A COMPUTATIONAL PHANTOM OF HEAD AND NECK - SISCODES

2007 International Nuclear Atlantic Conference - INAC 2007 Santos, SP, Brazil, September 30 to October 5, 2007 ASSOCIAÇÃO BRASILEIRA DE ENERGIA NUCLEAR - ABEN ISBN: 978-85-99141-02-1 A COMPUTATIONAL PHANTOM OF HEAD AND NECK - SISCODES Larissa Thompson, Bruno Machado Trindade and Tarcísio Passos Ribeiro Campos Departamento de Engenharia Nuclear Programa de Pós-Graduação em Ciências e Técnicas Nucleares Universidade Federal de Minas Gerais Av. Antônio Carlos, 6627 31.270-901 Belo Horizonte, MG larissa.thompson@ig.com.br bmtrindade@yahoo.com campos@nuclear.ufmg.br ABSTRACT A computational voxel model of a head and neck adult male was built through the SISCODES code, in order to complement and to optimize the radiotherapic treatment in head and neck cancer, in terms of dosimetric evaluation and prediction. This computational simulator object, namely computational phantom, is a useful tool for the elaboration and simulation of the three-dimensional radiation planning. The present phantom was assembled based on information generated by photographic images of the visible human project. The images were digitalized and converted one by one to a matrix of voxel, in which tissues and its respective chemical composition were identified, with the helping of a biomedical and nuclear data bank including in SISCODES. A computational simulation of an external beam mimicking a Co-60 irradiator was prepared, reproducing a radiation window equivalent to one found in a lateral irradiation of a nasopharyngeal tumour in situ. Absorbed dose evaluation in the internal regions of the phantom are presented and demonstrated through spatial dose distribution, superimposed with the 2D phantom sections. The present article illustrates the possibility of generating information of the spatial dose distribution including all adjacent tissues, far from tumour, improving the radiodosimetry and creating the possibility of investigating neurological deficits, dysphagia, and speaking and hearing alterations, non evaluated in the present stage of radiation therapy of head and neck. 1. INTRODUCTION According to the National Institute of Cancer INCA, the cancer is among of the main reasons of illness death in the world. Its occurrence is the second among the illness that produces death in Brazil (close to 10.86%) [1]. The majority of the head and neck tumors occurs in superior aerodigestive region, mainly in the mouth, pharynx and larynx, and rare in the nasal pits and in the sinus paranasal. The aerodigestive s tumours in non developed countries are outmoded by uterus lap tumours in terms of incidence. The cancers of mouth and pharynx occupy the third position in non developed countries and the eighth position in developed ones in terms of incidence [2]. The radiotherapy is a modality used in the treatment of the cancer in which ionizing radiation is applied to the destruction of the tumor, or acts simply as obstacle to the growth of the tumor [3-7]. However, much has been questioned about the deleterious effect (sequels) caused by the radiotherapic treatment in adjacent tissues. For generating an optimization of the treatments in radiotherapy for head and neck cancer, a computational phantom was developed, using the SISCODES tools. The main goals are the evaluations of the absorbed dose and the probability of the deleterious

effects in neighboring tissues. It can be done through a three-dimensional planning in radiotherapy. 2. SISCODES System of Codes for Absorbed Dose Calculation by Stochastic Method The SISCODES is a computational system for a 3D planning in radiotherapy that plugged data in the Monte Carlo code to execute nuclear particle transport in heterogeneous geometry. This system was developed for calculating the absorbed dose and also for simulating radiotherapy protocols. It is a system with user-friendly interface that optimizes the planning of various modalities of radiotherapic treatment. The SISCODES uses the nuclear code MCNP-5 (Monte Carlo N-Particle Transport Code, version 5) for the calculation of the nuclear particle transport [5-]. Such system uses the model of 3D computational planning defined by the ICRU-50 (8). The SISCODES shows with accuracy the dose absorbed in the tumor and organs and neighboring tissues, taking in consideration the tissue heterogeneities and the presence of emptiness. Thus, the computational simulation becomes the most efficient way of predicting the doses applied in a radiotherapic treatment. Considering that the existing computational systems normally are limited by specific types of radioactive sources and equipment, besides being expensive, these tools become inaccessible for many centers of treatment of cancer. Being thus, the SISCODES becomes a method of optimization in radiotherapic planning that takes care of to all these peculiarities [9-14]. 3. MATERIALS AND METHODS The present work was divided in 4 stages: the choice of the anatomical region, the choice of images taken from the computerized tomography (CT), the treatment of the images, and the development of the tissue voxel model. 3.1. Choice of the anatomical region The main focus is the study of the larynx and pharynx cancers and the effect of the radiotherapy in these structures. Therefore, the region, enclosing down the lungs up to the brain, was selected for the development of computational phantom of the head and neck. 3.2. Choice of the Computerized Tomography (CT) Images 51 cuts of CT had been used involving in sequence the nasal-pharynx until the apex of the lungs. These CT images were taken from a non identified adult male, 45 years old. This profile is relevant because the male sex and the 40-50 age range are the most relevant factors for incidences in the larynx and pharynx tumors. These axial images of head and neck, taken at the 3mm intervals, had been kept in a data base. Figure 1 shows some of the digitalized CT images used for the development of the phantom.

Figure 1. Two arbitrary CT sections used for the development of computational phantom. 3.3. Image manipulation All the 51 CT images had been transferred from the X-ray film to a computational file, through direct digitalization and image treatment. Figure 1 illustrates some of those images. The images in gray tones had been converted to the American National Standard Code for Information Interchange and edited into the Editplus program. A digital image, in mesh format, was constructed in which the number of subdivisions was equal to the number of voxels of the model. This image (mesh) was overlapped to the digital CT images. Observing each individual volume or voxel, a referring code was adopted that better represented the tissue position. The code number was digitalized on the voxel. This code was represented by a number, in which the structures were differentiated through the various grays tones, having thus specified different tissues (organs and structures). This procedure was repeated for all the treated CT sections. Air was included to make easy the differentiations of the structures. The model was constructed based on the the codes on the tissue table, forming a well defined three-dimensional mesh of tissues. 3.4. Development of the model of voxels - SISCODES The development of the voxel model of the head and neck was divided in 3 stages: Stage I - generation of the model of voxels in grays tones, Stage II - generation of the voxel model of of tissues, and Stage III - generating the ROIs (Region of Interest) in the voxel model of the tissues, in which calculations will take place. 3.4.1. Voxel model generation in gray level The CT images of the subject had been converted into a voxel model based in the tonality of these images, in grays tones. The CT images, previously treated, had been loaded in the SISCODES interface "trata_imagem". The images had been manipulated condensing the gray level information in each voxel in a image with less resolution. The gray level voxel model is represented in the grays tones. The flow of data is as follow: the CT image files are the inputs and the gray tone model is the output.

3.4.2. Tissue voxel model generation The gray level model was converted into a voxel model of organs and tissues. The voxel model of grays tones is loaded by a SISCODES interface to be manipulated. In this stage, the reconstructions of the voxel model are shown in three views (plane X, plane Y and plane Z). As an input file, the voxel model of gray tone is filled with data generated by the tissue voxel model. From there, two archives of configuration are used: "config.csv" and "tecidos.csv". In the archive "config.csv" is stored a small portion of the database on the server, from which the available information of the tissues are imported, together with its corresponding codes and colors. "tecidos.csv" stores the codes and colors of tissues used in the tissue voxel model. Information on these archives helps the user to specify the tissue colors that better corresponds to the gray level voxel on its position. As output file, the two models of voxels (tissue and grays level) are placed together, in the same format. This archive represents the voxel model of tissues, which two sections are depicted at Fig.2. Thus, the tissue voxel model is exported to a file whose syntax is compatible with the one equivalent to the input file of the MCNP-5 code. The voxel model of tissue is then plugged into the MCNP-5 in which the simulation of the radiotherapic treatment cases will be executed. The tissue voxel model had access through an executable file that is running in graphical X environment for Linux. 3.4.3. Selecting the ROIs in the tissue voxel model The delineation of the regions of interest (ROI) for the radiotherapic treatment in the voxel model of organs and tissues had been prepared in this case selecting the muscle tissue. The reconstructions of the voxels model are seen in 3 plans (plane X, plane Y and plane Z). It is in the plane Z that is the ROIs marks, in which points in the plane XY belong to the ROI were selected. The selection of a unique point into a volume selects this volume, through the linking of each neighboring similar voxel in a 3D fashion. This subsystem of ROI delineation operates two files: an input file that has the tissue voxel model and the output which has the ROI region. This subsystem reads three file for configuration: "config.csv", "tecidos.csv" and "rois.csv". In the archive "config.csv" is stored the data of accessing the data base of the server, in which the available tissues are imported, as well as its corresponding codes and colors. In the file "tecidos.csv" and "rois.csv" are stored the codes and colors of tissues and ROIS, respectively. From there, after the system imports the tissues and ROIs of the data base, it can modified the colors of these tissues and ROIs in way that it can adjust themselves to a real necessity, being able to function independently of the connection with the server. Figure 2. Two images of final computational phantom - SISCODES.

3.5. Transference of the model for MCNP-5 (Monte Carlo N-Particle Transport Code, version 5) The three-dimensional mesh was transferred to MCNP-5, plugging it on the input data of the MCNP-5. The voxel model has been recorded and then they had been identified by the program and congregated into the code. After this procedure, a fine adjustment of alignment of the tissues images was made. Figure 3 illustrates two section of the voxel model on MCNP5. A specific color was attributed to each tissue, to facilitate the visibility of the computer phantom, in relation the desired structure. Figure 3. Two images of final computational phantom MCNP-5. 3.6 A simulating case Two portals were applied in opposite-side on the pharynx and larynx region, in the x- direction at 5x5 cm field. A polyenergetic spectrum from a 6MV megavoltage irradiator was assumed in mono directional to the normal to each portal toward the neck. 4.1. Voxel model results 4. RESULTS The developed voxel model has 73x69x52 elements. Each voxel has 3x3x3mm. This phantom was developed in the SISCODES based on 51 images of computerized CT that holds the nasal-pharynx until the apex of the lungs, keeping all the superior digestiverespiratory system. The anatomy s knowledge is required to fill the voxel model. The time consuming to develop the model of voxel for a person with good anatomic knowledge was about 10 min per section of the head and neck voxel model. Thus, the working period of 8h had been enough to identify each tissue or organ promptly.

4.2. Simulations results Figure 4 A set of images from the nasopharyngeal taken from the interface of the SISCODES, which provides the profile of absorbed dose on the region related to the clinical case running. Figure 5 Section of the voxel model in which the maximum absorbed dose occurs.

Due to the fact that the number of particles running on the example case was not enough to find an accurate result of the absorbed dose, the absolute values of dose on the region simulated will be not presented at this moment. 5. CONCLUSION The present paper addresses a voxel model involving the pharynx and larynx for radiotherapy planning. It showed how it was built and how it be used on an radiation therapy planning. A simulation involving two portals has been applied and preliminary results are shown. The numerical results were not conclusive since the uncertain on the voxels kept larger than 30%. A running with large number of particle shall be done soon. The developed voxel model helps the understanding of the structures of the head and neck. It also will help in studying the deleterious effect of swallowing in radiotherapy protocols. ACKNOWLEDGMENTS My gratefulness to the Professor Tarcísio Passos Ribeiro Campos for the orientation. We thanks Bruno Trindade for the work in team. The authors are thankful to CNPq and CAPES due to the grant and scholarship support. REFERENCES 1. Falando de câncer se seus fatores de risco, http://www.inca.org.br (2006) 2. Câncer de cabeça e pescoço, http://www.hcanc.org.br (2006) 3. Trindade, B.M.; Campos, T.P.R., Sistema de construção de modelos de voxel a partir de imagens de CT ou MR para simulação de tratamento radioterápico, International Nuclear Atlantic Conference INAC 2005, Santos SP, 2005. 4. Trindade, B.M.; Campos, T.P.R., Modelo antropomórfico e antropométrico de órgãos e tecidos em voxels 3D, 3º Congresso Latino Americano de Órgãos Artificiais e Biomateriais COLAOB, Campinas SP, 2004. 5. Trindade, B.M.; Campos, T.P.R., Sistema para planejamento computacional 3D em radioterapia baseado em método determinístico e estocástico, XXIV Iberian Latin America Congress on Computational Methods in Engineering CILANCE, Ouro Preto - MG, 2003. 6. Trindade, B.M.; Campos, T.P.R., Sistema computacional para planejamento 3D em braquiterapia, Congresso Regional sobre Seguridad Radiológica y Nuclear IRPA, Lima Peru, 2003. 7. Trindade, B.M., Desenvolvimento de sistema computacional para dosimetria em radioterapia por nêutrons e fótons baseado em método estocástico - SISCODES, Dissertação de Mestrado, Universidade federal de Minas Gerais, Belo Horizonte - MG, Brasil (2004). 8. ICRU 50, The International Commission on Radiation Units and Measurements Report 50. Prescribing recording and reporting photon beam therapy. Maryland: Bethesda, 1993. 9. Maia, D. F., Estudo das alterações auditivas em radioterapia de cabeça e pescoço, Dissertação de Mestrado, Universidade federal de Minas Gerais, Belo Horizonte - MG, Brasil (2006).

10. Thompson, L., Desenvolvimento de um fantoma antropomórfico e antropométrico de cabeça e pescoço infanto-juvenil e de um fantoma computacional para estudo radiodosimétrico em câncer de laringe e faringe, Dissertação de Mestrado, Universidade Federal de Minas Gerais, Belo Horizonte MG, Brasil (2004). 11. Maia, M., Fantoma antropomórfico e antropométrico de tórax para fins de radioproteção e dosimetria, Dissertação de Mestrado, Universidade federal de Minas Gerais, Belo Horizonte - MG, Brasil (2004). 12. Costa, I.T.; Campos, T.P.R., Resposta radiodosimétrica de implantes de sementes de biovidros radioativos no cérebro de coelhos, 4º Congresso Latino Americano de Órgãos Artificiais e Biomateriais COLAOB, Caxambu MG, 2006. 13. Duarte, I.L.; Campos, T.P.R., Avaliação dosimétrica clínica de implantes de biovidros incorporados SM-153 para tratamento de tumores cerebrais, 4º Congresso Latino Americano de Órgãos Artificiais e Biomateriais COLAOB, Caxambu MG, 2006. 14. Lopes, J.P.A.; Campos, T.P.R., Avaliação da energia específica absorvida gerada de biovidros radioativos implantados em fígado em modelo animal, 4º Congresso Latino Americano de Órgãos Artificiais e Biomateriais COLAOB, Caxambu MG, 2006.