On the Influence of HDR Ir-192 Source Position Uncertainty on Dose Variation within the Varian Ring Applicator

Abstract

The tandem and ring (T&R) applicator is used as an alternate to tandem and ovoid (T&O) applicators to deliver HDR Ir-192 brachytherapy for cervical cancer. Since the ring interior lumen is larger than the HDR Ir-192 source capsule outer diameter, the source path within the ring is hard to predict. Some studies have shown that the actual source positions could be several millimeters different from planned positions, with potential for inducing significant dose deviation. However, other studies have demonstrated minimal dosimetric impact (Craciunescu et al 2012). Though most centers choose to implement a global source shift during commissioning (McMahon et al 2011) to limit the effect of inherent offsets, potential dose variations due to source position uncertainty are not yet well quantified. The purpose of this study was to investigate these dose variations for the worst geometry condition, and also to assess the validity of implementing a global offset using Monte Carlo (MC) simulations for dose calculation. The MCNP5 radiation transport code was used for all MC simulations. The GammaMedPlus HDR Ir-192 source core was simulated as a 0.09 cm diameter and 0.35 cm long right cylinder. Dose distributions were simulated in water (12 x 20 cm3 right cylinder) using the *F4 FMESH tally in both transverse (X-Y) and coronal (X-Z) plane. With a (0.1 cm)3 grid size, 109 photon histories gave a typical statistical error (k=1) of <1%. The Varian ring was simulated as a torus (0.29 cm and 0.35 cm inner and outer diameters) with a 0.03 cm thick titanium wall. The black buildup cap (0.5 cm) was simulated with polyoxymethylene (1.425 g/cm3) and dwell positions started from 129.5 cm with 0.5 cm step size in the ring lumen center. Dose variations for eccentric source positions were investigated by simulating dwell positions ± 0.007 cm (largest distance) from the lumen center along the Z-direction as shown in Figure 1a-c. To mimic a T&O geometry, dwell positions were simulated only in 135° to 225° and 315° to 45° as shown in Figure 1.e. To assess the influence of a global shift, ± 0.1, ± 0.3, ± 0.5 cm (Figure 1.d and 1.f), shifts were simulated distal and proximal from the ring end. Dose was normalized to 100% at 0.5 cm from the cap on the X-axis. For dose comparison, Point A was defined 2 cm up from the buildup cap and 2 cm laterally from the ring CAX on right (A_RT) and left side (A_LT). A typical dose distribution (X-Z plane) for the source in the ring lumen center was shown in Figure 1.g. Compared to sources in the ring lumen center, dose to points A_LT and A_RT changed ˗3.2% and ˗3.8%, respectively, for sources 0.007 cm away from Point A, and by 2.4% and 4.0% toward Point A, respectively. Compared to no shift, average dose to Point A was 1.6%, 2.3%, and 3.8% lower for the 0.1, 0.3, and 0.5 cm distal global source shifts, respectively. Doses were 1.2%, 1.4%, and 2.6% lower for the proximal shifts. Neither source position uncertainty within ring lumen nor global source shift caused significant dose variation to either point A_RT or point A_LT. Dose variation due to source position uncertainty was small (< 4%), and can be attributed mostly to inverse square law. Dose variations due to global source shifts also were not substantial. Applying a global shift will improve the accuracy of the dose distribution, but not applying the shift or applying a shift in wrong direction may not cause substantial dose calculation errors.