International intercomparison of absolute quantification in nuclear medicine

55 Objectives Absolute quantification of activity in nuclear medicine images is essential for accurate dosimetry of targeted radionuclide therapies. The ability to perform absolute quantification at multiple sites with different equipment has not been previously studied. The goal of this IAEA coordinated project was to test the ability to accurately estimate activity using SPECT and planar imaging at sites with a variety of experience and equipment. Methods Participants were from 11 countries on 5 continents. Four epoxy-filled rod sources (3.8cm long, diameters between 0.794 to 2.86 cm) containing Ba-133 with activities traceable to US NIST not disclosed to the sites were shipped to participants. The sources were to be placed in water-filled cylindrical phantoms. Two trials were performed, each with 3 SPECT(/CT) and planar acquisitions. In the first trial, training took the form of a draft of an IAEA technical document on quantification and a meeting of all participants. The protocol only specified the acquisition durations and the calibration procedure. After the first trial a second meeting was held to standardize the protocol, taking into account the variety of equipment and image analysis tools available. Results Activities in both trials were underestimated by 15-20% using planar imaging, despite the use of improved attenuation and scatter compensation methods in the second trial. The coefficient of variation was 32% for the smallest source and 12-20% for the other sources. For SPECT, in both trials the activity was underestimated by 5-10%, but the first trial showed substantially larger variability in activity estimates. Conclusions Absolute quantification of simple objects is possible in a multi-center setting with errors of 15-20% with planar imaging and better than 10% for SPECT imaging. Experience from the project suggests that careful training and protocol definition and validation is needed to achieve these levels of absolute accuracy.