Partial volume corrected image derived input functions for dynamic PET brain studies: Methodology and validation for [11C]flumazenil

Abstract

Extraction of arterial input functions from dynamic brain scans may obviate the need for arterial sampling and would increase the clinical applicability of quantitative PET studies. The aim of the present study was to evaluate applicability and accuracy of image derived input functions (IDIFs) following reconstruction based partial volume correction (PVC).Settings for the PVC ordered subset expectation maximization (PVC-OSEM) reconstruction algorithm were varied. In addition, different methods for defining arterial regions of interest (ROI) in order to extract IDIFs were evaluated. [11C]flumazenil data of 10 subjects were used in the present study. Results obtained with IDIFs were compared with those using standard on-line measured arterial input functions. These included areas under the curve (AUC) for peak (1–2 min) and tail (2–60 min), volume of distribution (VT) obtained using Logan analysis, and VT and K1 obtained with a basis function implementation of a single tissue compartment model.Best results were obtained with PVC-OSEM using 4 iterations and 16 subsets. Based on 11C point source measurements, a 4.5 mm FWHM (full width at half maximum) resolution kernel was used to correct for partial volume effects. A ROI consisting of the four hottest pixels per plane (over the carotid arteries) was the best method to extract IDIFs. Excellent peak AUC ratios (0.99±0.09) between IDIF and blood sampler input function (BSIF) were found. Furthermore, extracted IDIFs provided VT and K1 values that were very similar to those obtained using BSIFs.The proposed method appears to be suitable for analysing [11C]flumazenil data without the need for online arterial sampling.