Quantitative accuracy in cardiac studies with the GE discovery RX PET/CT scanner

Abstract

A phantom study was carried out to investigate the accuracy of attenuation and scatter correction for cardiac scanning on the Discovery RX PET/CT in 2D and 3D modes. Methods: A body phantom (RSD/Alderson) containing heart, liver, lungs and skeleton was scanned with 18F in the “myocardium” and (a) with various concentrations of 18F in the liver (placed, alternately, mostly inside or outside the PET FOV) and (b) with, successively, extra attenuation (arms and “fat”) in place. CT scans for attenuation correction were generated at 100, 120 and 140 kV and 10–50 mA. A 20 cm diam multi-compartment phantom was scanned with a cold insert in an active background and varying activity in an adjacent end compartment inside and outside the PET FOV. Images were reconstructed with reprojection (FBP) and OSEM methods. Results: For stable attenuation correction it was found that a tube voltage and current of at least 120 kV and 20 mA should be used. In the 2D mode, myocardial recovery coefficient (RC, image counts/true radioactivity concentration) was largely unaffected by body size and increasing liver activity (placed inside or outside the FOV). However, apparent increases in RC were observed in the 3D mode. This was indicative of higher residual scatter counts from measurement of the LV chamber/myocardium ratio. The 20 cm phantom gave a similar picture. Conclusion: Accurate attenuation correction in chest scanning on the Discovery RX can be obtained with low dose CT but there is some variation at the lowest doses. Scatter correction in the 2D mode is accurate for a range of body sizes and radioactivity distributions but in 3D the correction is less accurate for large body sizes. FBP and OSEM reconstructions gave broadly similar results.