Impact of the method for calculating voxel S-values on 3D dose distributions in radionuclide therapy

AIM-Voxel dosimetry by MIRD formalism requires extensive calculations of voxel S-values (VSVs), due to different geometries, reconstruction matrices and zoom factors. Alternatively to Monte Carlo (MC) simulations, VSVs can be calculated by Dose Point Kernel (DPK) convolution, and analytical models (AMs). 3D-dosimetry can be also performed with Local Energy Deposition (LED), or a rescaled LED method recently proposed. The aim of this study is to investigate the influence of calculation methods on the 3D-dosimetry. MATERIALS & METHODS-VSVs were calculated for soft tissue by DPK convolution using published data for water (Cross, AECL–Report 1992; Furhang et al, Med Phys 1996), and by analytically modelling the deposited energy as a function of the distance (Amato et al, Med Phys 2012). Dosimetry was also performed with LED, or with a novel method (NM) proposed by Traino et al (Med Phys 2013), based on rescaled OLINDA/EXM self- irradiation S-factors for the sphere model. Soft tissue spheroidal clusters with uniform activity distributions and various masses were simulated by the software CALDOSE (Pacilio et al, Med Phys 2009). Moreover, 10 treatments with 90Y derivatives (voxel size: 4.42 mm) were considered for comparisons in clinical settings. An IDL-based software (Torres Aroche et al, ALASBIMN Journal 2011) was used for 3D-dosimetry, and VSVs from the website www.medphys.it were assumed as reference. Comparisons were performed for: 1) doses associated to 95%, or 50% of the volume (D95%, or D50%) in cumulative DVHs, and 2) dose profiles. For clinical cases, volumes of interest (VOIs) were defined by isocount levels of 10%, 30% and 50% of the maximum count. RESULTS-For system models, the DPK method evidenced D95%, and D50% differences in the range (-1.5%/0%), for 131I, 188Re, and 90Y (always negative). Differences for the AM ranged between -3.4% and 4.4%. Clinical dosimetry yielded D95% and D50% differences down to -1.7% for DPK, -3.5% for AM, between -3.0% and 2.7% for NM and up to 2.7% for LED. Differences of dose profiles with reference data (for doses>10Gy) were systematic for DPK (about -1.7%) and AM (-3.5%), or between -20% to 10% for LED and NM. LED and NM dose profiles were identical in some cases, or rescaled by a constant value in others. CONCLUSION-Dosimetric estimates was scarcely affected by using either the DPK or the AM method. For 90Y treatments, LED and NM yielded very similar results, with higher differences with respect to reference data.