FINITE ELEMENT BIOMECHANICAL MODELING OF PAROTID GLAND MORPHING IN PATIENTS UNDERGOING RADIOTHERAPY FOR HEAD AND NECK TUMORS

Patients treated with radiation therapy (RT) for head and neck cancer are exposed during the course of several weeks of treatment (6-7 weeks) to several physiological processes that can induce inter-fraction deformation of the parotid salivary glands (PGs). Gland morphing tends to make them move towards the high dose region of the dose distribution. As a consequence, parotids may be irradiated more than initially planned, leading to increased toxicity. We implemented a biomechanical model of the parotid morphing process that can serve as the basis to optimize adaptive RT protocols, leading to further reductions of dose to these structures. Raystation (R) hybrid deformation algorithms and 3DSlicer (R) tools were employed to obtain a mesh representation of PG anatomy and deformation from CT-series of eight patients treated with highly modulated radiotherapy (Tomotherapy (R)). The biomechanical model was computationally implemented with the finite element software COMSOL (R) multiphysics. Gland tissue was modeled as a linear elastic material with a Poisson ratio of 0.49 and a density value of 1 g/cm(3). A radial force field was introduced to mimic parotid shrinkage due to radiation exposure. Fixed constraints were placed taking into account the anatomical barrier effect of parotid-surrounding structures (mandible) during the shrinkage process. The implemented biomechanical model was able to predict PG morphing with a mean volume difference of 1.36% [0.9-2.0] % between real and modeled deformed anatomy for the first half of treatment and 1.5% [0.8-2.1]% for the second half of treatment. Prediction of geometric overlap of observed versus simulated geometry as quantified using DICE was acceptable for the first half of the treatment but still unsatisfactory for the second half. Adding model components taking into account other elements that define PG morphing (facial geometry changes due to general weight loss and - in patients with significant macroscopic tumor load - local tumor volume loss) may accurately predict parotid morphing and thus permit further optimization of frequency and timing of off-line ART protocols in patients undergoing radiotherapy for head and neck cancer. This may result in a further reduction of treatment toxicity and these approaches are becoming even more important as the use of particle therapy with its increased dosimetric sensitivity towards geometric changes of the patient is slowly increasing.