Generation of clinical ¹⁷⁷Lu SPECT/CT images for dosimetry in Molecular Radiotherapy based on Monte Carlo simulation with GATE

Introduction: Patient-specific dosimetry in molecular radiotherapy relies on quantitative imaging, pharmacokinetic assessment and absorbed dose calculation. The DosiTest project aims to evaluate the uncertainties associated with each step of the clinical dosimetry protocol, via a virtual multicentric intercomparison based on Monte Carlo modeling. The first phase of the project considered only planar imaging. This work presents the generation of ‘virtual’ clinical SPECT-CT datasets based on GATE Monte Carlo modelling, which are subsequently integrated in commercial image workstations.

Methods: This study considered a therapy cycle of 6.85 GBq ¹⁷⁷Lu-labelled DOTATATE derived from an IAEA-CRP project (E23005) on “Dosimetry in Radiopharmaceutical therapy for personalised patient treatment”. Patient images were obtained from a GE Infinia gamma camera using a MEGP collimator. Simulated SPECT projections (matrix size of 128 x 128, 15 seconds per projection and energy window at 208 keV ± 20%) were generated based on experimental time points at 1h, 4h, 24h, 48h and 96h. Simulated projections were validated against clinical SPECT projections by comparing flattened profiles (on x-axis by summing all the counts on y-axis) and using the gamma index. Subsequently, these simulated projections were incorporated in the patient SPECT-CT DICOM envelopes for its integration into a commercial image workstation. Images were reconstructed using the OSEM reconstruction algorithm on Hermes (with CT-based attenuation correction and manufacturer-based Monte Carlo scatter correction) and Xeleris (with CT derived attenuation map for attenuation correction and TEW scatter correction).

Results: Gamma index passing rate between 95 - 97% and average gamma of 0.26- 0.30 was obtained among different time points, which revealed high similarity between simulated and experimental SPECT images. Image reconstruction of the simulated projections using standard approaches was successful on a HERMES and Xeleris workstation, making it possible to initiate the multicentric virtual intercomparison with simulated SPECT images.

Conclusions: We simulated realistic clinical ¹⁷⁷Lu SPECT-CT acquisitions with GATE. Modelled datasets will be circulated to different clinical departments to benchmark their clinical dosimetry procedure on a common reference dataset with the goal to assess uncertainties along the entire dosimetric chain.