<- Back to main page

Isodose 3D

Introduction

Radiation therapy is a set of modern techniques, based on the use of ionizing radiations (beta, gamma, or X-rays), for treating cancerous tumours. Their goal is to apply a prescribed radiation dose to a Primary Target Volume (PTV), while trying to avoid irradiating Organs At Risk (OAR) as most as possible. Two major techniques of radiation therapy exist, differing in the way the radiation is applied to the tumour. External radiotherapy consists in applying ionizing rays from outside the patient's body whereas brachytherapy uses radioactive sources which are placed within the body.

Intensity-modulated radiation therapy (IMRT) is an advanced mode of external radiotherapy that utilizes computer-controlled X-ray accelerators to deliver precise radiation doses to a malignant tumor or specific areas within the tumor. The radiation dose is designed to conform to the three-dimensional (3-D) shape of the tumor by modulating or controlling the intensity of the radiation beam to focus a higher radiation dose to the tumor while minimizing radiation exposure to the surrounding normal tissue.

Low dose rate brachytherapy (LDR) is a technique which is widely used for treating prostate cancer. In that specific case, it is made use of several iodine 125 radioactive sources which are being inserted in a final way within the patient's body. The sources activities are initially low, and decrease fast. Thus, the complete treatment is done when the activity of the sources has decreased to zero, making the removal of the sources useless. An other LDR brachytherapy techniques make use of wire-shaped iridium 192 sources, like those used in the Paris System, which is described by Dutreix et al. [2]. They are inserted within the tumour through catheters also called “vectors”, then removed once the prescribed dose has totally been delivered to the tumour.

High dose rate brachytherapy (HDR) makes use of a device called remote afterloader which is able to direct a single punctual iridium 192 source into catheters/vectors. The source may be stopped at several regularly spaced locations (dwell positions) within a catheter during determined times called dwell times. The total dose distribution is determined by the dwell positions the source stops at and the amount of time it dwells at each different position.

Current optimization tools for HDR brachytherapy are efficient for the determination of the values of the dwell times. Their input data are the 3D representation of the PTV and the already implanted catheters. However, the rules for the implantation of the catheters are not clearly defined. Generally, radiation oncologists place the catheters using empirical methods based on experience. Other methods are inspired from the rules defined in LDR implantation systems such as the Paris System.

These latter methods seem to offer more warranties concerning the quality of the resulting treatment, since they use strict rules providing satisfying dose distribution and target volume coverage.

The goal of my study was to establish the basics of a new methodology based on optimization, which provides implantation schemes adapted to the modern treatment technologies. This methodology must take into account the constraints which are included in the existing LDR implantation methods, in order to ensure the reliability of any treatment obtained from an implantation scheme.

We developed Isodose 3D, an optimization and visualization software program which displays the result of the dwell times optimization on an arbitrary specified volume, and in addition enables to test a specified implantation of the catheters in a very straightforward way. This software is released under the GNU General Public License, which allows anybody to use it and modify it freely.

Description

Isodose 3D has been developed in order to obtain a visualization of the searched results. This provides the two following advantages: first, for an optimization researcher to get a comprehensive visualization of physical and physiological phenomena for which he doesn't master all the aspects; and second, to show in a straightforward way the obtained results to physicists and radiation oncologists.

Globally, the idea behind this software is to provide all the visual and computation tools, allowing to obtain isodose surfaces for implantations defined under the terms of the Paris System. The software optimizes, for a standard ¹⁹²Ir source type, the dwell times in such a way that the prescribed dose on the target volume equals to 1 Gy. Isodoses are the same whatever the prescribed dose, because the dose calculation function is linear. Thus, a proportionality factor equal to any prescribed dose can be applied to the obtained dwell time values, in order to get the dwell times corresponding to the given prescribed dose.

The application is constituted by a graphical interface for the input of the parameters of the implantation problem, associated to a 3D visualization module. It has been developed in C language on a GNU/Linux PC. The graphical interface API is GTK+, and the 3D rendering is being done through the use of OpenGL.

Functionalities

Isodose 3D allows the definition of a phantom (named target volume in the software) of the kind of those defined in the Paris System: right-angled parallelepiped or prism with isosceles trapezoidal base. The user freely chooses the dimensions of target volume, in the three space directions. The target volume is directly visible on the screen.

The user chooses a positioning for the vectors (squares or triangles organization, a number of planes, vectors per plane, length and spacing of the vectors), and can directly observe their placement in superposition with the target volume. From this positioning, it is possible to generate the reference isodose surface (surface of points receiving a dose equal to the prescribed dose), according to various possibilities:

• by manually specifying a constant dwell time value;
• by calculating the basic dose rate as defined by the Paris System [2], and by deducing a constant dwell time value allowing to obtain a reference dose of 1 Gy on the isodose of 85% of the basic dose rate;
• by optimizing the dwell times with the linear programming formulation defined in [1].

With each isodose generation a DVH is generated, and the values of c₁, c₂, and the COIN are calculated (for more information, see chapter on dose evaluation in [1]). This provides, in addition to the visual appreciation of reference isodose obtained, analysis tools the radiation oncologists are accustomed to handling.

Figure 1 presents a general view of the application interface.

The left part of the screen displays a 3D representation of the problem. We can simultaneously see the target volume, the generated reference isodose (3D surface), and the vectors of the application, represented by the black segments. The white points distributed on the vectors are the possible dwell positions of the source. All these objects are visible simultaneously on the 3D representation, thanks to the use of object transparency. That makes possible to easily observe and appreciate the covering of the target volume by the reference isodose.

The generated isodose surface is an approximation made of small triangles. The approximation method is a well-known technique in 3D modeling, called the Marching Cubes algorithm [3]. The smooth shading lighting technique is a visual effect to give the user the impression the surface is actually smooth.

It is possible to observe the problem from various angles, which has the advantage of potentially not missing any aspect of the problem. Moreover, certain view angles make it possible to better illustrate one aspect of the problem or another. For example, figure 2 shows the same 3D scene under two different angles. That makes visible various parts of the target volume not covered by the reference isodose, according to the selected angle.

The right-hand side of the screen features a set of controls split inside a tabbed notebook (Figure 3). Each tab groups the following functionalities:

• target specification
• vector positioning specification
• optimization parameters settings
• display settings

Target Volume specification

In this tab (Figure 3(a)), the user chooses the type of target volume. The two possible volume types are right-angled parallelepiped or prism with isosceles trapezoidal base. The dimensions of the target volume can be freely chosen, and any modification is directly visible on the 3D representation. Figure 4 illustrates the available target volume types.

Definition of the Vectors

In this tab (Figure 3(b)), it is possible to specify the positioning of the vectors. The possible settings are the number of planes, the number of vectors per plane, the constant spacing between vectors, and the length of the vectors.

The type of vector disposition (square or triangle) depends on the specified target volume type. Concerning the number of vectors per plane in the case of a vector triangle disposition, the “number of vectors per plane” parameter concerns the first plane of the application. Whenever a second plane is added, it will feature one more vector, and so on for every additional plane.

The user can also choose the spacing of dwell positions inside the vectors. The two possible choices (0.5 cm and 0.25 cm) correspond to the most common real-case values, depending on the afterloader in use.

As well as in the first tab, any modification is directly visible on the 3D display.

Figure 5 shows two different vector dispositions for the same target volume.

Optimization Settings

In this tab (Figure 3(c)), the user can first specify the optimization space dimensions, the number of uniformity and contour points.

For our studied cases (see Results section of [1]), the expression of dose constraints in the linear program is limited to a lower dose bound for points inside the PTV, and an upper dose bound for points inside the surrounding space. Thus there are only two weighting coefficients of the objective function, one applied to the lower dose bound variations in the PTV (noted α_PTV), and the other to the upper dose bound variations, noted β_NT. The values for these parameters can also be input from this settings tab.

The optimization space is represented by an imaginary box created around the target volume and the normal tissue (see figure 1). The uniformity points are dose reference points generated uniformly in the calculation box. The contour points are generated at the surface of the target volume.

The volume of this box, defined by the user, must be large enough to represent a significant amount of normal tissue around the target volume. The volume should as well be small enough, in order to allow a maximal density of dose reference points, while limiting their number. In general, once this volume has been specified, the user chooses to hide it, in order not to overload the display.

Display settings

The available controls in this tab (Figure 3(d)) enable to setup the display.

Two different 3D object lighting methods are proposed: smooth shading and flat shading. Object lighting is used to emphasize the 3D volume shape, which gives the user good visual information about the volume shape. If no lighting is used, only the contour of single-color shapes would be visible. Smooth shading simulates natural reflection of a light source on a smooth object, even though the represented object is made of polygons. In the opposite, flat shading emphasizes the polygons by setting a single color to each polygon, depending on its orientation relatively to the light source and the point of view.

The space discretization which is used by the Marching Cubes algorithm for the generation of the isodose surface can be modified by the user. A smaller discretization provides a more accurate representation of the surface, because the approximation is done with smaller triangles. Though, as the number of triangles increases quickly when the discretization is reduced, the amount of necessary memory and computation power increases in the same way.

For the different generated 3D objects, it is possible to set an opacity parameter. The opacity determines the transparency of objects, and setting different opacity values for the different objects allows to view different objects at the same time, even when some are partially or totally covering others. The opacity value for an object varies from zero, where the object is invisible, to one, where the object is fully opaque and no other object can be seen through it.

Another display setting is the isodose level. For example, choosing an isodose level of 2 will diplay the isodose volume which contains 200% of the prescribed dose, which corresponds to the hyperdose sleeves defined in the Paris System. Figure 6 shows generated 100%, 150% and 200% isodoses for the same application.

Various points of interest can also be displayed: the dose reference points used in the optimization (uniformity and contour points), the hyperdose points, and the calculation points used for the DVH generation. Each sort of points is displayed in different colors, which can for example provide good visual information complementarily to the calculated dose conformality indices.

Download

You can download Isodose 3D from here :

• isodose_3d-1.0.tar.gz

All required software packages and compilation instructions are mentioned in the INSTALL file in the archive.

Isodose 3D is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version.

References

1. Galea, F. 2006. Problèmes d'optimisation en curiethérapie, PhD thesis, Université de Versailles, 45 avenue des Etats-Unis, F-78035 Cedex, France.
2. Dutreix, A., Marinello, G. Wambersie, A. 1982. Dosimétrie en curiethérapie, Masson, Paris.
3. Lorensen, W. E., Cline, H. E. 1987. Marching cubes: A high resolution 3d surface construction algorithm, Computer Graphics (Proceedings of SIGGRAPH '87) 21(4): 163--169

©2008-2013 François Galea - Last Modified 01/29/12 22:45