Sperduto_Khan's Treatment Planning in Radiation Oncology, 5e

Treatment Planning Algorithms: Photon Dose

Calculations 20

John P. Gibbons

calculation models are an area of continuous development, and it is likely that each commercial vendor’s implemen tation of one or more of these models will differ in many respects. Nevertheless, the intent is to provide a basic under standing of the principles behind each of these algorithms. REPRESENTATION OF THE PATIENT FOR DOSE PLANNING Patient representation has evolved dramatically over the past 50 years. Initially, patients were considered as a flat water phantom of a specific source-to-surface distance (SSD) and depth for use in simple dose or monitor unit calculations. Development of external contour tools aided the treatment planner in determining patient-specific dose distributions. Such procedures resulted in the patient being represented as a homogeneous composition (i.e., water) but did allow for the application of surface corrections to the calculation. Patient heterogeneities could be represented in simple ways, such as using internal contours with assigned densities. The electron density to assign to the region could be inferred from CT atlases or, if available, the mean patient-specific CT number within the contoured structure. 3 The problem with this approach was that tissues such as lung and bone are not themselves homogeneous, and their density varia tions would not be taken into account using this approach. All modern radiotherapy systems use volumetric imag ing data to characterize the patient in a 3D voxel-by-voxel description. The most common imaging dataset used for radiotherapy treatment planning is a treatment-planning CT scan, obtained using a conventional CT simulator. A CT dataset of the treatment region constitutes the most accurate representation of the patient applicable for dose computation, primarily because of the one-to-one rela tionship between CT number and physical and/or elec tron density. 3 Dose algorithms that can use the density representation on a point-by-point basis are easier for heterogeneous calculations because contouring of the het erogeneities is typically not required. An exception to this occurs when data are present within the CT scan which will not be present for the treatment. One obvious exam ple is the CT-simulator couch, which is either manually or

INTRODUCTION Computerized treatment planning systems have been uti lized in radiotherapy planning since the 1950s. The first computer algorithm used has been attributed to Tsien 1 who used punch cards to store isodose distributions to allow for the addition of multiple beams. Since that time, advancements in computer speeds and algorithm devel opment have vastly improved our capability to predict photon dose distributions in patients. In an early attempt to classify computer planning algorithms, the International Commission on Radia tion Units and Measurement (ICRU) Report 42 2 divided photon dose calculation methods into two categories: empirical and model-based algorithms. Early empirical algorithms such as Bentley–Milan were developed using clinical beam data measured on a flat water phantom as input. Corrections were then made to incorporate various effects, such as changes in patient external contour, block ing or physical wedges, and so forth. Eventually, patient heterogeneity correction factors were incorporated, but these were applied afterward, that is, after water-based calculations were performed assuming a homogeneous patient geometry. Most of this development occurred prior to the advent of computed tomography (CT), or at least before the incorporation of CT images into the radiotherapy planning process. However, eventually the commercial utilization of empirical algorithms faded. In the early 1990s, three dimensional conformal radiation therapy (3D CRT) began to use patient-specific CT-image data in the plan ning process. Initially, this was limited to virtual simula tion. At that time computer-based algorithms that could incorporate the newly available volumetric density infor mation and compute true 3D dose distributions in a rea sonable amount of time were not yet available. In order to fully utilize this new information, it was necessary to develop new algorithms that could incorporate variations in individual patient anatomy. As a result, most, if not all, commercial treatment planning systems have moved to model-based photon calculation methods. In this chapter, we describe three photon calculation models currently in use in radiotherapy clinics. Photon

Copyright © 2021 Wolters Kluwer, Inc. Unauthorized reproduction of the content is prohibited.

437