Identifying subtypes of mild cognitive impairment from healthy aging based on multiple cortical features combined with volumetric measurements of the hippocampal subfields.

BACKGROUND: Mild cognitive impairment (MCI) is subtle cognitive decline with an estimated 10-15% yearly conversion rate toward Alzheimer's disease (AD). It remains unexplored in brain cortical association areas in different lobes and its changes with progression and conversion of MCI. METHODS: Brain structural magnetic resonance (MR) images were collected from 102 stable MCI (sMCI) patients. One hundred eleven were converted MCI (cMCI) patients, and 109 were normal control (NC). The cortical surface features and volumes of subcortical hippocampal subfields were calculated using the FreeSurfer software, followed by an analysis of variance (ANOVA) model, to reveal the differences between the NC-sMCI, NC-cMCI, and sMCI-cMCI groups. Afterward, the support vector machine-recursive feature elimination (SVM-RFE) method was applied to determine the differences between the groups. RESULTS: The experimental results showed that there were progressive degradations in either range or degree of the brain structure from NC to sMCI, and then to cMCI. The SVM classifier obtained accuracies with 64.62%, 78.96%, and 70.33% in the sMCI-NC, cMCI-NC, and cMCI-sMCI groups, respectively, using the volumes of hippocampal subfields independently. The combination of the volumes from the hippocampal subfields and cortical measurements could significantly increase the performance to 71.86%, 84.64%, and 76.86% for the sMCI-NC, cMCI-NC, and cMCI-sMCI classifications, respectively. Also, the brain regions corresponding to the dominant features with strong discriminative power were widely located in the temporal, frontal, parietal, olfactory cortexes, and most of the hippocampal subfields, which were associated with cognitive decline, memory impairment, spatial navigation, and attention control. CONCLUSIONS: The combination of cortical features with the volumes of hippocampal subfields could supply critical information for MCI detection and its conversion.

Discrimination and conversion prediction of mild cognitive impairment using convolutional neural networks.

BACKGROUND: Recently, studies have demonstrated that machine learning techniques, particularly cutting-edge deep learning technology, have achieved significant progression on the classification of Alzheimer's disease (AD) and its prodromal phase, mild cognitive impairment (MCI). Moreover, accurate prediction of the progress and the conversion risk from MCI to probable AD has been of great importance in clinical application. METHODS: In this study, the baseline MR images and follow-up information during 3 years of 150 normal controls (NC), 150 patients with stable MCI (sMCI) and 157 converted MCI (cMCI) were collected from the Alzheimer's Disease Neuroimaging Initiative (ADNI). The deep convolutional neural networks (CNNs) were adopted to distinguish different stages of MCI from the NC group, and predict the conversion time from MCI to AD. Two CNN architectures including GoogleNet and CaffeNet were explored and evaluated in multiple classifications and estimations of conversion risk using transfer learning from pre-trained ImageNet (via fine-tuning) and five-fold cross-validation. A novel data augmentation approach using random views aggregation was applied to generate abundant image patches from the original MR scans. RESULTS: The GoogleNet acquired accuracies with 97.58%, 67.33% and 84.71% in three-way discrimination among the NC, sMCI and cMCI groups respectively, whereas the CaffeNet obtained promising accuracies of 98.71%, 72.04% and 92.35% in the NC, sMCI and cMCI classifications. Furthermore, the accuracy measures of conversion risk of patients with cMCI ranged from 71.25% to 83.25% in different time points using GoogleNet, whereas the CaffeNet achieved remarkable accuracy measures from 95.42% to 97.01% in conversion risk prediction. CONCLUSIONS: The experimental results demonstrated that the proposed methods had prominent capability in classification among the 3 groups such as sMCI, cMCI and NC, and exhibited significant ability in conversion risk prediction of patients with MCI.

Corrigendum: Conversion Discriminative Analysis on Mild Cognitive Impairment Using Multiple Cortical Features from MR Images.

[This corrects the article on p. 146 in vol. 9, PMID: 28572766.].

Conversion Discriminative Analysis on Mild Cognitive Impairment Using Multiple Cortical Features from MR Images.

Neuroimaging measurements derived from magnetic resonance imaging provide important information required for detecting changes related to the progression of mild cognitive impairment (MCI). Cortical features and changes play a crucial role in revealing unique anatomical patterns of brain regions, and further differentiate MCI patients from normal states. Four cortical features, namely, gray matter volume, cortical thickness, surface area, and mean curvature, were explored for discriminative analysis among three groups including the stable MCI (sMCI), the converted MCI (cMCI), and the normal control (NC) groups. In this study, 158 subjects (72 NC, 46 sMCI, and 40 cMCI) were selected from the Alzheimer's Disease Neuroimaging Initiative. A sparse-constrained regression model based on the l2-1-norm was introduced to reduce the feature dimensionality and retrieve essential features for the discrimination of the three groups by using a support vector machine (SVM). An optimized strategy of feature addition based on the weight of each feature was adopted for the SVM classifier in order to achieve the best classification performance. The baseline cortical features combined with the longitudinal measurements for 2 years of follow-up data yielded prominent classification results. In particular, the cortical thickness produced a classification with 98.84% accuracy, 97.5% sensitivity, and 100% specificity for the sMCI-cMCI comparison; 92.37% accuracy, 84.78% sensitivity, and 97.22% specificity for the cMCI-NC comparison; and 93.75% accuracy, 92.5% sensitivity, and 94.44% specificity for the sMCI-NC comparison. The best performances obtained by the SVM classifier using the essential features were 5-40% more than those using all of the retained features. The feasibility of the cortical features for the recognition of anatomical patterns was certified; thus, the proposed method has the potential to improve the clinical diagnosis of sub-types of MCI and predict the risk of its conversion to Alzheimer's disease.

[Tract-based spatial statistics analysis on the white matter of patients with temporal lobe epilepsy and automatic recognition].

This study aims to determine the salient brain regions with abnormal changes in white matter structures from diffusion tensor imaging (DTI) images of the patients with temporal lobe epilepsy (TLE), and to discriminate the patients with TLE from normal controls (NCs). Firstly, the DTI images from 50 subjects (28 NCs and 22 TLE) were acquired. Secondly, the four measures including the fractional anisotropy (FA), the mean diffusivity (MD), the axial diffusivity (AD) and the radial diffusivity (RD) were calculated. Thirdly, the tract-based spatial statistics (TBSS) was adopted to extract the measures in brain regions with significant differences between the two compared groups. Fourthly, the obtained measures were used as input features of the support vector machine (SVM) for classification, and the support vector machine-recursive feature elimination (SVM-RFE) was compared with the support vector machine-tract-based spatial statistics (SVM-TBSS) method. Finally, the essential brain regions and their spatial distribution were analyzed and discussed. The experimental results showed that the FA measures of the TLE group decreased significantly in the corpus callosum, superior longitudinal fasciculus, corona radiata, external capsule, internal capsule, inferior fronto-occipital fasciculus, fasciculus uncinatus and sagittal stratum, which were nearly bilaterally distributed, while the MD and RD increased significantly in most of these brain regions of the TLE group. Although the AD also increased, the differences were not statistically significant. The SVM-TBSS classifier obtained accuracies of 82%, 76% and 76% using the FA, MD and RD for classification, respectively, and 80% using combined measures. The SVM-RFE classifier obtained accuracies of 90%, 90% and 92% using the FA, MD and RD respectively, while the highest accuracy was 100% using combined measures. These results demonstrated that the SVM-RFE outperformed the SVM-TBSS, and the dominant characteristic influencing classification in brain regions were in associative and commissural fibers. These results illustrated that the measures of DTI images could reveal the abnormal changes in white matter structure of patients with TLE, providing effective information to clarify its pathological mechanism, localize the focus and diagnose automatically.