Importance of optimization of OS-EM reconstruction for improved dose volume histogram estimates in radionuclide therapy dosimetry

580 Objectives Dose-volume histograms are useful for computing tumor control and normal tissue complication probabilities. Dose distribution estimates are based on activity distribution estimates obtained from emission computed tomography(ECT) imaging, which are degraded by noise and limited spatial resolution. The goal of this work was to investigate the effect of these degrading factors and SPECT reconstruction parameters on the reliability of dose-volume histograms estimates. Methods We used the 3D NCAT phantom to model patient anatomy and organ activities based on In-111 Zevalin and In111-DTPA-octreotide patient scans. Projection data at 5 time points were generated using Monte Carlo simulation. Images were reconstructed using OS-EM with model-based compensation. Up to 150 iterations and Gaussian post-reconstruction filtering with 14 different FWHMs were used. Dose-rate volume histograms were computed and compared to the ones obtained from the true activity distribution.The mean squared error (MSE) over 50 Poisson noise realizations averaged over histogram bins served as a figure of merit. Results The parameters required to obtain the optimal histogram estimates were organ and time specific. For large organs like the liver, noise was the major limiting factor: fewer iterations and moderate FWHMs gave the minimum MSE. For small organs like kidney, both noise and partial volume effect (PVE) degraded the histograms,thus requiring larger numbers of iterations and smaller FWHMs. These results were quite different than previous results for organ total activity estimation, where 50 or more iterations and no filtering were optimal. Conclusions Careful attention to the number of iterations and post-reconstruction filtering is important in order to obtain high quality dose-volume histograms, but optimal parameters are noise-level and object dependent. Development of more advanced techniques for noise regularization and PVE compensation is important for 3D dosimetry