of CTAC.
METHODS: Our recently developed MCNP4C-based Monte Carlo X-ray CT simulator for modelling both fan- and cone-beam CT scanners and the Eidolon dedicated 3D PET Monte Carlo simulator were used to generate realigned PET/CT data sets. The impact of X-ray scattered radiation on the accuracy of CTAC was investigated through simulation of a uniform cylindrical water phantom for both a commercial fan-beam multi-slice and a prototype cone-beam flat panel detector-based CT scanner. The influence of contrast medium was studied by simulation of a cylindrical phantom containing different concentrations of contrast medium. Moreover, an experimental study using an anthropomorphic striatal phantom was conducted for quantitative evaluation of errors arising from the presence of contrast medium by calculating the apparent recovery coefficient (ARC) in the presence of different concentrations of contrast medium.
RESULTS: The analysis of attenuation correction factors (ACFs) for the simulated cylindrical water phantom in both fan- and cone-beam CT scanners showed that the contamination of CT data with scattered radiation in the absence of scatter removal causes underestimation of the true ACFs, namely by 7.3% and 28.2% in the centre for the two geometries, respectively. The ARC was 190.7% for a cylindrical volume of interest located in the main chamber of the striatal phantom containing contrast medium corresponding to 2,000 Hounsfield units, whereas the ARC was overestimated by less than 5% for the main chamber and by approximately 2% for the left/right putamen and caudate nucleus compared with the absence of contrast medium.
CONCLUSION: Without X-ray scatter compensation, the visual artefacts and quantitative errors in flat panel detector-based cone-beam geometry are substantial and propagate cupping artefacts to PET images during CTAC. Likewise, contrast-enhanced CT images may create considerable artefacts during CTAC in regions containing high concentrations of contrast medium.