An improved Virtual Torso phantom

INTRODUCTION Lung counting has many sources of uncertainty that can impact on the activity estimate such as different deposition patterns of radioactivity in the lung, rib size and placement, chest wall composition and profile, effect of detector size and thickness. For example, if a lung counter is calibrated using a lung set that has the activity homogeneously distributed throughout the tissue substitute material and the deposition in the subject is heterogeneous there will be a large uncertainty introduced. However, to investigate it experimentally would be too expensive and time-consuming, moreover, in some details impossible. Therefore, a virtual phantom derived from the MIRD (1) (Medical Internal Radiation Dose) and the LLNL phantom (2) was designed to study the effect of a heterogeneous deposition on the activity estimate using Monte Carlo code simulation (3). Unfortunately, the virtual phantom had certain limitations. It contained no sternum and the ribs extended all the way round the torso, whereas in reality the central part of the chest is covered with a mixture of cartilage (ribs) and bone (sternum). The ribs were located below the chest wall which added to the thickness of the chest wall. The lungs did not touch the inner surface of the chest wall along their length due to the differences in curvature between the ellipsoidal lungs and the ellipsoidal cylinder that defined the torso. As a result, there was extra intervening tissue between the lungs and the chest wall. This was shown to have a noticeable effect on the simulation of low energy photons; 17 keV photons were predicted to have a counting efficiency approximately a factor of three lower than experimentally observed. This paper describes the improvements and modifications to the virtual phantom that minimise some of these limitations. Experimental data are presented that show the modifications have greatly improved the virtual phantoms characteristics for low energy photons.