Current calibration methods for body counting offer personalisation for lung counting predominantly with respect to ratios of body mass and height. Chest wall thickness is used as an intermediate parameter. This work revises and extends these methods using a series of computational phantoms derived from medical imaging data in combination with radiation transport simulation and statistical analysis. As an example, the method is applied to the calibration of the In Vivo Measurement Laboratory (IVM) at Karlsruhe Institute of Technology (KIT) comprising four high-purity germanium detectors in two partial body measurement set-ups. The Monte Carlo N-Particle (MCNP) transport code and the Extended Cardiac-Torso (XCAT) phantom series have been used. Analysis of the computed sample data consisting of 18 anthropometric parameters and calibration factors generated from 26 photon sources for each of the 30 phantoms reveals the significance of those parameters required for producing an accurate estimate of the calibration function. Body circumferences related to the source location perform best in the example, while parameters related to body mass show comparable but lower performances, and those related to body height and other lengths exhibit low performances. In conclusion, it is possible to give more accurate estimates of calibration factors using this proposed approach including estimates of uncertainties related to interindividual anatomical variation of the target population.