Lung Density and Algorithm Choice in Radiotherapy: Implications for Accurate Lung Cancer Treatment Planning

Abstract

Accurate dose calculation is crucial for effective lung cancer radiotherapy, particularly in heterogeneous media where density variations can result in significant deviations between planning algorithms. This study systematically compared the fast collapsed cone convolution (CCC) algorithm against a high-accuracy Monte Carlo (MC) dose engine within the RayStation treatment planning system. Two main investigations were conducted: (1) a controlled phantom study consisting of a water block with lung-equivalent inserts of varying densities (1.0, 0.4, 0.3, 0.2 g/cm3 ) irradiated by 6 MV and 10 MV photon beams across four field sizes (2×2, 3×3, 5×5, 10×10 cm2), and (2) a clinical patient study involving three anonymised VMAT plans for small, medium, and large lung tumours. In the phantom study, PDD, lateral profiles, and sagittal dose slices revealed that CCC overestimates superficial (skin) dose and in-lung dose relative to MC. These effects intensified with smaller field sizes, lower lung densities, and higher beam energy due to lateral electronic disequilibrium. Quantitative ROI analyses showed that, although out-of-field differences were notable, they were statistically indistinguishable when accounting for MC noise. CCC’s deviations in lung-insert dose became significant for small fields (≤ 3×3 cm2) in low-density media. In the patient study, CCC overestimated entrance skin dose yet underestimated tumour and organs at risk coverage by up to 5.9 % of the clinical goal. These findings align with prior literature on lateral scatter limitations in CCC. Clinically, our results affirm that CCC plans should be verified by MC, especially for small-field, low-density lung targets, to ensure accurate tumour coverage and normal tissue sparing.