[ePoster] Accuracy Assessment of Absorbed Dose Rate Calculation: Evaluation of Two Treatment Planning Systems for Molecular Radiotherapy

Accuracy assessment of absorbed dose rate calculation: evaluation of two treatment planning systems for molecular radiotherapy

ABSTRACT
Aim/Introduction: Assessing the accuracy of clinical dosimetry in molecular radiotherapy is a challenging task, since the different steps that contribute to the determination of the absorbed doses (clinical dosimetry workflow – CDW) have to be considered. We initiated a study of the variability of each CDW step on two software, by analyzing the impact of different approaches/methodologies. In this work, we first present results obtained for the absorbed dose computation step, by comparing local energy deposition (LED) and dose voxels kernel (DVK) convolution, with or without media density correction. Monte Carlo radiation transport modelling was set as the reference.

Materials and Methods: Clinical dosimetry was performed on a group of patients who received Lutathera^® treatment at the Institut Régional du Cancer de Montpellier (ICM). The CDW was implemented using PLANET^® Dose (DOSIsoft SA) software. It included image registration, segmentation, absorbed dose rates (ADR) computation, and integration over time to obtain the absorbed doses. The overall variability of the CDW was compared with OpenDose3D software. Monte Carlo simulations were conducted using GATE version 9.1. 

Results: By looking first at the absorbed dose computation step, the initial differences on ADR between software were in the range of 4% to 11% (for kidneys and liver) depending on the algorithm used and media density management. Further studies and software comparison put in evidence differences in density correction implementation. By using a similar Hounsfield Unit-densities calibration function, the observed differences were reduced. For example, the difference for LED with media density correction decreased from 4% to 1%. The final comparison of convolution vs. direct Monte Carlo simulations shown a very good agreement (around 2% of difference at maximum). We are now studying the impact of registration and VOI definition across time. For lesions, a 20% difference in volumes was obtained, inducing discrepancies up to 30% on the final absorbed dose between both software. This is being further studied, as well as integration over time.

Conclusion: This work assesses the accuracy and validates the absorbed dose computation approaches implemented in the 2 software for 177Lu-based radiopharmaceutical therapies. It is gradually extended: first, to other steps of the CDW, then on other isotopes (e.g. ¹³¹I). In perspective, the study of the impact of registration and VOI definition across time, is currently ongoing. In that context, the availability of an open-source software, freely available, is an invaluable asset when benchmarking clinical dosimetry software.