The UF/MSK adult mesh-based computational human phantom library: applications to organ dosimetry in computed tomography.

Abstract

<i>Objective.</i>To develop a comprehensive mesh-based computational human phantom library of adult males and females that span current US body-size distributions. The phantom library was developed jointly between the University of Florida and Memorial Sloan-Kettering Cancer Center and termed the UF/MSK Phantom Library.<i>Approach.</i>Adult phantoms were modeled by rescaling and systematically deforming the International Commission on Radiological Protection Publication 145 mesh-type reference computational phantoms to match heights, body mass, and CDC National Health and Nutrition Examination Survey-derived secondary anthropometric parameters. Low body-mass phantoms underwent uniform two-dimensional scaling of abdominal organs to avoid collision with subcutaneous fat. To accommodate larger body circumferences, a rigging procedure was developed to smoothly deform the limbs while maintaining anatomical realism and preserving tissue. A companion 'arms-up' version of the library appropriate for computed tomography (CT) dosimetry simulation was also constructed. A validated CT source term was developed for the Canon Aquilion ONE/GENESIS considering technique factors such as tube voltage, bowtie filter size, and beam collimation, and was used to compute slice-specific organ doses for a subset of library phantoms.<i>Main results.</i>A large library of 186 male and 171 female (357 total) phantoms was created. Standing height ranged from 150-200 cm for males and 150-195 cm for females, and total body mass ranged from 40-150 kg for males and 40-155 kg for females. Simulated CT organ doses varied smoothly with BMI for a fixed phantom height.<i>Significance.</i>The UF/MSK adult phantom library has been constructed with body morphometries representative of the US population. Organ doses computed for phantoms of varying body sizes demonstrate the importance having a morphometrically robust selection of computational phantoms for patient matching. While the immediate application of this library is to expand phantom selection options in MIRDct software (www.mirdsoft.org), the library can be further utilized in other applications including radiopharmaceutical, radiography, fluoroscopy, and radiotherapy organ dose reconstructions.