Abstract
Since 2004, the Research Group Dosimetry Computing and Embedded Systems (GPDC & SE) publish works using objects simulators (computational representations of the human body) based on anatomical and physiological information provided mainly by the ICRP (International Commission on Radiological Protection). However, such anthropomorphic models are "healthy", ie have no pathologies (eg tumor) in their organs and tissues. A possible solution to tumorize these phantoms in a realistic way is to use PET imaging (positron emission tomography), since these functional images enable detection of tumor regions in the human body. For purpose of diagnosis, images obtained are artificially colored so that each color represents the concentration of radioactive material activity metabolised in the region in question (Bq / ml). To separate the tumor from the other information contained in PET imaging is necessary to make a color image segmentation. The targeting criteria used in this study is based on the similarity between the color intensities in RGB space using a mathematical rule for defining the nearest neighbors. From 20 images available from a chest of an adult tumor was performed targeting with the DIP software (Digital Image Processing)..