Towards the standardization of SIRT dosimetry: A method to evaluate protocols.

Objectives To propose a method to evaluate SIRT dosimetry protocols for clinical centers and multi-center trials.

Methods A Jaszczak phantom with 6 spheres (1 to 113 mL) was filled with 195 MBq of ^99mTc, a contrast of 10.1 and imaged on a Symbia T2 (Siemens) camera (128x128, 30 s, 60 proj.). The phantom was simulated in MCNP to obtain reference absorbed fractions (φ_r). φ_r values were used to compute Average Absorbed Doses of reference (AADr) for any injected activity and contrast. The protocol evaluation consists in comparing AADr to AAD of any protocol. The evaluation process was tested on 4 protocols. Two reconstructed datasets were performed using OSEM3D with attenuation and scatter corrections. R₁: 15 subs., 1 it., 8 mm Gaussian post-filtering (PF); R2: 15 subs., 10 it. no PF. For P1 to 3, ROI were defined by manual threshold. P1 and P2 used R1 and R2 respectively and φ=1. P3 was P2 including partial volume effect correction (PVEC) using recovery coefficients and φ from Siegel et al (JNM 1994). P3* was P3 using ideal ROIs and PVEC using the geometric transfer function.

Results The AADr range in the spheres was [47.8;62.7] Gy/GBq. The errors found with P3* were between -5 and 5%, except -6.8% for the smallest sphere. P1 gave poor results [-13;-77%]. P2 gave better results: error range on 5 spheres [0;14%] and -31% for the smallest one. The best results among protocols (P1-P3) were found with the P3 method integrating a simple PVEC: error range on 5 spheres [-4;9%] and -37% for the smallest sphere.

Conclusions A method to evaluate SIRT dosimetry protocols has been developed and applied to 4 protocols showing its ability to be a valuable tool for commissioning dosimetry protocols.