Quantitative image validation of PET-MR against stand-alone PET - A dynamic 11C-PE2I PET study.

Objectives For PET-MR, challenges incorporating bone in attenuation maps remain, which may affect quantitative accuracy of PET data. The stand-alone PET ECAT HR+ scanner, using rotating ⁶⁸Ge rod sources for attenuation correction, is often considered the gold standard for quantitative brain PET imaging. The aim of the present work was to compare quantitative measures of dopamine transporter availability and relative cerebral blood flow based on dynamic ¹¹C-PE2I-PET data acquired on a time-of-flight integrated PET-MR and a stand-alone PET scanner.

Methods Five patients evaluated for Parkinsonism underwent two 80 min dynamic PET scans using the specific dopamine transporter radioligand ¹¹C-PE2I (about 5 MBq/kg ). First they were investigated on a stand-alone PET scanner (ECAT HR+) and within 6 months they underwent a second investigation on a time-of-flight integrated PET-MR scanner (Signa PET-MR). ECAT images were reconstructed using OSEM (6 iterations, 8 subsets, 4 mm Hanning post-filter). PET-MR images were reconstructed using a resolution-matched protocol (OSEM; 2 iterations, 28 subsets, 5 mm Gaussian filter). PET-MR attenuation correction was based on a head atlas method. Both dynamic PET series were co-registered to T1-MRI images using SPM8. Volumes of interest were defined using a probabilistic template (PVElab) on the T1- MRI images and transferred to both sets of PET images. Binding potential (BP_ND), reflecting dopamine transporter availability, and relative cerebral blood flow (R₁) were derived from the PET images for a number of clinically relevant tissues using the simplified reference tissue model, with cerebellar grey matter as reference tissue. BP_ND in putamen, and R₁ in striatum as well as anterior cortical, posterior cortical, and limbic regions, were compared between both modalities using regression and Bland-Altman analysis.

Results PET-MR BP_ND in putamen agreed well with stand-alone PET BP_ND (R² 0.97), with a slope of 0.84 (CI 0.55-1.13) and intercept of 0.40 (CI -0.54 - 1.34), and no significant bias. PET-MR R₁ correlated best with stand-alone PET in striatum and limbic regions (R² 0.94 and 0.89, respectively), with lower values in anterior and posterior cortex (R² 0.76 and 0.50). A significant positive bias of PET-MR R₁ values was found in posterior cortex (bias 0.13, CI 0.03-0.24). Correlation plots are shown in Figure 1.

Conclusions This study showed a good agreement between the PET-MR and stand-alone PET for BP_ND in putamen and for R₁ especially in striatal and limbic regions. Cortical regions presented relatively low correlations between both modalities,which might be explained by the chosen attenuation correction method. Besides extension of the number of patients, additional investigations assessing, other attenuation correction methods for further improvement of the quantitative accuracy of the dynamic PET-MR data are required.

• Download figure
• Open in new tab
• Download powerpoint