Brain MR image classification for Alzheimer's disease diagnosis using structural hippocampal asymmetrical attributes from directional 3-D log-Gabor filter responses

Author(s)

Poloni, Katia M. ^[1] ; de Oliveira, Italo A. Duarte ^[1] ; Tam, Roger ^[2] ; Ferrari, Ricardo J. ^[1] ; Initi, Alzheimers Dis Neuroimaging

Affiliation

^[1] Univ Fed Sao Carlos, Dept Comp, Rod Washington Luis, Km 235, BR-13565905 Sao Carlos, SP - Brazil

^[2] Univ British Columbia, Sch Biomed Engn, Dept Radiol, Djavad Mowafaghian Ctr Brain Hlth, 2215 Wesbrook Mall, Vancouver, BC V6T 2B5 - Canada

• Document type: Journal article
• Source: Neurocomputing; v. 419, p. 126-135, JAN 2 2021.
• Web of Science Citations: 1

Abstract

Alzheimer's disease (AD) is a progressive and irreversible neurodegenerative condition whose development is characterized by lateralized brain atrophies. In AD, the hippocampus is the first brain structure to present atrophy, which, although to a lesser extent, is also a precursor to the broader asymmetrical development of the human brain. Structural magnetic resonance (MR) imaging is capable of detecting the disease-induced anatomical changes in the brain, thus aiding the diagnosis of AD. MR image attributes extracted from the hippocampal regions are commonly used for the AD classification task. However, most of the published methods do not explore hippocampal asymmetries for image classification. In this study, we propose a new technique for performing the classification of MR images for AD using only hippocampal asymmetrical attributes. By using the new proposed asymmetry index (AI), we assessed the attributes and the ones that passed the analysis of variance test, i.e., showing statistically mean differences among the classes (CN, MCI, and AD), were selected for classification. As a result of our study, the statistical analysis of our AI has shown a significant increase in hippocampal asymmetry as disease progress (CN < MCI < AD). Moreover, for the classification using clinical MR images, we obtained accuracy values of 69.44% and 82.59%; and AUC values of 0.76 and 0.9 for CN x MCI and CN x AD, respectively. Last, we found the results of our asymmetry analysis consistent with other statistical assessments and our classification results, using only asymmetry attributes comparable to (or even higher than) existing hippocampus studies. (c) 2020 Published by Elsevier B.V. (AU)

• FAPESP's process: 18/09972-9 - Detection and analysis of hippocampal structural asymmetries in magnetic resonance images with application to aid in the diagnosis of Alzheimer's Disease
• Grantee: Italo Antonio Duarte de Oliveira
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• FAPESP's process: 18/06049-5 - Automatic computational scheme for the detection, identification and classification of cerebral structural changes in magnetic resonance images to aid the diagnosis of patients with mild cognitive impairment and mild Alzheimer's disease
• Grantee: Katia Maria Poloni
• Support Opportunities: Scholarships in Brazil - Doctorate
• FAPESP's process: 18/08826-9 - Development of feature engineering and deep learning techniques applied to the classification of magnetic resonance images in healthy cognitive aging, mild cognitive impairment and Alzheimer's Disease
• Grantee: Ricardo José Ferrari
• Support Opportunities: Regular Research Grants