Development of a computer-aided diagnostic system for Alzheimer's disease using magnetic resonance imaging

TL;DR

This study developed a computer-aided diagnostic system for Alzheimer's disease using MRI, focusing on automatic volumetry and cortical thickness analysis, demonstrating robustness against magnetic field inhomogeneity and potential for early diagnosis.

Contribution

The paper introduces a novel MRI-based diagnostic system utilizing automated volumetry and PDE-based cortical thickness measurement for Alzheimer's detection.

Findings

Robustness of volumetry against magnetic field inhomogeneity

Maintained accuracy of volumetry with low to moderate noise levels

Clear demonstration of cortical thinning in clinical AD patients

Abstract

Alzheimer's disease (AD) is the most common type of dementia accompanied with brain atrophy. Structural measurements of brain atrophy in specific brain structures such as hippocampus using magnetic resonance imaging (MRI) have been reported to detect the development of dementia early in the course of the disease. The purpose of this study was to develop a computer-aided diagnostic system for AD using MRI, which is based on the automatic volumetry of segmented brain images and generation of three-dimensional cortical thickness images using the Eulerian partial differential equation (PDE) approach. We investigated the effect of the inhomogeneity of magnetic field strength and statistical noise on the accuracy of our automatic volumetry and the PDE approach using the simulated MR images generated from BrainWeb. Our automatic volumetry and PDE approach were robust against inhomogeneous magnetic field strength. Although the accuracy of our automatic volumetry decreased with increasing statistical noise, it was maintained when the statistical noise was less than 7-8%. The cortical thickness obtained by the PDE method tended to decrease with increasing statistical noise. When we applied our method to clinical data, the cortical thinning due to brain atrophy was clearly demonstrated in patients with brain atrophy. These results suggest that our system appears to be useful for the diagnosis of AD, because it allows us to automatically evaluate the extent of brain atrophy and cortical thinning in a three-dimensional manner.