Premium
Poster — Thur Eve — 60: Development of an Empirical Model for Respiratory Motion
Author(s) -
Quirk S,
Becker N,
Smith W
Publication year - 2010
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.3476165
Subject(s) - akaike information criterion , goodness of fit , breathing , statistics , mathematics , sigmoid function , population , regression analysis , position (finance) , linear model , computer science , artificial intelligence , medicine , environmental health , finance , artificial neural network , economics , anatomy
In order to simulate the effects of respiratory chest motion on breast cancer radiotherapy treatments, an accurate model of clinically relevant breathing patterns is needed. Current models of the chest wall are simple and ignore population variation in breathing shape and the asymmetries in inhale and exhale phases of the breathing cycle. This study aims to develop a realistic, empirical breathing model of the external motion of the chest wall. Development of this model will allow examination of how different breathing patterns may affect treatment deliverability and patient outcomes. We fit multiple sigmoidal functions to clinical respiratory data acquired by Varian's Real‐Time Position Management System (RPM, Varian Medical Systems, California). Nonlinear regression was used to quantify the goodness of fit parameters for the empirical model. Increasing the number of variables in the model increases the goodness of fit, and we used Akaike's Information Criterion (AIC) to make direct comparison of models accounting for the difference in the number of parameters. Using AIC, we found that for the inhale portion, a sigmoidal fit with four parameters was the most appropriate, while the exhale portion fit with a sigmoidal curve the three could be used instead of the four parameter fit. AIC information theory proved to be a useful tool in determining the likelihood of fit while accounting for the number of parameters. Further investigation will attempt to reduce the number of parameters by investigating trends in the coefficients of the fit.