Patient Specific Delivery of Proton Beam Radiation

Overview

Proton beam radiation therapy, because of its high precision, is used to treat tumors near critical organs, e.g., tumors at the base of the skull. Proton beams lose the majority of their energy in a concentrated region in space called the Bragg Peak. In order to precisely deliver the radiation dose to the tumor, the Bragg Peak needs to be positioned inside the tumor. This is done with a patient-specific block of material called a radiation compensator, placed in the path of the radiation. The thickness of the compensator at each point over the tumor determines the location of the Bragg Peak at each point in the tumor. Conventionally, compensators are made in a milling machine. This method is time consuming, expensive, and wasteful. Furthermore, a given patient can require up to 30 compensators. This project explores a novel, more efficient and cost effective method for constructing patient-specific radiation compensators. This method uses a reusable apparatus, consisting of an array of movable rods whose heights can be controlled with precision. These rods would form the contour of the compensator onto which a sheet of acrylic can be molded. The resulting shell can be filled with water, which is an ideal compensating material. This would significantly reduce the cost of the materials and the time it takes to fabricate the compensators. This project demonstrates the feasibility of the new method with a working proof of concept. This project also determines the shape of the compensator by solving the Bethe-Bloch equation for stopping power and finding its inverse. This project explores two approaches to solve the equation: a Riemann Sum approximation, and a neural network based approach. The neural network is also used to find the inverse function. An original program utilizing the backpropagation algorithm for finding the weights and biases for a multilayer neural network is demonstrated, resulting in a very close approximation of the solution of the Bethe-Bloch equation and its inverse. An alternate approach using the Scikit-Learn Library to determine the weights and biases is also explored.

Competition history

• AJAS 2020 Category not listed

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• AJAS None projects
• Every AJAS 2020 project