@article{53e7c2803d34428aa63bc4ffba2265e2,
title = "The reliability of optimization under dose-volume limits",
abstract = "Purpose: An optimization algorithm improves the distribution of dose among discrete points in tissues, but tolerance depends on the distribution of dose across a continuous volume. This report asks whether an exact algorithm can be completed when enough points are taken to accurately model a dose-volume constraint. Methods and Materials: Trials were performed using a 3-dimensional model of conformal therapy of lung cancer Trial were repeated with different limits placed on the fraction of lung which could receive > 20 Gy. Bounds were placed on cord dose and target dose inhomogeneity. A mixed integer algorithm was used to find a feasible set of beam weights which would maximize tumor dose. Tests of feasibility and optimality are introduced to check the solution accuracy. Results: Solutions were optimal for points used to model tissues. An accuracy of 3-4% in a volume condition could be obtained with models of 450-600 points. The error improved to 2% with 800 points to model the lung. Solution times increased six-fold at this level of accuracy. Conclusion: The mixed integer method can find optimum weights which respect dose-volume conditions in usually acceptable times. If constraints are violated by an excessive amount, the optimization model should be rerun with more points.",
keywords = "Computer treatment planning, Dose-volume limits, Optimization",
author = "Mark Langer and Richard Brown and Peter Kijewski and Chul Ha",
note = "Funding Information: Supported by Theratrcnics International and the Manitoba Health Research Council. Funding Information: Hardware to calculate 3D dose distributions rapidly is too expensive to be owned by a radiation therapy department as are 3D graphic display devices. Recently the Defense Advanced Research Projects Agency (DARPA) and the National Science Foundation (NSF) allotted funds through the Corporation for National Research Initiatives (CNRI) for five sites to be used as “testbeds” for ultrahigh speed networking. These “gigabit” (IO9 bits/second) networks would be capable of sending or receiving data 1000 times as fast as current high speed networks. A consortium, known as VISTAnet, composed of the Departments of Radiation Oncology and Computer Science at the University of North Carolina, the North Carolina Supercomputing Center, BellSouth and GTE was formed to implement such a testbed network using interactive 30 treatment planning as the driving application. VISTAnet became one of the five funded sites and began work in June, 1990. It should be emphasized that no gigabit network yet exists in the USA outside of the laboratory. Results: We have now demonstrated that it is feasible to calculate more than 50,000 radiation dose points/second from four radiation beams using a CRAY YMPTM supercomputer located at the North Carolina Supercomputing Center 20 miles away. The techniques to do this involve a careful mix of calculation and interpolation, taking full advantage of the vector processing available on the CRAY. After the dose grid is calculated it will be interpolated up to 2563 or 16 million points, and shipped via the gigabit network to Pixel-Planes 5, an ultra high speed multiprocessor graphics engine. There the dose grid is combined with the patient anatomy and is rendered into a 3D display which is sent to the Department of Radiation Oncology for viewing by the clinician. At present, VISTAnet is operating at conventional networking speeds and will do so until the high speed network components are available. Using Sherouse{\textquoteright}s Virtual Simulator as an interface the clinician should be able to set up hundreds of beam configurations and receive immediate feedback as to the 3D dose distribution in the form of high quality 30 graphic displays. In addition, it is planned to have dose-volume histograms, and normal tissue complication probabilities (NTCP) and tumor control probabilities (TCP) calculated and displayed in realtime as well. Armed with these biological models and the ability to examine a 3D dose distribution in detail a clinician should be able to study hundreds of potential treatment plans to find one that is good for his or her patient.",
year = "1993",
month = jun,
day = "15",
doi = "10.1016/0360-3016(93)90972-X",
language = "English (US)",
volume = "26",
pages = "529--538",
journal = "International journal of radiation oncology, biology, physics",
issn = "0360-3016",
publisher = "Elsevier Inc.",
number = "3",
}