A Dynamic Programming Approach to Adaptive Radiation Therapy

Abstract:

The objective of external beam radiation therapy is to deliver beams of radiation to a patient from different angles so that the prescribed dose (as specified by a doctor) to the tumor is met. Conventional treatment planning procedures deliver equal dose fractions every day over the course of 30-40 days. In this work, we consider delivering different dose fractions each day. We focus on a model of the day-to-day variations of the gap between the tumor and organ-at-risk (OAR), which is usually the limiting factor in escalating the dose to the tumor. Using an adaptive dose fraction rather than a fixed one can allow us to take advantage of a "favorable" or large tumor-OAR gap by increasing the dose to the tumor. Similarly, we can decrease the dose fraction for an "unfavorable" or a small tumor-OAR gap. This results in a lower cumulative dose to the OAR over the course of the treatment. We formulate the problem using a dynamic programming framework and discuss some interesting results.

Biography:

Jagdish Ramakrishnan is currently a Ph.D. student working with (i) Prof. John Tsitsiklis from MIT and (ii) Dr. Thomas Bortfeld and Dr. David Craft from the Massachusetts General Hospital (MGH). His current research interest is optimization of radiation therapy planning. He received his Bachelor's and Master's degrees in Electrical and Computer Engineering (ECE) from Georgia Tech.