Assessment of detection accuracy of changes in thickness of human brain cortex according to MRI data

Abstract

Purpose. The study of the stability of the results of MRI data processing at different noise levels, parameters of the scanning protocol and the position of the patient's head. Data processing was carried out by voxel-based and surface morphometry methods. Materials and methods. Data from six healthy volunteers were measured using various scanning protocols on a 1.5T Siemens Symphony clinical scanner with an isotropic resolution of 1 mm3. The measurement of the of the cortical thickness was carried out using two software packages - "CAT12" and "FreeSurfer". For an adequate comparison of the methods, all preliminary data processing was carried out only using standard software package functions. Results. The obtained values of the cortical layer thickness decreased by 5% with an increase of the noise level by 60% (signal-to-noise ratio 15), and did not depend on the used parameters of the scanning protocol and the position of the patient's head. All changes in the values of the cortical layer thickness were within the error limits of the data processing methods. Conclusion. Voxel-based and surface morphometry methods implemented in the «CAT12» and «FreeSurfer» software packages allow obtaining stable results regardless of the quality of the initial data.