Impact of Biomechanical Modeling of Anatomical Variations on the Uncertainties of the Delivery and Understanding of Radiation Therapy

Abstract

External beam radiation therapy is an effective and widely used focal cancer therapy. However, due to anatomical changes during radiation therapy, both in the tumor and in the normal tissue, the delivered radiation dose can deviate from the planned radiation dose. These responses may compromise the delivery of the most effective treatment and lead to an increased risk of complications in normal tissues. The ability to estimate the delivered radiation to the tumor and normal tissues with high accuracy requires modeling the patient response to dose. Modern medical imaging, such as computed tomography (CT) and medical resonance imaging (MRI), provides a method to evaluate spatial and functional changes of the tumor and normal tissue over the course of radiation therapy. A comprehensive evaluation of these changes requires identification of the tumor and normal tissue, through image segmentation, and accurate alignment of images, through image registration. In the head and neck region, varying angles of neck flexion, rapid tumor response and weight loss cause early changes in healthy tissue. In the abdominal region, motion due to breathing and digestion cause changes in the tumor position and normal structures. When the deviations between delivered and planned dose are great enough, the radiation treatment plan should be reoptimized, in order to ensure that the tumor is adequately treated and that the normal tissue is maximally avoided. Estimating the delivered dose to sufficient accuracy is therefore an important requirement for effective adaptive replanning. This dissertation work develops different techniques based on biomechanical models of the anatomical changes to improve estimates of delivered dose, which can ultimately lead to improvements in treatment adaptation strategies as well as a better understanding of toxicity. A series of experiments based on finite element modeling were conducted to model the uncertainties between planned and delivered dose, as well as the potential impact of modeling on different organ sites. Abdominal normal tissue complication probability models were developed based on estimated delivered dose and their accuracy compared to traditional models based on planned dose. Following this study, a predictive model was developed for the head and neck site, in order to find how early in treatment significant deviations in planned and delivered dose could be predicted. After seeing the large potential deviations between planned and delivered dose in the head and neck site, a comprehensive study was conducted to model the changes that potentially cause these large deviations. This comprehensive head and neck model was developed in two steps; first, the positional changes due to flexion were resolved and second, the dose response to the parotid glands was modeled using finite element modeling. Each clinical site poses different challenges, and this dissertation work highlights two areas in which modeling the deviations between planned and delivered dose will improve advanced adaptive radiation therapy.