Premium
A method for physically based radiotherapy optimization with intelligent tissue weight determination
Author(s) -
Jones Lois,
Hoban Peter
Publication year - 2002
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 - 10.1118/1.1428756
Subject(s) - weighting , dosimetry , imaging phantom , intensity modulation , mathematical optimization , computer science , mathematics , algorithm , nuclear medicine , medicine , physics , optics , phase modulation , phase noise , radiology
An important aspect of the intensity modulated radiotherapy (IMRT) process is that of optimizing the beam intensity profiles. Most such methods use a comparison between the defined dose distribution (including prescription dose and dose limits for critical structures) and the current distribution. The comparison may involve dose differences or dose ratios. This paper investigates four different ratio‐based methods, with a hypothetical U‐shaped target and cylindrical phantom as the test case. All methods are shown to give satisfactory results. There are differences in the dose distributions produced and these can be related to the formulation of the respective fluence‐update methods. In common with other IMRT beam optimization methods the techniques used here include tissue‐weighting factors which express the relative importance of achieving the goal dose level or limit. Altering the values of these weights can improve the ability of the optimization algorithm to adhere to the goal dose levels. Typically the weights are adjusted on a “trial and error” basis until the planner is satisfied with the results. In this paper a technique for automatic updating of the weights according to the optimization results is applied to each of the four fluence update methods. With this technique the difference between the current dose level in each tissue and the specified dose constant for that tissue is used to increase the weight should the constraint not be met. It is found that the four methods yield similar results if the weights are updated in this way.