Multi-Method Analysis of MRI Images in Early Diagnostics of Alzheimer's Disease

Abstract

The role of structural brain magnetic resonance imaging (MRI) is becoming more and more emphasized in the early diagnostics of Alzheimer's disease (AD). This study aimed to assess the improvement in classification accuracy that can be achieved by combining features from different structural MRI analysis techniques. Automatically estimated MR features used are hippocampal volume, tensor-based morphometry, cortical thickness and a novel technique based on manifold learning. Baseline MRIs acquired from all 834 subjects (231 healthy controls (HC), 238 stable mild cognitive impairment (S-MCI), 167 MCI to AD progressors (P-MCI), 198 AD) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database were used for evaluation. We compared the classification accuracy achieved with linear discriminant analysis (LDA) and support vector machines (SVM). The best results achieved with individual features are 90% sensitivity and 84% specificity (HC/AD classification), 64%/66% (S-MCI/P-MCI) and 82%/76% (HC/P-MCI) with the LDA classifier. The combination of all features improved these results to 93% sensitivity and 85% specificity (HC/AD), 67%/69% (S-MCI/P-MCI) and 86%/82% (HC/P-MCI). Compared with previously published results in the ADNI database using individual MR-based features, the presented results show that a comprehensive analysis of MRI images combining multiple features improves classification accuracy and predictive power in detecting early AD. The most stable and reliable classification was achieved when combining all available features.

On the early diagnosis of Alzheimer's Disease from multimodal signals: A survey

“…Yepes-Calderón et al [105] achieved 87.3% of accuracy distinguishing AD and MCI cohorts and 90.64% in the case of MCI and control. MCI and control subjects have also been distinguished in other studies with accuracy rates of 85.4% [28], 84% [56], 81.3% [106] and 78.22% [26].…”

“…Recently, Yepes-Calderón et al [105] have developed a relatively simple classification system to distinguish between AD, MCI and control patients with MRI and they achieved classifications accuracies of 98.95% when distinguishing AD from control patients. Others have reported accuracies of 92% [30], 89.22% [112], 89% [56], 88.9% [106], 88.49% [28], 87% [26] and 83% [113]. Farzan et al [114] have achieved comparable results in AD diagnosis from longitudinal MRI data.…”

“…Especially the latter are being developed and improved straight off, and this is giving way to AD related physiological changes getting better identified. The volumetric analysis of the brain, which allows to detect atrophy, has been the main objective of imaging, and even if it has been done manually for many years, nowadays, it is evolving and has started to be done automatically, thanks to techniques like the voxel-based morphometry [53,54], tensor-based morphometry [55,56], object-based morphometry [57] and feature-based morphometry [58]. Currently, "neuroimaging plays a central role in the clinical research of cognitive disorders" [59].…”

Characterization of brain anatomical patterns by comparing region intensity distributions: Applications to the description of Alzheimer's disease

“…(2011) using a different database reported up to 81% sensitivity and 95% specificity. Similar or slightly lower results were found for methods relying on tissue segmentation (Davatzikos et al., 2008; Fan, Resnick, Wu, & Davatzikos, 2008; Westman, Aguilar, Muehlboeck, & Simmons, 2013; Zhang, Wang, Zhou, Yuan, & Shen, 2011), elastic deformations (Magnin et al., 2009), semiautomatic segmentation of the hippocampus (Barnes et al., 2004), or combinations of one or more of them (Farhan, Fahiem, & Tauseef, 2014; Kloppel et al., 2008; Plant et al., 2010; Teipel et al., 2007; Wolz et al., 2011). …”

Instantiated mixed effects modeling of Alzheimer's disease markers

“…Manifold learning features from MR images have been shown to contain valuable information about disease severity and progression [18,16,17,49]. Here, the aim is to learn the underlying low-dimensional manifold that best represents the population.…”