Dosimetry of regressing tumors in 131I internal emitter therapy using patient data from multiple, integrated CT-SPECT images1

Accurate tumor dosimetry in internal emitter therapy requires modeling both the spatial and temporal variation of absorbed dose rates in tumor volumes. Generally, 3D absorbed dose distributions can be computed only under the approximation that the time dependence of tumor activity is spatially uniform, since 3D, time-varying patient measurement data is not available. However, in a pilot study at our clinic involving follicular lymphoma patients being treated with 131I tositumomab, registered SPECT and CT images are acquired with an integrated scanner at multiple times after both tracer and therapy administration, thus providing the extensive data required for detailed absorbed dose computations. In a previous work we described a method for the Monte Carlo computation of 3D absorbed dose distributions with spatially varying time-activity distributions in tumors. Multiple time point CT images were registered to a reference CT image which was used to define a fixed patient geometry, and voxel-based time-activity curves were derived from the registered SPECT images and numerically integrated to yield a 3D integrated activity map. Because this method relies on a single CT image to define the patient, it is not applicable for regressing, proliferating, or deforming tumor volumes. In the current work, we present results using a new method for computing absorbed dose that accounts for tumor deformation. Absorbed dose rate maps are calculated via Monte Carlo at each time point using the registered SPECT images to describe the activity distribution and the CT images to define the patient volume. These time-dependent 3D absorbed dose rate maps are then registered using transformation variables determined by a mutual information registration computation applied to the CT images. Time-integrated absorbed dose distributions are then computed by modeling the time dependence of dose rate between time steps and the tumor volume in the presence of deformation.