Kirjahaku

Brain tumor is one of the most life threatening diseases and hence its detection should be fast and accurate. This can be achieved by the execution of automated tumor detection techniques on medical images. Many automated techniques which are being used for image segmentation have been proposed. Here we propose an automated and efficient brain tumor detection technique implementing on Positron Emission Tomography (PET) images. Simulation of the proposed work is done in MATLAB. The brain tumor segmentation on Positron Emission Tomography (PET) images is very difficult and important task for medical diagnosis. This thesis describes the processes and techniques to detect the brain tumor from PET images using ANN (Artificial Neural Network) which is applied most of the artificial intelligence in biomedical image for classification and recognition. In the proposed system, at first pre-processing and post- processing of PET images is performed to enhance it then the processed image is being more suitable to analysis and classifies the tumor images. Here sobel edge detection is used to segment the PET images. In the second stage, statistical feature analysis is extracted from PET images.

Edges define the boundaries between regions in an image, which helps with segmentation and object recognition. They can show where shadows fall in an image or any other distinct change in the intensity of an image. Edges in an image are formed due to variations of some of the physical properties, like surface illumination, shadows, geometry and reflectance of objects in scene. Clear edges can show where shadows fall in an image, any other distinct change in the intensity of an image. The process of extraction of these feature points is called edge detection. This feature plays an important role in object identification methods used in machine vision, image segmentation, 3D reconstruction and image processing systems of an edge. Morphological edge detectors involve simple addition/subtraction operations and max/min operations. Since different edge detectors work better under different conditions, it would be ideal to have an algorithm that makes use of multiple edge detectors, applying each one when the scene conditions are most ideal for its method of detection. In order to create this system, it is first required to know which edge detectors perform better under which conditions.