Compared with conventional, whole-organ, S-factor-based dosimetry, 3-dimensional (3D), patient-specific dosimetry better accounts for radionuclide distribution and anatomic patient variability. Its accuracy, however, is limited by the quality of the cumulated activity information that is provided as input. This input has typically been obtained from SPECT and planar imaging studies. The objective was to implement and evaluate PET-based, patient-specific, 3D dosimetry for thyroid cancer patients.
Methods: Three to 4 PET imaging studies were obtained over a 7-d period in 15 patients with metastatic thyroid carcinoma after administration of (124)I-NaI. Subsequently, patients were treated with (131)I on the basis of established clinical parameters. Retrospective dosimetry was performed using registered (124)I PET images that were corrected for the half-life difference between (124)I and (131)I. A voxel-by-voxel integration, over time, of the resulting (131)I-equivalent PET-derived images was performed to provide a single 3D dataset representing the spatial distribution of cumulated activity values for each patient. Image manipulation and registration were performed using Multiple Image Analysis Utility (MIAU), a software package developed previously. The software package, 3D-Internal Dosimetry (3D-ID), was used to obtain absorbed dose maps from the cumulated activity image sets.
Results: Spatial distributions of absorbed dose, isodose contours, dose-volume histograms (DVHs), and mean absorbed dose estimates were obtained for a total of 56 tumors. Mean absorbed dose values for individual tumors ranged from 1.2 to 540 Gy. The absorbed dose distribution within individual tumors was widely distributed ranging from a minimum of 0.3 to a maximum of 4,000 Gy.
Conclusion: (124)I PET-based, patient-specific 3D dosimetry is feasible, and sequential PET can be used to obtain cumulated activity images for 3D dosimetry.