Integració a MATLAB de la visualització volumètrica mitjançant l'algorisme de Ray Casting basat en GPU

Abstract

Nuclear medicine has been a great advance for medicine in general. Thanks to this discipline is possible accurately detect pathological areas in different parts of the body using devices such as scanners, RMI­PETs, SPECT or any other device that uses radiation. This become a great benefit for diagnose, detection and disease prevention. To facilitate the task of nuclear hysicians, in this project a software application will be developed to able to manage RMI­PET medical volumes and to deal with a graphic interface that renders the volume with distinct pathological areas. Moreover, it will allow to interact with the volume modifying several parameters such as the view position, zoom areas, contrast parameters... etc, in order to enhance the professionals' experience where they will be able to classify the patient's condition based on the results. One of the biggest problems in MATLAB are the slow speed in the operations with hugh matrices. Consequently, the display of volumes is slow and tedious. To speed and to solve this problem, it is necessary an external rendering algorithm capable to increase image processing in MATLAB. RayCasting's algorithm is one of the best algorithms for image processing due to its quality. Moreover, the use of the paralelism that offers recent GPUs allows to process graphics speedier than CPU would do. Along this project we will see how to integrate in this application an external Ray Casting algorithm based on GPU which will be capable of improving the problem of slow display. Furthermore, it will be analyzed and discussed in detail with real datasets and we present the obtained results, improvements and developed implementation.