Comprehensive SPECT/CT system characterization and calibration for 177Lu quantitative SPECT (QSPECT) with dead-time correction

Abstract

Background: Personalization of 177 Lu-based radionuclide therapy requires implementation of dosimetry methods that are both accurate and practical enough for routine clinical use. Quantitative single-photon emission computed tomography/ computed tomography (QSPECT/CT) is the preferred scanning modality to achieve this and necessitates characterizing the response of the camera, and calibrating it, over the full range of therapeutic activities and system capacity. Various methods to determine the camera calibration factor (CF) and the deadtime constant (τ) were investigated, with the aim to design a simple and robust protocol for quantitative 177 Lu imaging. Methods: The SPECT/CT camera was equipped with a medium energy collimator. Multiple phantoms were used to reproduce various attenuation conditions: rod sources in air or water-equivalent media, as well as a Jaszczak phantom with inserts. Planar and tomographic images of a wide range of activities were acquired, with multiple energy windows for scatter correction (double or triple energy window technique) as well as count rate monitoring over a large spectrum of energy. Dead time was modelled using the paralysable model. CF and τ were deduced by curve fitting either separately in two steps (CF determined first using a subset of low-activity acquisitions, then τ determined using the full range of activity) or at once (both CF and τ determined using the full range of activity). Total or segmented activity in the SPECT field of view was computed. Finally, these methods were compared in terms of accuracy to recover the known activity, in particular when planar-derived parameters were applied to the SPECT data. Results: The SPECT camera was shown to operate as expected on a finite count rate range (up to~350 kcps over the entire energy spectrum). CF and τ from planar (sources in air) and SPECT segmented Jaszczak data yielded a very good agreement (CF < 1% and τ < 3%). Determining CF and τ from a single curve fit made deadtime-corrected images less prone to overestimating recovered activity. Using tripleenergy window scatter correction while acquiring one or more additional energy window(s) to enable wide-spectrum count rate monitoring (i.e. ranging 55-250 or 18-680 keV) yielded the most consistent results across the various geometries. The final, planar-derived calibration parameters for our system were a CF of 9.36 ± 0.01 cps/MBq and a τ of 0.550 ± 0.003 μs. Using the latter, the activity in a Jaszczak phantom could be quantified by QSPECT with an accuracy of 0.02 ± 1.10%.

Influence of sampling schedules on [177Lu]Lu-PSMA dosimetry

“…Assuming a minimal kidney activity at 192 h p.i. of 3.1 MBq, a sensitivity of 9.4 cps/MBq [34], a field-of-view in z-direction of 38.7 cm and a bed speed of 10 cm/min, a maximal Poisson noise of 1.2% can be expected for the measured counts within the kidney ROIs for planar imaging. Thus, increases in Poisson noise over time could be neglected for the kidneys.…”

Impact of dead time on quantitative 177Lu-SPECT (QSPECT) and kidney dosimetry during PRRT

“…As others have found, the impact of dead time is limited when IA does not exceed the widely adopted empiric IA of 7.4 GBq [11,12]. Nevertheless, simple dead-time correction methods can be easily implemented to maximize the accuracy of QSPECT and remove a layer of uncertainty which would otherwise sum up with many others when performing internal dosimetry [8,[12][13][14].…”

“…177 Lu-QSPECT/CT was acquired and reconstructed as previously described using a Symbia T6 system (Siemens Healthineers, Erlangen, Germany), with the only modulated parameter being the time per projection (15 s for acquisitions the same day and the day following the injection, and 20 s for following days) [8]. Since photons of any energy can cause dead time, the dead-time correction factor was deduced from the average acquisition wide-spectrum (18-680 keV) counting rate using a lookup table [7,8]. The dead time corresponds to one minus the inverse of the dead-time correction factor.…”

“…The dead time corresponds to one minus the inverse of the dead-time correction factor. We used the dead-time constant and calibration factor that were recently obtained (0.55 μs and 9.4 cps/MBq, respectively [8]), which were smaller than those initially estimated (0.78 μs and 10.8 cps/MBq, respectively [7]). Of note, when multiple bed positions were acquired, we considered the dead time of that encompassing the kidney, which was typically the greatest.…”

“…The accuracy of dosimetry is dependent on that of quantitative imaging. We previously suggested, and recently updated, a practical method for 177 Lu quantitative SPECT (QSPECT) with a dead-time correction that we implemented for routine clinical dosimetry [7,8]. Our primary aim was to quantify, in a large cohort of patients undergoing personalized PRRT, the dead time-and the impact of not correcting for it-on the accuracy of 177 Lu-QSPECT and dosimetry.…”