Dose calculation in nuclear medicine with magnetic resonance imaging images using Monte Carlo method.

This article focuses on a novel method for converting magnetic resonance imaging (MRI) signals into computed tomography (CT) Hounsfield unit (HU) values to enable accurate dose calculations in nuclear medicine (NM). The study introduces the "voxels spawn voxels" technique combined with the "orthonormalize" feature in Carimas software to build MRI-to-HU conversion tables for abdominal organs, including the gluteus maximus, liver, kidneys, spleen, pancreas, and colon. Dose calculations using the Geant4/Gamos Monte Carlo simulation showed good agreement (<5% difference) between converted MRI and original CT images for most organs, except the colon, which exhibited an 18% discrepancy likely due to its complex structure and contents. This approach aims to overcome limitations of existing bulk density, atlas registration, and machine learning methods by providing individualized, voxel-level conversion without requiring large datasets or assumptions of organ homogeneity.