Parametric imaging of [11C]flumazenil binding in the rat brain

I. Lopes Alves, D. Vállez García, A. Parente, J. Doorduin, R. Dierckx, A. M. Marques da Silva, M. Koole, A. T. M. Willemsen, R. Boellaard

Research output: Contribution to journalMeeting AbstractAcademic


This study evaluates the performance of several parametric methods for assessing [11C]flumazenil binding distribution in the rat brain.

Dynamic (60 min) PET data with metabolite corrected plasma input function was retrospectively analyzed (male Wistar rats, n = 10). Distribution volume (VT) images were generated from the basis function method (BFM), the Logan Graphical Analysis (LGA) and the Spectral Analysis (SA). Using the pons as pseudo-reference tissue, binding potential (BPND and DVR-1) images were obtained from two receptor parametric imaging algorithms (RPM and SRTM2) and Reference Logan (RLogan). Standardized Uptake Value images (SUV and SUVR) were also computed for different post-injection intervals. From these images, regional averages were extracted based on pre-defined volumes of interest (VOIs). In addition, the corresponding non-linear regression (NLR) version of each method was applied to the time-activity curves extracted of each VOI from the dynamic image. Parametric data was compared to the NLR counterparts as well as to 2TCM based values (previously defined as the model of choice for rats). Parameter agreement was assessed by linear regression analysis and Bland-Altman plots.

All parametric methods correlated strongly to their NLR counterparts (R2 > 0.8). However, RPM and SRTM2 overestimated NLR values (slope = 1.1 and slope = 1.3, respectively). Compared to 2TCM VT, all parametric methods showed excellent correlation (R2 ≥ 0.9). However, BFM and SRTM2 underestimated VT and BPND respectively (slope = 0.8), while SUVR-1 overestimated BPND (slope = 1.2). Bland-Altman plots showed LGA and RPM had the best agreement both to NLR counterparts (−0.20 and 0.03 bias) and to 2TCM values (−0.19 and −0.1 bias), while SA showed a bias of 0.57 to NLR and of −0.49 to 2TCM VT.

In conclusion, all parametric methods showed good performance, although LGA and RPM outperformed other methods. Yet, BFM and SA are of interest because they also provide K1 and model order (SA only) images.

Original languageEnglish
Article numberPS01-088
Pages (from-to)152-153
Number of pages2
JournalJournal of Cerebral Blood Flow and Metabolism
Issue number1_suppl
Publication statusPublished - 1-Apr-2017
Event28th International Symposium on Cerebral Blood Flow, Metabolism and Function / 13th International Conference on Quantification of Brain Function with PET - Berlin, Germany
Duration: 1-Apr-20174-Apr-2017

Cite this