Abstract
Intensity modulated radiation therapy (IMRT) of nasopharyngeal cancer (NPC) involves delivering a high radiation dose to the nasopharynx, and parts of the skull base. As the head and neck region is characterized by complex anatomical structures that involve tissue, bone, and air interfaces, the first objective of this study was to study the effects of tissue heterogeneity on dose distribution in NPC patients treated with IMRT. Moreover, the ears are a very small critical organ located near the high dose area and are prone to radiation-induced toxicity. The second objective of this study, therefore, was to develop a high-resolution (HR) voxel phantom to evaluate the dose distribution of the auditory apparatus during IMRT. The measurement-based Monte Carlo (MBMC) method was adopted for the study. The major components of the MBMC technique involves (1) the BEAMnrc code for beam transport through the treatment head of a Varian 21 EX linear accelerator, (2) the DOSXYZnrc code for patient dose simulation and (3) an electronic portal imaging device (EPID) for measuring the efficiency map which describes non-uniform fluence distribution of the IMRT treatment beam. Beam parameters and the aS1000 EPID at Tungs’ MetroHarbor hospital were commissioned for this study. Ten NPC IMRT plans were evaluated for dose effect due to tissue heterogeneity by comparing dose distributions calculated with Eclipse plans and those obtained from the Measurement-based Monte Carlo (MBMC) method. The Eclipse plans were based on the anisotropic analytical algorithm (AAA). For further evaluation of dose effect to small critical structures, MBMC simulations using HR phantoms (HR-MBMC) were applied to the IMRT treatment plans of three of the ten NPC patients and an additional skull base tumor. In-house MATLAB (R2010a) program was developed to create the HR voxel phantoms. CT slices 3mm in thickness in the patient’s head and neck region and 1 mm in the skull base area were obtained. The CT images were rescaled using bicubic interpolation of the 'imresize' function in MATLAB. The voxel size of the HR phantoms was 0.05 x 0.05 x 0.1 cm3 in the skull base area and 0.05 x 0.05 x 0.3 cm3 for other parts of the phantoms. MBMC simulation on 10 NPC patients revealed that in PTV1 the mean value of the volume receiving at least 95% of the prescribed dose (VPTV95), was slightly lower for MBMC (98.7%) than that for AAA (99.0%). The dose to 95% of PTV1 (D95%) also showed lower mean dose for MBMC (6832 cGy) when compared with AAA (6895 cGy). On the other hand, MBMC simulation predicted a higher dose distribution to the optic nerves, lens, eyeball, spinal cord, temporomandibular joints, parotid glands, and middle ears than AAA. The difference in the mean doses between MBMC and AAA suggests that critical organ doses should be confirmed in order to avoid serious complication from overdose. HR-MBMC simulation of three NPC patients revealed that in PTV1 the mean D95% for HR-MBMC (6763.3 cGy) was less than that for AAA (6847.1 cGy). Small volume organs (volume < 1 cm3) such as the eighth cranial nerve, semicircular canal, and cochlea showed a mean dose increase of 287.5 cGy when compared with AAA. HR phantom simulation of the skull base tumor also showed higher dose to the ear structures than AAA. The mean doses predicted by the HR-MBMC for the right 8th cranial nerve, right cochlea, and right semicircular canal compared with AAA were 751.5 cGy vs. 732.2 cGy, 532.5 cGy vs.468.2 cGy, and 870.7 cGy vs 817.0 cGy, respectively. It suggests that HR-MBMC has the potential to assist in detailed dose analysis for small critical organs (<1 cm3) in high dose gradient skull base area. The MBMC dose simulation method can serve as a good dose evaluation reference for IMRT plans having complex tissue composition and small critical structures since it applies EPID measured efficiency maps with very fine spatial resolution. The HR-MBMC method is well suitable for detailed evaluation of dose distribution for small volume organs such as the ears.