Voxel-based morphometry and a deep learning model for the diagnosis of early Alzheimer's disease based on cerebral gray matter changes

Abstract

This study aimed to analyse cerebral grey matter changes in mild cognitive impairment (MCI) using voxel-based morphometry and to diagnose early Alzheimer's disease using deep learning methods based on convolutional neural networks (CNNs) evaluating these changes. Participants (111 MCI, 73 normal cognition) underwent 3-T structural magnetic resonance imaging. The obtained images were assessed using voxel-based morphometry, including extraction of cerebral grey matter, analyses of statistical differences, and correlation analyses between cerebral grey matter and clinical cognitive scores in MCI. The CNN-based deep learning method was used to extract features of cerebral grey matter images. Compared to subjects with normal cognition, participants with MCI had grey matter atrophy mainly in the entorhinal cortex, frontal cortex, and bilateral frontotemporal lobes (p < 0.0001). This atrophy was significantly correlated with the decline in cognitive scores (p < 0.01). The accuracy, sensitivity, and specificity of the CNN model for identifying participants with MCI were 80.9%, 88.9%, and 75%, respectively. The area under the curve of the model was 0.891. These findings demonstrate that research based on brain morphology can provide an effective way for the clinical, non-invasive, objective evaluation and identification of early Alzheimer's disease.

Keywords: cerebral grey matter; convolutional neural network; deep learning; mild cognitive impairment; voxel-based morphometry.

Fig. 1

Flowchart depicting the VBM processing. The original T1-weighted magnetic resonance image was registered to the standard brain template for head movement correction and skull dissection. Then, GM segmentation was performed using the configured standard cerebral GM. After Gaussian smoothing, the final cerebral GM image was obtained. Abbreviations: GM, gray matter; VBM, voxel-based morphometry.

Fig. 2

The convolutional neural network architecture. Abbreviations: MCI, mild cognitive impairment; NC, normal control.

Fig. 3

Comparison of GM volumes between patients with MCI and NC participants (transverse plane). Significant changes (P < 0.0001) are indicated as colored brain areas where GM atrophy of the MCI group exceeds that of the NC group. Abbreviations: GM, gray matter; MCI, mild cognitive impairment; NC, normal control.

Fig. 4

GM volume changes related to MMSE scores in patients with MCI (transverse plane). Significant correlations (P < 0.01) between MCI atrophy and MMSE score are indicated as colored brain areas. Abbreviations: GM, gray matter; MCI, mild cognitive impairment; MMSE, mini-mental state examination.

Fig. 5

The solid lines represented the accuracy of the CNN-based deep learning model for training and validation sets. The dotted line in the background represented the curve fitted by the model during the training and validation data set. Abbreviations: CNN, convolutional neural network.

Fig. 6

ROC curve of the deep learning model. The solid line represents the ROC curve of the test data set. Abbreviations: AUC, area under the curve; CNN, convolutional neural network; ROC, receiver operating characteristics.