A system of transport equations and an inverse problem from radiation therapy

TL;DR

This paper studies the existence of solutions for a complex system of transport equations modeling radiation therapy, and explores an inverse problem for treatment planning as an optimal control problem.

Contribution

It establishes existence and uniqueness results for a coupled Boltzmann transport system and formulates the inverse radiation treatment planning as an optimal control problem.

Findings

Proved existence of solutions for the coupled transport equations.

Formulated the inverse problem as an optimal boundary control problem.

Provided variational equations for the optimal control.

Abstract

The paper considers existence results of solution for a linear coupled system of Boltzmann transport equations and related inverse problem. The system models the evolution of three species of particles, photons, electrons and positrons. Hyper-singularities of differential cross sections associated with charged particle transport cause that modelling contains the first order partial differential term with respect to energy and the second order partial differential term with respect to velocity angle. The overall system is a partial integro-differential equation. The model is intended especially for dose calculation in the forward problem of radiation therapy. Firstly we consider a single transport equation for charged particles. After that we verify under physically relevant assumptions that the coupled equation together with relevant initial and inflow boundary values has a unique solution in appropriate $L^2$-based spaces. The existence of solutions for the adjoint problem is verified as well. Moreover, we deal with the related inverse problem, a so called inverse radiation treatment planning problem. It is formulated as an optimal boundary control problem. Variational equations for an optimal control related to an appropriate differentiable strictly convex object function are verified. Its solution can be used as an initial point for an actual optimization which needs global optimization methods.