Shortened whole body dynamic PET 18F-FDG Patlak imaging using a population input function and the Biograph Vision Quadra PET/CT

Introduction: Non-invasive quantification of ¹⁸F-FDG kinetics throughout the body is possible using a Siemens Biograph Vision Quadra PET/CT system. Its 106 cm long axial coverage captures the heart and all other organs of interest, including possible tumor lesions, within a single bed position. The excellent performance characteristics of the Vision Quadra, such as a substantial increase in sensitivity, allows for precise measurements of image derived input functions (IDIF) and tissue time activity curves. Previously we have proposed a method for a reduced 30 min (as opposed to 60 min) whole body ¹⁸F-FDG Patlak PET imaging procedure using an external population averaged input function (PIF) scaled to IDIF values at 30-60 min post-injection (pi) (van Sluis et al. EJNMMI Phys. (2020)). The aim of the present study was to validate Patlak analysis of whole body ¹⁸F-FDG images using the Vision Quadra, including the use of a PIF for reduced scan duration.

Methods: Lung cancer patients, referred for clinical purposes, were included and received a weight-based (3 MBq/kg) injection of ¹⁸F-FDG. Each patient underwent a 65 min dynamic PET acquisition. Dynamic PET images were reconstructed using European Association of Nuclear Medicine Research Ltd. (EARL) standards 2 reconstruction settings to obtain data that complied with European guidelines for multicenter PET image quantification and harmonization (3D OP-OSEM with 4 iterations, 5 subsets, a matrix size of 220x220 with a voxel size of 3.3x3.3x1.5 mm, time of flight, resolution modelling, and a 5 mm Gaussian filter). For each dynamic PET dataset, volumes of interest (VOI) were placed in the ascending aorta (AA) and in the left ventricle of the heart (LV) to derive IDIFs. Subsequently, a previously published PIF (Cheebsumon et al. Eur J Nucl Med Mol Imaging. (2011)) was scaled to AA and LV IDIF values at 30-60 min pi, resulting in four input functions (IF) per dataset: IDIF AA, IDIF LV, PIF AA, and PIF LV. For each IF, parametric ¹⁸F-FDG influx rate (K_i) images were generated using only the 30 to 60 min pi uptake data. In these K_i images, tumor lesions and, as reference, cerebral grey matter were segmented using a semi-automated segmentation method (i.e., 50% of SUVpeak isocontour). In addition, a 3 cm diameter spherical VOI was placed in the liver and 2 cm diameter spherical VOIs were placed in the spleen and muscle tissue.

Agreement between K_i obtained using the IDIF and corresponding PIF derived from both the AA and LV was assessed using Bland-Altman plots and correlation coefficients (R²). Moreover, R² was obtained to evaluate variation in K_i obtained by IDIF VOI positioning (i.e., to compare AA and LV IDIF).

Results: So far, six lung cancer patients were included with a median weight of 74 kg and a median injected dose of 240 MBq ¹⁸F-FDG. Bland-Altman plots to assess the agreement between K_i obtained using IDIF and PIF showed lower and upper limits of agreement (LOA) of -0.002 and 0.001 for the AA, and -0.004 and 0.001 for the LV, respectively. In both cases, 31 out of 33 datapoints (in total nine tumor lesions and four healthy tissues per patient) were located between those LOAs. Tumor lesion and healthy tissue K_i R² obtained from the IDIF versus PIF comparison for the AA were 0.997 and 0.998, respectively. For the LV, tumor lesion and healthy tissue K_i R² were 0.997 and 0.987, respectively. Scatter plots in Figure 1 show tumor lesion and healthy tissue K_i obtained from the IDIF versus PIF comparison for the AA and LV. With regard to positioning the IDIF VOI in AA or LV, R² for all (both tumor lesion and healthy tissue) K_i was 0.980.

Conclusions: From these first results, it appears to be feasible to generate parametric ¹⁸F-FDG Patlak K_i images non-invasively using a Vision Quadra system. In addition, using a PIF allows for a substantial (factor 2) reduction in scan time (i.e., from 30 – 60 min pi) without loss of accuracy.

• Download figure
• Open in new tab
• Download powerpoint