Development of methods for MRI longitudinal studies. (September 2016)

• Record Type: Journal Article
• Authors: Grucka, A.; Sensi, Francesco; Brun, Francesco; Ukmar, M.; Chincarini, A.; Longo, R.
• Is Part Of: Physica medica. Volume 32(2016)Supplement 3
• Page Start: 310
• Publication Date: 2016-09
• Subjects: Medical physics -- Periodicals; Biophysics -- Periodicals; Biophysics -- Periodicals; Imagerie médicale -- Périodiques; Radiothérapie -- Périodiques; Rayons X -- Sécurité -- Mesures -- Périodiques; Physique -- Périodiques; Médecine -- Périodiques; 610.153
• DOI: 10.1016/j.ejmp.2016.07.175
• Languages: English
• ISSNs: 1120-1797

Abstract

Abstract : Introduction: MRI longitudinal studies have been proposed to detect abnormal brain changes in the early phase of the Alzheimer's disease (AD) during its asymptomatic phase. Purpose: In order to obtain a very high sensitivity in MRI longitudinal studies a number of technical issues have to be addressed. Materials and methods: In the present study a high resolution MRI protocol focused on the medial temporal lobe mainly conceived for 1.5T MRI device is developed. A grey-level reference object was designed to be used during acquisition to help image post-processing when gray-level comparison is required. A prospective study on healthy volunteers is ongoing. An image analysis protocol for longitudinal studies has been developed, including pre-processing (gray level normalization, image registration) and voxel-based time-course analysis. The protocol has been tested on the MIRIAD dataset. Results: The new acquisition and gray-level normalization technique was applied and comparison indexes (PSNR, SSI, NCC) have been calculated to demonstrate the increased reproducibility of the images. The image processing developed for this study demonstrates that the time-course of brain atrophy let us differentiate normal and AD subjects with repeated images acquired in less than 2 years. Conclusion: Applying our imaging techniques, longitudinal studies will likely be more accurate than the study based on MIRIAD dataset. Therefore we expect that repeated images with our acquisition and normalization protocol will allow the discrimination between normal and AD subjects with a reduced number of images acquired in a period of about one year. Disclosure: Nothing to declare.