Poster - Thur Eve - 63: Prostate IMRT: Product-Mixture model of a two-dimensional probability density function integrating the variability of the motion of the rectum and the rectal wall thickness

Abstract

This study investigated the dependence of the probability density function (pdf) describing rectal wall geometry on the rectum position (Rm) and the rectal wall thickness (tw). Probability density functions describing the organ motion uncertainties of the rectum (pdfM) have been reported by many authors. In this study, we further proposed a pdf describing the changes in rectal wall thickness (pdfTW) and hence introduced a two-dimensional function pdfM&TW, incorporating the variability of RM and tW using their pdfM and pdfTW, respectively. Our study is based on the average, , of 587 prostate patients. The new pdfTW was established as a mixture of a three-mode distribution with specific mean value (μ), standard deviation (σ) and weight (w), namely, (μF = 3.31, σF = 1.82 and wF = 77.1%), (μPF = 7.7, σPF = 0.809 and wPF = 15.2%) and (μE = 10.27, σ = 0.906 and wE = 9.4%) for the full, partially full and empty state of the rectum, respectively. The pdfM&TW function was introduced as a product-mixture model of the two functions pdfM and pdfTW and it has been graphically and mathematically reviewed. Our pdfM&TW model is a more realistic representation of the probability of the geometric rectal configuration that can occur during prostate IMRT than the model using only the motional pdfM and assuming a constant rectal wall thickness in the plan optimization.