Aportaciones del tratamiento de imágenes a la dosimetría en radioterapia

Laburpena

Modern radiotherapy depends on digital imaging to acquire exact anatomic information, so there is an increasing need of having quality control systems and procedures to assure image fidelity. In this dissertation, a set of digital image processing techniques developed to solve real clinical problems in the Radiation Oncology Dept., Hospital Clinico Universitario. Valencia. Spain, is presented. The main goal of this work is to show that image processing is of dosimetric interest in all stages of the treatment and helps to improve treatment quality. Some basic radiation physic principles help to understand that geometric quality of the images are of more importance than visual quality. Several imaging tools have been developed for this work. These include a complete medical imaging software, a tomographic reconstruction system for radiotherapy simulators and a videodensitometer. It has been developed a method for the tomographic reconstruction of big volumes using partial slices, allowing the scanning of patients up to 60 cm diameter, together with an algorithm to select and locate the minimum number of tomographic slices required to perform a true 3D dosimetry and evaluation of structure volumes. Following this tomographic acquisition enhancement, work has been done to manage physiological movement in radiotherapy treatments. Using the fluoroscopy facility of the radiotherapy simulation, a procedure has been invented to automatically analyze an evaluate the extent of organ motion within the patient due to respiration. Using an algorithm based on correlation with adapted filters, the frequency and amplitude of a pseudo periodic movement is detected and measured. Different image resolutions, time resolutions and adapted filters have been tested until clinically acceptable processing times and results have been achieved. The procedure has also been clinically tested. Finally, image comparison techniques have been implemented. By means of a spatial warping algorithm, images form different imaging modalities (CT, MR, DR, CR, Portal Vision etc.) can be objectively compared and fused if needed.