Uncertainty and sensitivity analysis in nuclear medicine imaging and dosimetry

Objectives The main objective of this study is to quantify the absorbed dose uncertainty and identify the most influential factors in nuclear medicine towards improving the hardware and software system for the reduction of dose uncertainty to the patients. The uncertainty and sensitivity analysis is applied to quantitatively evaluate the overall uncertainties in the processes of molecular imaging acquisitions and image-based compartmental model setup, as well as medical radiation internal dose calculation.

Methods The main uncertainty of organ absorbed dose to patients is attributed to three parts. The first part is the physical measurement uncertainty in imaging acquisition, which is propagated to the second part of uncertainty of radioactivity in source organs in compartmental modelling. The third part of uncertainty is assigned to the specific absorbed fractions (SAFs) derived from computational voxel phantoms. Uncertainty propagation law is applied for calculating the overall uncertainty of the radiation internal dose to the patients. Moreover, a variance-based sensitivity analysis is applied to identify the most influential factor in the imaging acquisition process and the important parameter in dose calculation procedure. The developed method was preliminarily applied for calculating uncertainty of organ absorbed doses for radiopharmaceuticals of ¹⁸F-choline and ¹⁸F-FDG.

Results The uncertainty factor (UF, defined as the square root of ratio between 97.5^th and 2.5^th percentiles) for organ absorbed dose is in the range of 1.1 and 1.2 for lymph, and 2.1 and 1.7 for urinary bladder wall (UB-wall) for ¹⁸F-choline and ¹⁸F-FDG, respectively. The uncertainty contribution to absorbed dose varies in different target organs; however, for most target organs, the main uncertainty originates from SAFs with blood or other soft tissues as source organs. For ¹⁸F-choline, SAFs contribute about 96% for lung absorbed dose; whereas for absorbed dose in UB-wall, the activity integrated in UB-content and SAF for UB-content to UB-wall contribute 16% and 84%, respectively.

Conclusions The identification of the influential factors in the whole process from image acquisition in clinic practice to internal dose calculation may improve patient-specific dose assessment. Furthermore, the uncertainty information may help in choosing advanced technology that causes the lowest patient exposure with comparable diagnostic quality.