In recent years, Peptide Receptor Radionuclide Therapy (PRRT) with 177Lu-labelled somatostatine analogous has yield promising results in the treatment of neuroendocrine tumours. Despite observed progresses, PRRT has not been systematically optimized as the choice of the treatment schedule remains empirical. According to the studies conducted at IRST-IRCCS on patients with advanced, well-differentiated gastrointestinal neuroendocrine tumors (GI-NETs), a cumulative activity of 500 mCi (18.5 GBq) in five cycles could be considered the minimum effective dosage in GI-NETs, also for patients with a higher probability of developing kidney or bone marrow toxicity. This is important, as the possibility of sparing normal tissue from unnecessary radiation exposure is a must in radiation therapy. At IRST-IRCCS a SPECT/CT General Electric Discovery NM/CT 670 has been installed with the aim of tailoring the PRRT treatment on the basis of dosimetric data. The aim of present work is to investigate the performance of this device for measurements performed with 177Lu and to evaluate its quantitative accuracy.
The SPECT system is calibrated in reference conditions by using dedicated phantoms and validated in non-reference conditions using an anthropomorphic torso phantom. The images are acquired by the dual-head gamma camera equipped with 3/8”-thick NaI(Tl) crystals with medium energy high resolution collimators. A 20% energy window is applied at the 208 keV photopeak energy. The following reference geometries are considered: a point source, a 16 ml sphere in the Jaszczak phantom (with and without cold water as attenuating medium) and a cylindrical phantom. The reconstruction and analysis of the SPECT images are carried out on the Xeleris 3.0 workstation by using the Dosimetry Toolkit (GE), using the ordered subsets expectation maximization (3D OSEM) algorithm for image reconstruction. Corrections for attenuation, scatter fraction (triple energy window) and recovery resolution are included. The attenuation map is created with 16 slices CT measurement (120 kVp; 40-80 mA). The behavior of 3D OSEM reconstruction is studied for the point source and sphere geometries as a function of the number of iterations and subsets. For each geometry a calibration factor is estimated and validated on the anthropomorphic torso phantom.
The convergence of the 3D OSEM strongly depends on the volume of interest and different behaviors are observed for point source and sphere geometries. The accuracy of the activity quantification in the anthropomorphic phantom varies as a function of the calibration factors and improves when the calibration factor estimated for the sphere is used.
Phantom experiments and characterization of the 3D OSEM algorithm performance are mandatory for accurate personalized dosimetry for PRRT.