Use this URL to cite or link to this record in EThOS:
Title: The effect of photon dose calculation algorithms on the clinical outcome of radiotherapy as assessed by radiobiological models
Author: Chandrasekaran, Mekala
Awarding Body: University of Liverpool
Current Institution: University of Liverpool
Date of Award: 2012
Availability of Full Text:
Access from EThOS:
Access from Institution:
The accuracy of dose calculation algorithms used for radiotherapy treatment planning play a significant role in the clinical outcome of various treatment regimens. Heterogeneities in human anatomy such as lung, air cavities, bone, soft tissue and fat present challenges to the dose calculation algorithms as they are prone to disrupt the charged-particle equilibrium. Monte Carlo (MC) based dose calculation algorithms are proven to be superior to all the current analytical algorithms owing to their ability to account for all the physical interactions that are involved in radiation transport. Numerous publications have examined the differences in physical doses calculated by analytical algorithms when compared to MC in dealing with heterogeneities. However, before this work the clinical significance of these differences in physical dose has never been investigated in detail. An EGSnrc, BEAMnrc and DOSXYZnrc based MC dose calculation engine was set up in a parallel computing environment to simulate three-dimensional conformal radiotherapy (3DCRT) and intensity modulated radiation therapy (IMRT). A Varian 2100 C/D accelerator head was modeled and validated to match measurements of open and dynamic wedged fields in a homogeneous water phantom which was found to be in good agreement with measurements within 2%/2mm and 3%/3mm respectively. In addition, MC calculated doses in a heterogeneous lung phantom were compared to radiochromic film measurements. Overall, there was good agreement between the two, although large differences of upto 16% were found in some cases. This dose calculation system was used to perform MC simulations on computed tomography (CT) images. The clinical impact of the differences in absolute doses calculated by various photon dose calculation algorithms for two clinical tumour sites was investigated. The tumour control probability (TCP) and normal tissue complication probability (NTCP) were estimated using well established bio-mathematical radiobiological models. This work includes the analysis of 7 convolution (i.e. pencil-beam) and convolution-superposition (CS) based photon dose algorithms available in commercial treatment planning systems (TPSs) as well as MC, in treatment plans of non-small cell lung carcinoma (NSCLC) and nasopharyngeal carcinoma (NPC). In both NSCLC and NPC, the convolution algorithms overestimate the dose to the tumour and hence overestimate the TCP to up to 45%. Some of the CS algorithms were comparable to MC though others exhibit significant differences. In NSCLC, the absolute differences in the NTCP values with radiation pneumonitis and rib fracture as end points were not as large as the differences found in the TCPs. On the other hand, in NPC, the overestimation of probability of occurrence of xerostomia by some TPS algorithms may be preventing dose escalation. Parameters for the TCP model were derived by fitting the TCP predictions to published outcome for four widely varying dose-fractionation regimens for a patient cohort undergoing radical radiotherapy treatment for NSCLC. The derived parameter sets strongly depend on the accuracy of the dose calculation algorithm involved. Parameters derived based on dose-distribution data sets obtained using one particular dose calculation algorithm may not hold good when evaluating treatment plans calculated with a different algorithm. In this sub-study, the influence of dose calculation algorithms on TCP model parameters was evaluated. Significant differences were found in TCPs when calculated with inconsistent parameters. Hence, the choice of dose calculation algorithm is crucial and although some algorithms generally perform close to MC in handling inhomogeneities, it is necessary to understand how the underlying differences affect the predicted clinical outcome.
Supervisor: Nahum, Alan; Baker, Colin Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: RC0254 Neoplasms. Tumors. Oncology (including Cancer)