Correção de aquisição e automatização da reconstrução GRASP de imagens de RM

Abstract

The main objective of this work is to develop a system that allows the acquisition of magnetic resonance images of the thorax of patients with free breathing. Traditional techniques when used in patients who can not perform apnea result in high blurry images. Conventional MR imaging methods work by acquiring Cartesian- K space. In the proposed system we used the GRASP pulse sequence, Goldenangle RAdial Sparse Parallel, in conjunction with a reconstruction algorithm. The GRASP sequence acquires the images radially continuously, incremented by the radial golden angle constant. Subsequently, in the reconstruction process the respiratory movement is tracked, the separation of the spokes stars in respiratory stages and the reduction of subamplification artifacts using Compressed Senssing. After the acquisition of the RM data by the GRASP method, it is essential to optimize the acquired data to obtain adequate results. To perform this correction we reconstructed several exams of different patients, varying only this parameter in order to carry out more in depth analyzes of the attributes of these images. From these results, we developed a method that is based on the variance of the pixels of the images as a figure of merit in order to obtain the most reliable images, presenting totally satisfactory results. A total of 174 computerized tomography and magnetic resonance imaging examinations of 29 patients were performed. These examinations were evaluated clinically by two experienced radiologists. In the clinical evaluation, the GRASP method presented a high correlation with Computerized Tomography (reference) and quality equivalence with the conventional magnetic resonance imaging in apnea.