Online External Beam Radiation Planning and Training Felix Hamza-Lup a, Ivan Sopin a, Omar Zeidan b a Computer Science, Armstrong Atlantic State University, Savannah, Georgia, USA b MD Anderson Cancer Center Orlando, Orlando, Florida, USA Abstract. External beam radiation therapy (EBRT) is an effective and widely accepted form of treatment for many types of cancer that requires extensive planning. Unfortunately, current treatment planning systems have limited or no available visual aid that combines patient volumetric models extracted from patient specific CT data with the treatment device in an inclusive 3D interactive simulation. We are presenting new features of a web-enabled system emphasizing several new elements for user interaction and simulation accuracy assessment. We illustrate the potential of 3D simulation in radiation therapy, specifically radiation treatment planning. Embedding patient specific data (CT from DICOM-RT) in a web-based interactive simulation advances the radiation therapy planning by early detection of collision cases. The system may be used to support the decision making process as well as innovations in medical planning and training. Keywords: external beam radiation therapy, distributed training, three-dimensional simulation 1. Introduction External beam radiation therapy (EBRT) is an effective and widely accepted form of treatment for many types of cancer that requires extensive planning. Unfortunately, current treatment planning systems have limited or no available visual aid that combines patient volumetric models extracted from patient specific CT data with the treatment device in an inclusive 3D interactive simulation. Therefore, students and trainees often find it difficult during their external beam rotations to design optimal and deliverable patient-specific plans. In some cases, patient treatment is delayed or postponed due to unforeseen factors that arise on the first day of treatment such as setups issues or possible collision scenarios among the system hardware components or between the hardware components and the patient. In addition, the demand for better cancer targeting has created specific immobilization and on-board imaging devices which often add to the complexity of the planning process. We are presenting new features of the web-enabled system developed through a multidisciplinary research and development effort, emphasizing several new elements for user interaction and simulation accuracy assessment. The paper is structured as follows. Section 2 presents details of the simulator s scope as well as the most recent work by other research and development groups. In section 3 we illustrate the main components of the online system emphasizing the user interaction components. In section 4 we discuss a few simulation assessment strategies and current results. We conclude with a discussion on the importance of online simulation and training systems for the medical field.
2. Rationale Radiation therapy requires accurate graphical visualization of volumetric data to provide adequate analysis, and operation planning. Most of the processed data describes radiation doses which have to be delivered at specific locations to destroy cancerous tissue. Since the success of the treatment depends on the accuracy of planning and delivery, physicians need robust tools to assist them in the planning process. For the past fifteen years, numerous software modules have been added to the planning systems however none of them can provide adequate collision detection and room level setup visualization. A wide range of analytical methods for linac-based radiation surgery have been proposed in the past as a means to improve the EBRT planning process [1-4]. Some of these methods, even though accurate, are based on the hardware numeric rotational and translational values disregarding patient-specific as well as detailed hardware-specific geometry. Previous research and development concerning graphical simulations of linac systems have limitations such as: Simulations involve only generic patient body representations [5] and not accurate hardware 3D models; hence, collisions with patients are not accurately modeled or predicted; Simulations run as standalone applications and cannot be deployed over the web for potential collaboration with remote experts during treatment planning. One of the most accurate representations of the patient and hardware is under integration in a commercial system at the Hull university [6]; however, the current implementation cannot be deployed freely over the Internet. Our efforts, directed towards an online simulator have generated a first prototype in 2005. Since then we have improved the system to provide an accurate representation of the EBRT room setup based on patient-specific medical data. We have deployed a new version of the system on a secure website (http://www.3drtt.org) in 2007 and provided free registration for interested parties. 3. Simulator Description 3.1 Methods and Tools With the advent of the X3D [7] standard and its extended functionality, the Internet- Based systems for simulation gained momentum. X3D is being developed by the Web3D Consortium (originally the VRML Consortium) as a more mature and refined standard. The simulator implementation takes advantage of X3D, and in the development process we employ several software tools for 3D modeling. Fig.1 illustrates snapshots of the web-based simulator (denoted 3D Radiation Therapy Training 3DRTT) which provides a room-based view of the hardware and patient setup.
Fig 1. 3DRTT with Fixed (HTML based) and Floating (X3D based) Graphical User Interface Without going into deeper technical details, laser scanners may be employed to collect point clouds from several viewpoints. These points are merged into one cloud based on a set of specially designated markers. Filtering the noise and wrapping the valid points into a polygonal model is a very important process. Due to the inaccuracies of the laser scanner (approx. 3mm), we post-process the 3D models, obtaining better consistency with the real equipment. The polygonal model is exported into an X3D object and embedded in the simulator. Considering the geometrical complexity and the highpolygonal resolution of the model, we have to optimize the scene such that adequate frame rates (30 Frames per Second or more) are obtained on machines with less rendering power as the users navigate online in the simulated room. To further improve the rendering speed and reduce the file size, we make use of textures, simulating the geometry of complex areas. Special processing might be needed in some cases. For instance, the table contains a special glass-like component. Because the glass-like material does not attenuate the beam in reality, we must tune the transparency of the surface to resemble the glass characteristics. 3.2 Interaction with the Simulator 3DRTT provides an intuitive floating graphical user interface (GUI) for controlling the angles and locations of the machine s parts. The GUI is designed in the form of multiple semitransparent windows holding various volumetric controls (Fig. 2 a). The GUI components can be easily rearranged to avoid occlusions of important objects. Fig. 2 a) GUI with volumetric switches and dials b) Distance Measuring Tool The measuring tool (Fig.2 b) allows users to determine the exact distance between any two points in the virtual space. Such measurements are useful for simulation accuracy assessment and in collision scenarios, when distance misinterpretation is possible.
The collision mode activates an automatic collision warning system to provide information about potential collisions in the hardware setups (Fig.3). The collision detection system is based on polygonal primitives, and the algorithm has been optimized to work in an interactive web-based environment. As the collision detection system brings any small clearance case to the user s attention, the measuring tool can be used to obtain precise measurements following a collision warning. Fig. 3 Visual collision validation: the actual setup (left), and the simulator (right). 3.4 Patient 3D Model Generation from DICOM-RT CT An important issue we address in this project is the inclusion of real patient data in the simulation. We have to efficiently convert a set of CT scans of a patient to the polygonal model of the patient s body. The set of CT scans used is stored using DICOM RT standard [8]. As opposed to other types of image files, the DICOM RT file standard contains slice resolution, slice spacing, and pixel size which all are useful parameters in producing a realistic polygonal model of a patient. We process the CT scans using algorithms similar with the ones in the Visualization Toolkit (VTK) [9] with additional improvements for rendering accuracy. We apply the Marching Cubes algorithm with a value that selects the isosurface of the patient s skin Fig.4 (middle). Then the 3D model is embedded in the simulator (Fig.4 right). Fig. 4 CT data, 3D model reconstruction and 3D model inside simulator For calibration purposes (and training sometimes) predefined shapes are used by the medical personnel. An example is the elliptical phantom device (Fig. 5), used for testing and training purposes. Such phantoms have a predefined usually regular shape. We are in the process of developing a library of 3D models that will augment the existing simulation with such components.
Fig. 5 The phantom device: real environment (left) and simulated environment (right). 4. Simulator Assessment Preliminary To objectively test collision scenarios we asked the radiation therapy technician and the therapist to simulate a plan that contains collisions among the system components (illustrated in Fig.5). The simulator provides an accurate representation of the hardware (specifically the Varian 23iX LINAC) that can predict any collision scenarios within one centimeter accuracy. Fig.6 Visual Collision Validation The preliminary assessment process illustrated in Fig.6 provides an early validation for the accuracy of the simulator. The tool can be used off-line by planners to inspect their patient-specific beam arrangements. A comprehensive validation is underway involving a large set of hardware setups and measurements. We will report the results in future work this year. We are also in the process of measuring the impact of the simulator on the EBRT planning and execution efficiency. Also a large scale assessment of the system for teaching and training purposes is underway in parallel with the integration of the system in radiologic sciences departments curriculum.
5. Conclusion We have presented an online EBRT simulator that has the ability to detect/predict all possible collisions between LINAC components (including machine add-on accessories and immobilization devices) for a given patient. The simulator may eliminate the need for backup plans and saves planning time. In addition, it enables the planner to explore different and unconventional combinations of gantry-couch-collimator for treatment that may give rise to better quality plans. Hence the uses of the online EBRT simulator benefits are two fold: educational and clinical. For the education component, the targeted trainee groups includes physics residents, dosimetry trainees, and radiation therapy techs. The online web-based system will allow trainees and students to: Familiarize the trainee with the various components of the linac including table, gantry, and collimator translational and rotation motion limits and angle conventions Validate patient setup, plan deliverability, and to check for possible collision scenarios and beam-couch intersections Educate patients about their treatment delivery technique and help reduce pretreatment anxiety. In addition to its educational benefits, the web-based system will be used clinically to improve the overall quality of radiation therapy patient treatment, enabling the planner to explore differing and unconventional gantry-table-collimator combinations for treatment that may give rise to better quality plans. References: [1] Beange I, Nisbet A. Collision prevention software tool for complex threedimensional isocentric set-ups. Br J Radiol. 2000 May;73(869):537-41. [2] Hua C, Chang J, Yenice K, Chan M, Amols H. A practical approach to prevent gantry-couch collision for linac-based radiosurgery. Med Phys. 2004 Jul;31(7):2128-34 [3] Humm J, Pizzuto D, Fleischman E, Mohan R. Collision detection and avoidance during treatment planning. Int J Radiat Oncol Biol Phys. 1995 Dec 1;33(5):1101-8. [4] Purdy JA et al. In advances in 3-dimensional radiation treatment planning systems: room-view display with real time interactivity. Int J Radiat Oncol Biol Phys. 1993 Nov 15;27(4):933-44. [5] Tsiakalos MF, Scherebmann E, Theodorou K, Kappas C. In graphical treatment simulation and automated collision detection for conformal and stereotactic radiotherapy treatment planning. Med Phys. 2001 Jul;28(7):1359-63. [6] Beavis A, Ward J, Bridge P, Appleyard R, Phillips R, editors. An immersive virtual environment for training of radiotherapy students and developing clinical experience. Proceedings of AAPM; 2006 Jul 30 Aug 3; Orlando, FL, USA. Melville: Amer Assoc Physicists Medicine Amer Inst Physics; 2006. [7] Web3D Consortium. What is X3D? Available at http://www.web3d.org/about/overview. Accessed December 1, 2007. [8] NEMA. DICOM Homepage. Available at http://medical.nema.org. Accessed March 1, 2007. [9] Schroeder WJ, Martin KM, Lorensen WE. The design and implementation of an object-oriented toolkit for 3D graphics and visualization. Proceedings of VIS '96; 1996 Oct 27 Nov 1; San Francisco, CA, USA.