AUTHOR=Freire Marta , Cañizares Gabriel , Echegoyen Sara , Gonzalez-Montoro Andrea , Gonzalez Antonio J. 

TITLE=Reducing Calibration Time in PET Systems Based on Monolithic Crystals

JOURNAL=Frontiers in Medicine

VOLUME=Volume 8 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2021.734476

DOI=10.3389/fmed.2021.734476

ISSN=2296-858X

ABSTRACT=In the last years, gamma-ray detector designs based on large and thick monolithic crystals have demonstrated to be excellent candidates for the design of high-performance Positron Emission Tomography (PET) systems, such as for instance long axial field of view (FOV) PET. Monolithic crystals allow to achieve intrinsic detector resolutions well below state-of-the-art in whole-body PET of 3-4 mm Full Width at Half of the Maximum (FWHM). Moreover, using these crystals allow for an increased packing fraction thus, increasing system sensitivity. In addition, monolithic crystals offer intrinsic depth of interaction (DOI) capabilities, which helps improving lesion detectability at the edges of the scanner FOV. Regarding cost, they are cheaper than the traditional pixelated scintillators for pixel sizes smaller than 1.5  1.5 mm2, as the ones used in pre-clinical PET imaging.
The bottle-neck to translate to clinical PET systems based on large number of monolithic detectors is eventually the requirement of mechanically complex and time-consuming calibration processes. To mitigate this drawback, several methods have been already proposed, such as using non-physically collimated radioactive sources or implementing Neuronal Networks (NN) algorithms trained with simulated data.
In this work, we aim to simplify and fasten the calibration process of monolithic based systems. We provide a comparison of the standard calibration procedure of individually acquiring data for all the detector modules composing the PET system, to the case where: i) the calibration map of one detector is estimated and shared among all others, and ii) the calibration map is slightly modified for each module as a function of their detector uniformity map. Results exhibit negligible reconstructed image differences (13% deviation at most) with a significant reduction in calibration time. In terms of spatial resolution, variances of the proposed methods to the conventional one are within 0.1 mm error bars. Regarding count rate, deviations as small as 0.2 % were reported. Moreover, the calculated contrast values showed almost identical performance.