Comparative assessment of linear least‐squares, nonlinear least‐squares, and Patlak graphical method for regional and local quantitative tracer kinetic modeling in cerebral dynamic ¹⁸F‐FDG PET

Abstract

Purpose Dynamic 18F‐FDG PET allows quantitative estimation of cerebral glucose metabolism both at the regional and local (voxel) level. Although sensitive to noise and highly computationally expensive, nonlinear least‐squares (NLS) optimization stands as the reference approach for the estimation of the kinetic model parameters. Nevertheless, faster techniques, including linear least‐squares (LLS) and Patlak graphical method, have been proposed to deal with high resolution noisy data, representing a more adaptable solution for routine clinical implementation. Former research investigating the relative performance of the available algorithms lack precise evaluation of kinetic parameter estimates under realistic acquisition conditions. Methods The present study aims at the systematic comparison of the feasibility and pertinence of kinetic modeling of dynamic cerebral 18F‐FDG PET using NLS, LLS, and Patlak method, based on numerical simulations and patient data. Numerical simulations were used to study the bias and variance of K1 and Ki parameters estimation under representative noise levels. Patient data allowed to assess the concordance between the three methods at the regional and voxel scale, and to evaluate the robustness of the estimations with respect to patient head motion. Results and Conclusions Our findings indicate that at the regional level NLS and LLS provide kinetic parameter estimates (K1 and Ki) with similar bias and variance characteristics (K1 bias ± relative standard deviation [RSD] 0.0 ± 5.1% and 0.1% ± 4.9% for NLS and LLS respectively, Ki bias ± RSD 0.1% ± 4.5% and −0.7% ± 4.4% for NLS and LLS respectively). NLS estimates appear, however, to be slightly less sensitive to patient motion. At the voxel level, provided that patient motion is negligible or corrected, LLS offers an appealing alternative solution for local K1 mapping. It yields K1 estimates that are highly correlated, with high correlation with NLS values (Pearson's r = 0.95 on actual data) within computations times less than two orders of magnitude lower. Last, Patlak method appears as the most robust and accurate technique for the estimation of Ki values at the regional and voxel scale, with or without head motion. It provides low bias/low variance Ki quantification (bias ± RSD −1.5 ± 9.5% and −4.1 ± 19.7% for Patlak and NLS respectively) as well as smooth parametric images suitable for visual assessment.

Plasma radio–metabolite analysis of PET tracers for dynamic PET imaging: TLC and autoradiography

“…In the few cases where radio-metabolites overlapped with the parent tracer because of similar polarity and hence strength of adhesion to the silica media, the parent tracer peak could be adequately resolved by the curve fitting procedure discussed in §2.6. Taken the above results together, our method of combining TLC and the Beaver proprietary beta particle detector has the analyte resolution and sensitivity for blood metabolite determination for both [18 F]FAZA and [ 18 F]FEPPA in individual large (pig) or small (rodents) animals. Nevertheless, by comparing the [ 18 F]FEPPA dose used in our pig studies (13−27 MBq/kg) to the published patient dose (2.5 − 6.2 MBq/kg (22-25)) the sensitivity of the detector has to be increased by at least 5 times for the method to be used for blood metabolite determination in individual patients.…”

“…Significant inter-subject and intra-subject variability in radiometabolites production could be observed with both tracers. For [ 18 F]FEPPA, 50% of plasma activity was from radiometabolites as early as 5 min post injection while for [ 18 F]FAZA, significant metabolites did not appear until 50 min post. Simulation study investigating the effect of radio-metabolite in the estimation of kinetic parameters indicated that 32-400% parameter error can result without radio-metabolites correction.…”

“…Without this correction, the AIF will be overestimated leading to errors in the kinetic analysis of dynamic PET. Although we used two specific tracers, [ 18 Figure A2, the percentage bias of the same parameters as in Table 3 -1 , 2 , 3 , 4 , , , and , as estimated by AIFm and AIFc, differs widely among the different scale…”

“…The separation of radio-metabolites in blood plasma with different mixtures of methanol, hexane and ethyl acetate for both tracers is shown in Figure 2. The optimized mobile phase for [ 18 Figure 3. For the reference, a well-defined peak was observed due to high signal to noise ratio.…”

Plasma radio–metabolite analysis of PET tracers for dynamic PET imaging: TLC and autoradiography

“…Without this correction, the AIF will be overestimated leading to errors in the kinetic analysis of dynamic PET. Although we used two specific tracers, [ 18 Figure A2, the percentage bias of the same parameters as in Table 3 -1 , 2 , 3 , 4 , , , and , as estimated by AIF m and AIF c , differs widely among the different scale…”