Voxel phantoms are human models based on computed tomographic or magnetic resonance images obtained from high-resolution scans of a single individual. They consist of a huge number of volume elements (voxels) and are at the moment the most precise representation of the human anatomy. The purpose of this paper is to introduce the GSF voxel phantoms, with emphasis on the new ones and highlight their characteristics and limitations. The GSF voxel family includes at the moment two paediatric and five adult phantoms of both sexes, different ages and stature and several others are under construction. Two phantoms made of physical calibration phantoms are also available to be used for validation purposes. The GSF voxel phantoms tend to cover persons of individual anatomy and were developed to be used for numerical dosimetry of radiation transport but other applications are also possible. Examples of applications in patient dosimetry in diagnostic radiology and in nuclear medicine as well as for whole-body irradiations from idealized external exposures are given and discussed.

New organ equivalent dose conversion coefficients are presented for whole body irradiation with monoenergetic photons of energies between 10 keV and 10 MeV for idealized geometries and seven adult male and female voxel models. The geometries are broad parallel photon beams in anterior-posterior, posterior-anterior, left- and right-lateral direction and a full 360 degree rotation around the body length axis. Dose differences between the different voxel models are below approximately 30% for some organs and geometries in the energy range between 60 and 200 keV, but they can be up to 100% or more in single cases, due to differences in stature and individual anatomical details. For low photon energies, the differences may amount to hundreds of per cent. Extensive comparisons of the dose conversion coefficients with respective values calculated using mathematical body models revealed various degrees of unrealistic positioning of single organs in the latter models. Examples are the kidneys, spleen and stomach that are located too superficially in the mathematical models. Over- or underestimations of several tens of per cent may, thus, occur for the mathematical models, compared to the voxel models considered. In contrast to previous assumptions, when the mathematical models have been used to establish reference organ dose conversion coefficients, it can be concluded that they do not properly represent a large population of individuals.

Conversion Coefficients for Radiological Protection Quantities for External Radiation Exposures * If the monitoring devices are not designed to measure H 0 (3, X) or H p (3), H 0 (0.07, X) and H p (0.07) may be applied.

A new series of organ equivalent dose conversion coefficients for whole body external photon exposure is presented for a standardized couple of human voxel models, called Rex and Regina. Irradiations from broad parallel beams in antero-posterior, postero-anterior, left- and right-side lateral directions as well as from a 360 degrees rotational source have been performed numerically by the Monte Carlo transport code EGSnrc. Dose conversion coefficients from an isotropically distributed source were computed, too. The voxel models Rex and Regina originating from real patient CT data comply in body and organ dimensions with the currently valid reference values given by the International Commission on Radiological Protection (ICRP) for the average Caucasian man and woman, respectively. While the equivalent dose conversion coefficients of many organs are in quite good agreement with the reference values of ICRP Publication 74, for some organs and certain geometries the discrepancies amount to 30% or more. Differences between the sexes are of the same order with mostly higher dose conversion coefficients in the smaller female model. However, much smaller deviations from the ICRP values are observed for the resulting effective dose conversion coefficients. With the still valid definition for the effective dose (ICRP Publication 60), the greatest change appears in lateral exposures with a decrease in the new models of at most 9%. However, when the modified definition of the effective dose as suggested by an ICRP draft is applied, the largest deviation from the current reference values is obtained in postero-anterior geometry with a reduction of the effective dose conversion coefficient by at most 12%.

Backscatter factors were determined for x-ray beams relevant to diagnostic radiology using Monte Carlo methods. The phantom size considered most suitable for calibration of dosimeters is a cuboid of 30 x 30 cm2 front surface and 15 cm depth. This phantom size also provides a good approximation to adult patients. Three different media were studied: water, PMMA and ICRU tissue; the source geometry was a point source with varying field size and source-to-phantom distance. The variations of the backscatter factor with phantom medium and field geometry were examined. From the obtained data, a set of backscatter factors was selected and proposed for adoption as a standard set for the calibration of dosimeters to be used to measure diagnostic reference doses.