Análise comparativa e qualitativa de imagens geradas por exames de tomografia computadorizada e ressonância magnética utilizando etapas de pré-processamento digital de imagens

Abstract

Medical imaging exams are the basis of modern medicine, in which body structures can be analyzed more succinctly without the need to apply invasive methods to patients. In more complex diagnoses that require continuous and in-depth analyses, the patient's work in performing continuous imaging exams can hinder the achievement of results over time, which in certain cases is crucial for the patient's life. From this, this project aimed to study the generation and reproduction of medical images from computed tomography and magnetic resonance machines, collecting data from cranial exams and applying digital pre-processing steps, through the use of a Gaussian filter, filter de Laplace, Sobel operator and anisotropic diffusion filter. To obtain the medical images, a database of patient images in Dicom format was used, which is the standard format for this type of image. The program to process the images was developed in Python language, using Spyder software. To support this project, an in-depth research was carried out on computed tomography and magnetic resonance equipment, acquisition and manipulation of medical images, digital image processing and comparative methods of image quality. The project, therefore, is based on the acquisition of images from examinations of the skull area coming from both machines, reproducing them through the Python programming language with the support of specific libraries such as OpenCV, Pydicom Skimage and Medpy, where they are pre- processed using filtering methods and computational adjustments, so that their results can be compared qualitatively and computationally. After the tests with the codes finalized, it was possible to conclude that the images from the resonance machine produced clearer and more detailed results, despite the program's processing time being higher.