4D FDG-PET quantification in thoracic anatomical structures for anthropomorphic phantom measurements

A mandatory prerequisite for integration of PET image quantification in clinical applications is to evaluate the ability to accurately determine the volume and activity of the target structures. With this purpose, anthropomorphic thorax phantoms with compartments that can be independently filled have been recently developed. However, activity distributions generally employed in such studies are limited by being an average over different clinical scenarios and by being based on 3D PET image quantification, in which the compensation of motion effects is not involved. An additional prerequisite is an standardized procedure for image quantification. Common standard guidelines for activity and volume estimation has been provided, but to our knowledge there exists no evaluation of its robustness for ungated and retrospectively gated PET/CT images. In this work activity concentrations for several structures, required for experimental phantom measurements, are derived from retrospectively gated PET/CT quantification of 10 lung cancer patients. Volume definition for activity estimation is performed following the EANM procedure guidelines recommendations (mC50 and mC70) and their compatibility for ungated and retrospectively gated PET/CT images is evaluated. From our results, mC50 quantification is concluded to represent a good parameter to standardize ungated and gated PET activity quantification: with an average quotient between ungated and gated activity concentrations of 0.9 ± 0.3 (0.89 ± 0.11) and compatibility in 8 (9) patients in tumor (heart). From the retrospectively gated PET quantification results for different structures in lung cancer patients, it can be concluded that in order to cover the different clinical scenarios, it is recommended to employ activity distributions with different highest activity structures (liver, heart and tumor). Activity concentrations reported in this work could be employed as reference values.