Obtenção de volumes tumorais em pilhas de imagens PET usando técnicas de segmentação de imagens coloridas

Abstract

This work is demonstrated the step by step how to segment color images on the thoracic region of an adult in order to separate the tumor volume at present same conditions without significantly changing the values of R components (Red), G (Green) and B (Blue) color of the pixels . For information allowing build color map you need to segment and classify at appropriate intervals the colors present in the images. The used segmentation technique is to select a small rectangle with color samples of a given region and then erase with a specific color called "rubber" the other regions of the image. The tumor region was segmented into one of the images available and the procedure is shown in the tutorial format. All necessary computational tools have been implemented in DIP (Digital Image Processing), software developed by the authors. The results obtained besides permit the construction of the color map of the distribution of activity in PET images concentration will also be useful in future studies to insert tumors in voxel phantoms for the purpose of carrying out dosimetry evaluations.

Palavras-chave: segmentação, imagens PET, processamento de imagens