A detector block-pairwise dead time correction method for improved quantitation with a dedicated BrainPET scanner

Abstract

Objective: Dead time correction (DTC) is an important factor in ensuring accurate quantification in PET measurements. This is currently often achieved using a global DTC method, i.e., an average DTC factor is computed. For PET scanners designed to image dedicated organs, e.g., those used in brain imaging or positron emission mammography (PEM), a substantial amount of the administered radioactivity is located outside of the PET field-of-view (FOV). This activity contributes to the dead time (DT) of the scintillation detectors. Moreover, the count rates of the individual scintillation detectors are potentially very inhomogeneous due to the specific irradiation of each detector, especially for combined MR/PET systems, where radiation shields cannot be applied. Approach: We have developed a block-pairwise DTC method for our Siemens 3T MR BrainPET insert by extending a previously published method that uses the delayed random coincidence count rate to estimate the DT in the individual scans and planes (i.e., scintillation pixel rings). The method was validated in decay experiments using phantoms with a homogenous activity concentration and with and without out-of-FOV activity. Based on a three-compartment phantom, we compared the accuracy and noise properties of the block-pairwise DTC and the global DTC method. Main results: The currently used global DTC led to a substantial positive bias in regions with high activity; the block-pairwise DTC resulted in substantially less bias. The noise level for the block-pairwise DTC was comparable to the global DTC and image reconstructions without any DTC. Finally, we tested the block-pairwise DTC with a data set obtained from volunteer measurements using the mGluR5 (metabotropic glutamate receptor subtype 5) antagonist [11C]ABP688. When the relative differences in activity concentrations obtained with global DTC and block-pairwise DTC for the ACC and the cerebellum GM were compared, the ratios differed by a factor of up to 1.4 at the beginning – when the first injection is administered as a bolus with high radioactivity. Significance: In this work, global DTC was shown to have the potential to introduce quantification bias, while better quantitation accuracy was achieved with the presented block-pairwise DTC method. The method can be implemented in all systems that use the delayed window technique and is particulary expected to improve the quantiation accuracy of dedicated brain PET scanners due to their geometry.

Impact of improved dead time correction on the quantification accuracy of a dedicated BrainPET scanner

“…Therefore, the homogeneity of the single rates in the detector blocks is affected, causing variations in the DT depending on the physical position of each detector block. According to earlier results [15], the blockpairwise DTC method is particularly beneficial in this respect and has shown consistent results for brain-sized phantoms inserted in the BrainPET, leading to improved quantification accuracy. This is due to the fact that it is based on the estimation of the DTC factor as a nonlinear function derived from the random coincidence rate in individual detector rings [14,15].…”

“…A more accurate DTC method decreases quantification biases in image reconstructions and results in more accurate mean counts [11][12][13]. Based on the method developed by [14], we recently developed an improved block-pairwise DTC method for the BrainPET insert of a 3T MRI scanner [15]. In this previous study, phantom measurements were used to quantitatively evaluate the accuracy and precision of the block-pairwise DTC method with respect to the differences between the reconstructed activity concentrations and true activity concentrations in the phantoms and phantom compartments.…”

“…The global DTC method is dependent on the overall count rate, which is obtained by averaging over all individual scintillation detector blocks in the PET system, irrespective of the physical position of each block. Consequently, depending on the distribution of the activity inside and outside the FOV, quantification accuracy may be impaired since the global DTC applies the same correction factor to all detector blocks, although they receive different count rates [15]. In contrast, as the block-pairwise DTC method can obtain activity-independent and location-independent quantitation in phantom measurements, the mean of DT losses is estimated correctly [14,15].…”

“…Consequently, depending on the distribution of the activity inside and outside the FOV, quantification accuracy may be impaired since the global DTC applies the same correction factor to all detector blocks, although they receive different count rates [15]. In contrast, as the block-pairwise DTC method can obtain activity-independent and location-independent quantitation in phantom measurements, the mean of DT losses is estimated correctly [14,15]. Furthermore, as previously shown [15] also analyzed the propagation of both the DTC methods into image noise, and the results showed that it was only slightly higher in the case of the block-pairwise DTC method.…”

“…In contrast, as the block-pairwise DTC method can obtain activity-independent and location-independent quantitation in phantom measurements, the mean of DT losses is estimated correctly [14,15]. Furthermore, as previously shown [15] also analyzed the propagation of both the DTC methods into image noise, and the results showed that it was only slightly higher in the case of the block-pairwise DTC method. The dedicated BrainPET scanner used in this work is the Siemens 3T MR-BrainPET insert [16][17][18][19].…”