Multi-Modal Feature Selection with Feature Correlation and Feature Structure Fusion for MCI and AD Classification

Feature selection for multiple types of data has been widely applied in mild cognitive impairment (MCI) and Alzheimer’s disease (AD) classification research. Combining multi-modal data for classification can better realize the complementarity of valuable information. In order to improve the classification performance of feature selection on multi-modal data, we propose a multi-modal feature selection algorithm using feature correlation and feature structure fusion (FC2FS). First, we construct feature correlation regularization by fusing a similarity matrix between multi-modal feature nodes. Then, based on manifold learning, we employ feature matrix fusion to construct feature structure regularization, and learn the local geometric structure of the feature nodes. Finally, the two regularizations are embedded in a multi-task learning model that introduces low-rank constraint, the multi-modal features are selected, and the final features are linearly fused and input into a support vector machine (SVM) for classification. Different controlled experiments were set to verify the validity of the proposed method, which was applied to MCI and AD classification. The accuracy of normal controls versus Alzheimer’s disease, normal controls versus late mild cognitive impairment, normal controls versus early mild cognitive impairment, and early mild cognitive impairment versus late mild cognitive impairment achieve 91.85 ± 1.42%, 85.33 ± 2.22%, 78.29 ± 2.20%, and 77.67 ± 1.65%, respectively. This method makes up for the shortcomings of the traditional multi-modal feature selection based on subjects and fully considers the relationship between feature nodes and the local geometric structure of feature space. Our study not only enhances the interpretation of feature selection but also improves the classification performance, which has certain reference values for the identification of MCI and AD.

Information Flow Pattern in Early Mild Cognitive Impairment Patients

Purpose: To investigate the brain information flow pattern in patients with early mild cognitive impairment (EMCI) and explore its potential ability of differentiation and prediction for EMCI.Methods: In this study, 49 patients with EMCI and 40 age- and sex-matched healthy controls (HCs) with available resting-state functional MRI images and neurological measures [including the neuropsychological evaluation and cerebrospinal fluid (CSF) biomarkers] were included from the Alzheimer's Disease Neuroimaging Initiative. Functional MRI measures including preferred information flow direction between brain regions and preferred information flow index of each brain region parcellated by the Atlas of Intrinsic Connectivity of Homotopic Areas (AICHA) were calculated by using non-parametric multiplicative regression-Granger causality analysis (NPMR-GCA). Edge- and node-wise Student's t-test was conducted for between-group comparison. Support vector classification was performed to differentiate EMCI from HC. The least absolute shrinkage and selection operator (lasso) regression were used to evaluate the predictive ability of information flow measures for the neurological state.Results: Compared to HC, disturbed preferred information flow directions between brain regions involving default mode network (DMN), executive control network (ECN), somatomotor network (SMN), and visual network (VN) were observed in patients with EMCI. An altered preferred information flow index in several brain regions (including the thalamus, posterior cingulate, and precentral gyrus) was also observed. Classification accuracy of 80% for differentiating patients with EMCI from HC was achieved by using the preferred information flow directions. The preferred information flow directions have a good ability to predict memory and executive function, level of amyloid β, tau protein, and phosphorylated tau protein with the high Pearson's correlation coefficients (r > 0.7) between predictive and actual neurological measures.Conclusion: Patients with EMCI were presented with a disturbed brain information flow pattern, which could help clinicians to identify patients with EMCI and assess their neurological state.

Altered Functional Connectivity of the Basal Nucleus of Meynert in Subjective Cognitive Impairment, Early Mild Cognitive Impairment, and Late Mild Cognitive Impairment

Background: The spectrum of early Alzheimer's disease (AD) is thought to include subjective cognitive impairment, early mild cognitive impairment (eMCI), and late mild cognitive impairment (lMCI). Choline dysfunction affects the early progression of AD, in which the basal nucleus of Meynert (BNM) is primarily responsible for cortical cholinergic innervation. The aims of this study were to determine the abnormal patterns of BNM-functional connectivity (BNM-FC) in the preclinical AD spectrum (SCD, eMCI, and lMCI) and further explore the relationships between these alterations and neuropsychological measures.Methods: Resting-state functional magnetic resonance imaging (rs-fMRI) was used to investigate FC based on a seed mask (BNM mask) in 28 healthy controls (HC), 30 SCD, 24 eMCI, and 25 lMCI patients. Furthermore, the relationship between altered FC and neurocognitive performance was examined by a correlation analysis. The receiver operating characteristic (ROC) curve of abnormal BNM-FC was used to specifically determine the classification ability to differentiate the early AD disease spectrum relative to HC (SCD and HC, eMCI and HC, lMCI and HC) and pairs of groups in the AD disease spectrum (eMCI and SCD, lMCI and SCD, eMCI and lMCI).Results: Compared with HC, SCD patients showed increased FC in the bilateral SMA and decreased FC in the bilateral cerebellum and middle frontal gyrus (MFG), eMCI patients showed significantly decreased FC in the bilateral precuneus, and lMCI individuals showed decreased FC in the right lingual gyrus. Compared with the SCD group, the eMCI group showed decreased FC in the right superior frontal gyrus (SFG), while the lMCI group showed decreased FC in the left middle temporal gyrus (MTG). Compared with the eMCI group, the lMCI group showed decreased FC in the right hippocampus. Interestingly, abnormal FC was associated with certain cognitive domains and functions including episodic memory, executive function, information processing speed, and visuospatial function in the disease groups. BNM-FC of SFG in distinguishing eMCI from SCD; BNM-FC of MTG in distinguishing lMCI from SCD; BNM-FC of the MTG, hippocampus, and cerebellum in distinguishing SCD from HC; and BNM-FC of the hippocampus and MFG in distinguishing eMCI from lMCI have high sensitivity and specificity.Conclusions: The abnormal BNM-FC patterns can characterize the early disease spectrum of AD (SCD, eMCI, and lMCI) and are closely related to the cognitive domains. These new and reliable findings will provide a new perspective in identifying the early disease spectrum of AD and further strengthen the role of cholinergic theory in AD.

Random forest model for feature-based Alzheimer’s disease conversion prediction from early mild cognitive impairment subjects

Alzheimer’s Disease (AD) conversion prediction from the mild cognitive impairment (MCI) stage has been a difficult challenge. This study focuses on providing an individualized MCI to AD conversion prediction using a balanced random forest model that leverages clinical data. In order to do this, 383 Early Mild Cognitive Impairment (EMCI) patients were gathered from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Of these patients, 49 would eventually convert to AD (EMCI_C), whereas the remaining 334 did not convert (EMCI_NC). All of these patients were split randomly into training and testing data sets with 95 patients reserved for testing. Nine clinical features were selected, comprised of a mix of demographic, brain volume, and cognitive testing variables. Oversampling was then performed in order to balance the initially imbalanced classes prior to training the model with 1000 estimators. Our results showed that a random forest model was effective (93.6% accuracy) at predicting the conversion of EMCI patients to AD based on these clinical features. Additionally, we focus on explainability by assessing the importance of each clinical feature. Our model could impact the clinical environment as a tool to predict the conversion to AD from a prodromal stage or to identify ideal candidates for clinical trials.

Sex Differences of Brain Functional Topography Revealed in Normal Aging and Alzheimer’s Disease Cohort

We applied graph theory analysis on resting-state functional magnetic resonance imaging data to evaluate sex differences of brain functional topography in normal controls (NCs), early mild cognitive impairment (eMCI), and AD patients. These metrics were correlated with RAVLT verbal learning and memory scores. The results show NCs have better functional connectivity (FC) metrics than eMCI and AD, and NC women show worse FC metrics compared to men, despite performing better on the RAVLT. FC differences between men and women diminished in eMCI and disappeared in AD. Within women, better FC metrics relate to better RAVLT learning in NCs and eMCI groups.

Hyperconnectivity of Self-Referential Network as a Predictive Biomarker of the Progression of Alzheimer’s Disease

Background: Self-referential processing is associated with the progression of Alzheimer’s disease (AD), and cerebrospinal fluid (CSF) proteins have become accepted biomarkers of AD. Objective: Our objective in this study was to focus on the relationships between the self-referential network (SRN) and CSF pathology in AD-spectrum patients. Methods: A total of 80 participants, including 20 cognitively normal, 20 early mild cognitive impairment (EMCI), 20 late MCI (LMCI), and 20 AD, were recruited for this study. Independent component analysis was used to explore the topological SRN patterns, and the abnormalities of this network were identified at different stages of AD. Finally, CSF pathological characteristics (i.e., CSF Aβ, t-tau, and p-tau) that affected the abnormalities of the SRN were further determined during the progression of AD. Results: Compared to cognitively normal subjects, AD-spectrum patients (i.e., EMCI, LMCI, and AD) showed a reversing trend toward an association between CSF pathological markers and the abnormal SRN occurring during the progression of AD. However, a certain disease state (i.e., the present LMCI) with a low concentration of CSF tau could evoke more hyperconnectivity of the SRN than other patients with progressively increasing concentrations of CSF tau (i.e., EMCI and AD), and this fluctuation of CSF tau was more sensitive to the hyperconnectivity of the SRN than the dynamic changes of CSF Aβ. Conclusion: The integrity of the SRN was closely associated with CSF pathological characteristics, and these findings support the view that the hyperconnectivity of the SRN will play an important role in monitoring the progression of the pre-dementia state to AD.

Constructing Dynamic Functional Networks via Weighted Regularization and Tensor Low-Rank Approximation for Early Mild Cognitive Impairment Classification

Brain functional networks constructed via regularization has been widely used in early mild cognitive impairment (eMCI) classification. However, few methods can properly reflect the similarities and differences of functional connections among different people. Most methods ignore some topological attributes, such as connection strength, which may delete strong functional connections in brain functional networks. To overcome these limitations, we propose a novel method to construct dynamic functional networks (DFN) based on weighted regularization (WR) and tensor low-rank approximation (TLA), and apply it to identify eMCI subjects from normal subjects. First, we introduce the WR term into the DFN construction and obtain WR-based DFNs (WRDFN). Then, we combine the WRDFNs of all subjects into a third-order tensor for TLA processing, and obtain the DFN based on WR and TLA (WRTDFN) of each subject in the tensor. We calculate the weighted-graph local clustering coefficient of each region in each WRTDFN as the effective feature, and use the t-test for feature selection. Finally, we train a linear support vector machine (SVM) classifier to classify the WRTDFNs of all subjects. Experimental results demonstrate that the proposed method can obtain DFNs with the scale-free property, and that the classification accuracy (ACC), the sensitivity (SEN), the specificity (SPE), and the area under curve (AUC) reach 87.0662% ± 0.3202%, 83.4363% ± 0.5076%, 90.6961% ± 0.3250% and 0.9431 ± 0.0023, respectively. We also achieve the best classification results compared with other comparable methods. This work can effectively improve the classification performance of DFNs constructed by existing methods for eMCI and has certain reference value for the early diagnosis of Alzheimer’s disease (AD).

Random forest model for feature-based Alzheimer’s disease conversion prediction from early mild cognitive impairment subjects

Alzheimer’s Disease (AD) conversion prediction from the mild cognitive impairment (MCI) stage has been a difficult challenge. This study focuses on providing an individualized MCI to AD conversion prediction using a balanced random forest model that leverages clinical data. In order to do this, 383 Early Mild Cognitive Impairment (EMCI) patients were gathered from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Of these patients, 49 would eventually convert to AD (EMCI_C), whereas the remaining 335 did not convert (EMCI_NC). All of these patients were split into training and testing data sets with 95 patients reserved for testing. Nine clinical features were selected, comprised of a mix of demographic, brain volume, and cognitive testing variables. Oversampling was then performed in order to balance the initially imbalanced classes. Our results showed that a random forest model was effective (93.6% accuracy) at predicting the conversion of EMCI patients to AD based on these clinical features. Additionally, we assessed the importance of each clinical feature at both the individual and model level for interpretation of the prediction itself. Our model could impact the clinical environment as a tool to predict the conversion to AD from a prodromal stage or to identify ideal candidates for clinical trials.