Fusion of ULS Group Constrained High- and Low-Order Sparse Functional Connectivity Networks for MCI Classification.

Functional connectivity networks, derived from resting-state fMRI data, have been found as effective biomarkers for identifying mild cognitive impairment (MCI) from healthy elderly. However, the traditional functional connectivity network is essentially a low-order network with the assumption that the brain activity is static over the entire scanning period, ignoring temporal variations among the correlations derived from brain region pairs. To overcome this limitation, we proposed a new type of sparse functional connectivity network to precisely describe the relationship of temporal correlations among brain regions. Specifically, instead of using the simple pairwise Pearson's correlation coefficient as connectivity, we first estimate the temporal low-order functional connectivity for each region pair based on an ULS Group constrained-UOLS regression algorithm, where a combination of ultra-least squares (ULS) criterion with a Group constrained topology structure detection algorithm is applied to detect the topology of functional connectivity networks, aided by an Ultra-Orthogonal Least Squares (UOLS) algorithm to estimate connectivity strength. Compared to the classical least squares criterion which only measures the discrepancy between the observed signals and the model prediction function, the ULS criterion takes into consideration the discrepancy between the weak derivatives of the observed signals and the model prediction function and thus avoids the overfitting problem. By using a similar approach, we then estimate the high-order functional connectivity from the low-order connectivity to characterize signal flows among the brain regions. We finally fuse the low-order and the high-order networks using two decision trees for MCI classification. Experimental results demonstrate the effectiveness of the proposed method on MCI classification.

Cortical graph neural network for AD and MCI diagnosis and transfer learning across populations.

Combining machine learning with neuroimaging data has a great potential for early diagnosis of mild cognitive impairment (MCI) and Alzheimer's disease (AD). However, it remains unclear how well the classifiers built on one population can predict MCI/AD diagnosis of other populations. This study aimed to employ a spectral graph convolutional neural network (graph-CNN), that incorporated cortical thickness and geometry, to identify MCI and AD based on 3089 T1-weighted MRI data of the ADNI-2 cohort, and to evaluate its feasibility to predict AD in the ADNI-1 cohort (n = 3602) and an Asian cohort (n = 347). For the ADNI-2 cohort, the graph-CNN showed classification accuracy of controls (CN) vs. AD at 85.8% and early MCI (EMCI) vs. AD at 79.2%, followed by CN vs. late MCI (LMCI) (69.3%), LMCI vs. AD (65.2%), EMCI vs. LMCI (60.9%), and CN vs. EMCI (51.8%). We demonstrated the robustness of the graph-CNN among the existing deep learning approaches, such as Euclidean-domain-based multilayer network and 1D CNN on cortical thickness, and 2D and 3D CNNs on T1-weighted MR images of the ADNI-2 cohort. The graph-CNN also achieved the prediction on the conversion of EMCI to AD at 75% and that of LMCI to AD at 92%. The find-tuned graph-CNN further provided a promising CN vs. AD classification accuracy of 89.4% on the ADNI-1 cohort and >90% on the Asian cohort. Our study demonstrated the feasibility to transfer AD/MCI classifiers learned from one population to the other. Notably, incorporating cortical geometry in CNN has the potential to improve classification performance.

Maternal sensitivity predicts anterior hippocampal functional networks in early childhood.

Maternal care influences child hippocampal development. The hippocampus is functionally organized along an anterior-posterior axis. Little is known with regards to the extent maternal care shapes offspring anterior and posterior hippocampal (aHPC, pHPC) functional networks. This study examined maternal behavior, especially maternal sensitivity, at 6 months postpartum in relation to aHPC and pHPC functional networks of children at age 4 and 6 years. Maternal sensitivity was assessed at 6 months via the "Maternal Behavior Q Sort (MBQS) mini for video". Subsequently, 61 and 76 children underwent resting-state functional magnetic resonance imaging (rs-fMRI), respectively, at 4 and 6 years of age. We found that maternal sensitivity assessed at 6 months postpartum was associated with the right aHPC functional networks in children at both 4 and 6 years of age. At age 4 years, maternal sensitivity was associated positively with the right aHPC's functional connectivity with the sensorimotor network and negatively with the aHPC's functional connectivity with the top-down cognitive control network. At 6 years of age, maternal sensitivity was linked positively with the right aHPC's functional connectivity with the visual-processing network. Our findings suggested that maternal sensitivity in infancy has a long-term impact on the anterior hippocampal functional network in preschool children, implicating a potential role of maternal care in shaping child brain development in early life.

Enhancing the representation of functional connectivity networks by fusing multi-view information for autism spectrum disorder diagnosis.

Functional connectivity network provides novel insights on how distributed brain regions are functionally integrated, and its deviations from healthy brain have recently been employed to identify biomarkers for neuropsychiatric disorders. However, most of brain network analysis methods utilized features extracted only from one functional connectivity network for brain disease detection and cannot provide a comprehensive representation on the subtle disruptions of brain functional organization induced by neuropsychiatric disorders. Inspired by the principles of multi-view learning which utilizes information from multiple views to enhance object representation, we propose a novel multiple network based framework to enhance the representation of functional connectivity networks by fusing the common and complementary information conveyed in multiple networks. Specifically, four functional connectivity networks corresponding to the four adjacent values of regularization parameter are generated via a sparse regression model with group constraint ( l2,1 -norm), to enhance the common intrinsic topological structure and limit the error rate caused by different views. To obtain a set of more meaningful and discriminative features, we propose using a modified version of weighted clustering coefficients to quantify the subtle differences of each group-sparse network at local level. We then linearly fuse the selected features from each individual network via a multi-kernel support vector machine for autism spectrum disorder (ASD) diagnosis. The proposed framework achieves an accuracy of 79.35%, outperforming all the compared single network methods for at least 7% improvement. Moreover, compared with other multiple network methods, our method also achieves the best performance, that is, with at least 11% improvement in accuracy.

Fronto-parietal numerical networks in relation with early numeracy in young children.

Early numeracy provides the foundation of acquiring mathematical skills that is essential for future academic success. This study examined numerical functional networks in relation to counting and number relational skills in preschoolers at 4 and 6 years of age. The counting and number relational skills were assessed using school readiness test (SRT). Resting-state fMRI (rs-fMRI) was acquired in 123 4-year-olds and 146 6-year-olds. Among them, 61 were scanned twice over the course of 2 years. Meta-analysis on existing task-based numeracy fMRI studies identified the left parietal-dominant network for both counting and number relational skills and the right parietal-dominant network only for number relational skills in adults. We showed that the fronto-parietal numerical networks, observed in adults, already exist in 4-year and 6-year-olds. The counting skills were associated with the bilateral fronto-parietal network in 4-year-olds and with the right parietal-dominant network in 6-year-olds. Moreover, the number relational skills were related to the bilateral fronto-parietal and right parietal-dominant networks in 4-year-olds and had a trend of the significant relationship with the right parietal-dominant network in 6-year-olds. Our findings suggested that neural fine-tuning of the fronto-parietal numerical networks may subserve the maturation of numeracy in early childhood.

Novel Effective Connectivity Inference Using Ultra-Group Constrained Orthogonal Forward Regression and Elastic Multilayer Perceptron Classifier for MCI Identification.

Mild cognitive impairment (MCI) detection is important, such that appropriate interventions can be imposed to delay or prevent its progression to severe stages, including Alzheimer's disease (AD). Brain connectivity network inferred from the functional magnetic resonance imaging data has been prevalently used to identify the individuals with MCI/AD from the normal controls. The capability to detect the causal or effective connectivity is highly desirable for understanding directed functional interactions between brain regions and further helping the detection of MCI. In this paper, we proposed a novel sparse constrained effective connectivity inference method and an elastic multilayer perceptron classifier for MCI identification. Specifically, a ultra-group constrained structure detection algorithm is first designed to identify the parsimonious topology of the effective connectivity network, in which the weak derivatives of the observable data are considered. Second, based on the identified topology structure, an effective connectivity network is then constructed by using an ultra-orthogonal forward regression algorithm to minimize the shrinking effect of the group constraint-based method. Finally, the effective connectivity network is validated in MCI identification using an elastic multilayer perceptron classifier, which extracts lower to higher level information from initial input features and hence improves the classification performance. Relatively high classification accuracy is achieved by the proposed method when compared with the state-of-the-art classification methods. Furthermore, the network analysis results demonstrate that MCI patients suffer a rich club effect loss and have decreased connectivity among several brain regions. These findings suggest that the proposed method not only improves the classification performance but also successfully discovers critical disease-related neuroimaging biomarkers.

Multimodal hyper-connectivity of functional networks using functionally-weighted LASSO for MCI classification.

Recent works have shown that hyper-networks derived from blood-oxygen-level-dependent (BOLD) fMRI, where an edge (called hyper-edge) can be connected to more than two nodes, are effective biomarkers for MCI classification. Although BOLD fMRI is a high temporal resolution fMRI approach to assess alterations in brain networks, it cannot pinpoint to a single correlation of neuronal activity since BOLD signals are composite. In contrast, arterial spin labeling (ASL) is a lower temporal resolution fMRI technique for measuring cerebral blood flow (CBF) that can provide quantitative, direct brain network physiology measurements. This paper proposes a novel sparse regression algorithm for inference of the integrated hyper-connectivity networks from BOLD fMRI and ASL fMRI. Specifically, a least absolution shrinkage and selection operator (LASSO) algorithm, which is constrained by the functional connectivity derived from ASL fMRI, is employed to estimate hyper-connectivity for characterizing BOLD-fMRI-based functional interaction among multiple regions. An ASL-derived functional connectivity is constructed by using an Ultra-GroupLASSO-UOLS algorithm, where the combination of ultra-least squares (ULS) criterion with a group LASSO (GroupLASSO) algorithm is applied to detect the topology of ASL-based functional connectivity networks, and then an ultra-orthogonal least squares (UOLS) algorithm is used to estimate the connectivity strength. By combining the complementary characterization conveyed by rs-fMRI and ASL fMRI, our multimodal hyper-networks demonstrated much better discriminative characteristics than either the conventional pairwise connectivity networks or the unimodal hyper-connectivity networks. Experimental results on publicly available ADNI dataset demonstrate that the proposed method outperforms the existing single modality based sparse functional connectivity inference methods.

Behavioral Heterogeneity in Relation with Brain Functional Networks in Young Children.

This study aimed to identify distinct behavioral profiles in a population-based sample of 654 4-year-old children and characterize their relationships with brain functional networks using resting-state functional magnetic resonance imaging data. Young children showed 7 behavioral profiles, including a super healthy behavioral profile with the lowest scores across all Child Behavior CheckList (CBCL) subscales (G1) and other 6 behavioral profiles, respectively with pronounced withdrawal (G2), somatic complaints (G3), anxiety and withdrawal (G4), somatic complaints and withdrawal (G5), the mixture of emotion, withdrawal, and aggression (G6), and attention (G7) problems. Compared with children in G1, children with withdrawal shared abnormal functional connectivities among the sensorimotor networks. Children in emotionally relevant problems shared the common pattern among the attentional and frontal networks. Nevertheless, children in sole withdrawal problems showed a unique pattern of connectivity alterations among the sensorimotor, cerebellar, and salience networks. Children with somatic complaints showed abnormal functional connectivities between the attentional and subcortical networks, and between the language and posterior default mode networks. This study provides novel evidence on the existence of behavioral heterogeneity in early childhood and its associations with specific functional networks that are clinically relevant phenotypes for mental illness and are apparent from early childhood.

Structural Connectivity Guided Sparse Effective Connectivity for MCI Identification.

Recent advances in network modelling techniques have enabled the study of neurological disorders at a whole-brain level based on functional connectivity inferred from resting-state magnetic resonance imaging (rs-fMRI) scan possible. However, constructing a directed effective connectivity, which provides a more comprehensive characterization of functional interactions among the brain regions, is still a challenging task particularly when the ultimate goal is to identify disease associated brain functional interaction anomalies. In this paper, we propose a novel method for inferring effective connectivity from multimodal neuroimaging data for brain disease classification. Specifically, we apply a newly devised weighted sparse regression model on rs-fMRI data to determine the network structure of effective connectivity with the guidance from diffusion tensor imaging (DTI) data. We further employ a regression algorithm to estimate the effective connectivity strengths based on the previously identified network structure. We finally utilize a bagging classifier to evaluate the performance of the proposed sparse effective connectivity network through identifying mild cognitive impairment from healthy aging.

Multi-task diagnosis for autism spectrum disorders using multi-modality features: A multi-center study.

Autism spectrum disorder (ASD) is a neurodevelopment disease characterized by impairment of social interaction, language, behavior, and cognitive functions. Up to now, many imaging-based methods for ASD diagnosis have been developed. For example, one may extract abundant features from multi-modality images and then derive a discriminant function to map the selected features toward the disease label. A lot of recent works, however, are limited to single imaging centers. To this end, we propose a novel multi-modality multi-center classification (M3CC) method for ASD diagnosis. We treat the classification of each imaging center as one task. By introducing the task-task and modality-modality regularizations, we solve the classification for all imaging centers simultaneously. Meanwhile, the optimal feature selection and the modeling of the discriminant functions can be jointly conducted for highly accurate diagnosis. Besides, we also present an efficient iterative optimization solution to our formulated problem and further investigate its convergence. Our comprehensive experiments on the ABIDE database show that our proposed method can significantly improve the performance of ASD diagnosis, compared to the existing methods. Hum Brain Mapp 38:3081-3097, 2017. © 2017 Wiley Periodicals, Inc.

Neonatal neural networks predict children behavioral profiles later in life.

This study aimed to examine heterogeneity of neonatal brain network and its prediction to child behaviors at 24 and 48 months of age. Diffusion tensor imaging (DTI) tractography was employed to construct brain anatomical network for 120 neonates. Clustering coefficients of individual structures were computed and used to classify neonates with similar brain anatomical networks into one group. Internalizing and externalizing behavioral problems were assessed using maternal reports of the Child Behavior Checklist (CBCL) at 24 and 48 months of age. The profile of CBCL externalizing and internalizing behaviors was then examined in the groups identified based on the neonatal brain network. Finally, support vector machine and canonical correlation analysis were used to identify brain structures whose clustering coefficients together significantly contribute the variation of the behaviors at 24 and 48 months of age. Four meaningful groups were revealed based on the brain anatomical networks at birth. Moreover, the clustering coefficients of the brain regions that most contributed to this grouping of neonates were significantly associated with childhood internalizing and externalizing behaviors assessed at 24 and 48 months of age. Specially, the clustering coefficient of the right amygdala was associated with both internalizing and externalizing behaviors at 24 months of age, while the clustering coefficients of the right inferior frontal cortex and insula were associated with externalizing behaviors at 48 months of age. Our findings suggested that neural organization established during fetal development could to some extent predict individual differences in behavioral-emotional problems in early childhood. Hum Brain Mapp 38:1362-1373, 2017. © 2016 Wiley Periodicals, Inc.

Fusion of High-Order and Low-Order Effective Connectivity Networks for MCI Classification.

Functional connectivity network derived from resting-state fMRI data has been found as effective biomarkers for identifying patients with mild cognitive impairment from healthy elderly. However, the ordinary functional connectivity network is essentially a low-order network with the assumption that the brain is static during the entire scanning period, ignoring the temporal variations among correlations derived from brain region pairs. To overcome this weakness, we proposed a new type of high-order network to more accurately describe the relationship of temporal variations among brain regions. Specifically, instead of the commonly used undirected pairwise Pearson's correlation coefficient, we first estimated the low-order effective connectivity network based on a novel sparse regression algorithm. By using the similar approach, we then constructed the high-order effective connectivity network from low-order connectivity to incorporate signal flow information among the brain regions. We finally combined the low-order and the high-order effective connectivity networks using two decision trees for MCI classification and experimental results obtained demonstrate the superiority of the proposed method over the conventional undirected low-order and high-order functional connectivity networks, as well as the low-order and high-order effective connectivity networks when they were used separately.

Novel Effective Connectivity Network Inference for MCI Identification.

Inferring effective brain connectivity network is a challenging task owing to perplexing noise effects, the curse of dimensionality, and inter-subject variability. However, most existing network inference methods are based on correlation analysis and consider the datum points individually, revealing limited information of the neuron interactions and ignoring the relations amongst the derivatives of the data. Hence, we proposed a novel ultra group-constrained sparse linear regression model for effective connectivity inference. This model utilizes not only the discrepancy between observed signals and the model prediction, but also the discrepancy between the associated weak derivatives of the observed and the model signals for a more accurate effective connectivity inference. What's more, a group constraint is applied to minimize the inter-subject variability and the proposed modeling was validated on a mild cognitive impairment dataset with superior results achieved.

Multimodal Hyper-connectivity Networks for MCI Classification.

Hyper-connectivity network is a network where every edge is connected to more than two nodes, and can be naturally denoted using a hyper-graph. Hyper-connectivity brain network, either based on structural or functional interactions among the brain regions, has been used for brain disease diagnosis. However, the conventional hyper-connectivity network is constructed solely based on single modality data, ignoring potential complementary information conveyed by other modalities. The integration of complementary information from multiple modalities has been shown to provide a more comprehensive representation about the brain disruptions. In this paper, a novel multimodal hyper-network modelling method was proposed for improving the diagnostic accuracy of mild cognitive impairment (MCI). Specifically, we first constructed a multimodal hyper-connectivity network by simultaneously considering information from diffusion tensor imaging and resting-state functional magnetic resonance imaging data. We then extracted different types of network features from the hyper-connectivity network, and further exploited a manifold regularized multi-task feature selection method to jointly select the most discriminative features. Our proposed multimodal hyper-connectivity network demonstrated a better MCI classification performance than the conventional single modality based hyper-connectivity networks.

Improving Estimation of Fiber Orientations in Diffusion MRI Using Inter-Subject Information Sharing.

Diffusion magnetic resonance imaging is widely used to investigate diffusion patterns of water molecules in the human brain. It provides information that is useful for tracing axonal bundles and inferring brain connectivity. Diffusion axonal tracing, namely tractography, relies on local directional information provided by the orientation distribution functions (ODFs) estimated at each voxel. To accurately estimate ODFs, data of good signal-to-noise ratio and sufficient angular samples are desired. This is however not always available in practice. In this paper, we propose to improve ODF estimation by using inter-subject image correlation. Specifically, we demonstrate that diffusion-weighted images acquired from different subjects can be transformed to the space of a target subject to drastically increase the number of angular samples to improve ODF estimation. This is largely due to the incoherence of the angular samples generated when the diffusion signals are reoriented and warped to the target space. To reorient the diffusion signals, we propose a new spatial normalization method that directly acts on diffusion signals using local affine transforms. Experiments on both synthetic data and real data show that our method can reduce noise-induced artifacts, such as spurious ODF peaks, and yield more coherent orientations.

Joint feature-sample selection and robust diagnosis of Parkinson's disease from MRI data.

Parkinson's disease (PD) is an overwhelming neurodegenerative disorder caused by deterioration of a neurotransmitter, known as dopamine. Lack of this chemical messenger impairs several brain regions and yields various motor and non-motor symptoms. Incidence of PD is predicted to double in the next two decades, which urges more research to focus on its early diagnosis and treatment. In this paper, we propose an approach to diagnose PD using magnetic resonance imaging (MRI) data. Specifically, we first introduce a joint feature-sample selection (JFSS) method for selecting an optimal subset of samples and features, to learn a reliable diagnosis model. The proposed JFSS model effectively discards poor samples and irrelevant features. As a result, the selected features play an important role in PD characterization, which will help identify the most relevant and critical imaging biomarkers for PD. Then, a robust classification framework is proposed to simultaneously de-noise the selected subset of features and samples, and learn a classification model. Our model can also de-noise testing samples based on the cleaned training data. Unlike many previous works that perform de-noising in an unsupervised manner, we perform supervised de-noising for both training and testing data, thus boosting the diagnostic accuracy. Experimental results on both synthetic and publicly available PD datasets show promising results. To evaluate the proposed method, we use the popular Parkinson's progression markers initiative (PPMI) database. Our results indicate that the proposed method can differentiate between PD and normal control (NC), and outperforms the competing methods by a relatively large margin. It is noteworthy to mention that our proposed framework can also be used for diagnosis of other brain disorders. To show this, we have also conducted experiments on the widely-used ADNI database. The obtained results indicate that our proposed method can identify the imaging biomarkers and diagnose the disease with favorable accuracies compared to the baseline methods.

High-order resting-state functional connectivity network for MCI classification.

Brain functional connectivity (FC) network, estimated with resting-state functional magnetic resonance imaging (RS-fMRI) technique, has emerged as a promising approach for accurate diagnosis of neurodegenerative diseases. However, the conventional FC network is essentially low-order in the sense that only the correlations among brain regions (in terms of RS-fMRI time series) are taken into account. The features derived from this type of brain network may fail to serve as an effective disease biomarker. To overcome this drawback, we propose extraction of novel high-order FC correlations that characterize how the low-order correlations between different pairs of brain regions interact with each other. Specifically, for each brain region, a sliding window approach is first performed over the entire RS-fMRI time series to generate multiple short overlapping segments. For each segment, a low-order FC network is constructed, measuring the short-term correlation between brain regions. These low-order networks (obtained from all segments) describe the dynamics of short-term FC along the time, thus also forming the correlation time series for every pair of brain regions. To overcome the curse of dimensionality, we further group the correlation time series into a small number of different clusters according to their intrinsic common patterns. Then, the correlation between the respective mean correlation time series of different clusters is calculated to represent the high-order correlation among different pairs of brain regions. Finally, we design a pattern classifier, by combining features of both low-order and high-order FC networks. Experimental results verify the effectiveness of the high-order FC network on disease diagnosis. Hum Brain Mapp 37:3282-3296, 2016. © 2016 Wiley Periodicals, Inc.

Hyper-connectivity of functional networks for brain disease diagnosis.

Exploring structural and functional interactions among various brain regions enables better understanding of pathological underpinnings of neurological disorders. Brain connectivity network, as a simplified representation of those structural and functional interactions, has been widely used for diagnosis and classification of neurodegenerative diseases, especially for Alzheimer's disease (AD) and its early stage - mild cognitive impairment (MCI). However, the conventional functional connectivity network is usually constructed based on the pairwise correlation among different brain regions and thus ignores their higher-order relationships. Such loss of high-order information could be important for disease diagnosis, since neurologically a brain region predominantly interacts with more than one other brain regions. Accordingly, in this paper, we propose a novel framework for estimating the hyper-connectivity network of brain functions and then use this hyper-network for brain disease diagnosis. Here, the functional connectivity hyper-network denotes a network where each of its edges representing the interactions among multiple brain regions (i.e., an edge can connect with more than two brain regions), which can be naturally represented by a hyper-graph. Specifically, we first construct connectivity hyper-networks from the resting-state fMRI (R-fMRI) time series by using sparse representation. Then, we extract three sets of brain-region specific features from the connectivity hyper-networks, and further exploit a manifold regularized multi-task feature selection method to jointly select the most discriminative features. Finally, we use multi-kernel support vector machine (SVM) for classification. The experimental results on both MCI dataset and attention deficit hyperactivity disorder (ADHD) dataset demonstrate that, compared with the conventional connectivity network-based methods, the proposed method can not only improve the classification performance, but also help discover disease-related biomarkers important for disease diagnosis.

Multilevel Deficiency of White Matter Connectivity Networks in Alzheimer's Disease: A Diffusion MRI Study with DTI and HARDI Models.

Alzheimer's disease (AD) is the most common form of dementia in elderly people. It is an irreversible and progressive brain disease. In this paper, we utilized diffusion-weighted imaging (DWI) to detect abnormal topological organization of white matter (WM) structural networks. We compared the differences between WM connectivity characteristics at global, regional, and local levels in 26 patients with probable AD and 16 normal control (NC) elderly subjects, using connectivity networks constructed with the diffusion tensor imaging (DTI) model and the high angular resolution diffusion imaging (HARDI) model, respectively. At the global level, we found that the WM structural networks of both AD and NC groups had a small-world topology; however, the AD group showed a significant decrease in both global and local efficiency, but an increase in clustering coefficient and the average shortest path length. We further found that the AD patients had significantly decreased nodal efficiency at the regional level, as well as weaker connections in multiple local cortical and subcortical regions, such as precuneus, temporal lobe, hippocampus, and thalamus. The HARDI model was found to be more advantageous than the DTI model, as it was more sensitive to the deficiencies in AD at all of the three levels.

State-space model with deep learning for functional dynamics estimation in resting-state fMRI.

Studies on resting-state functional Magnetic Resonance Imaging (rs-fMRI) have shown that different brain regions still actively interact with each other while a subject is at rest, and such functional interaction is not stationary but changes over time. In terms of a large-scale brain network, in this paper, we focus on time-varying patterns of functional networks, i.e., functional dynamics, inherent in rs-fMRI, which is one of the emerging issues along with the network modelling. Specifically, we propose a novel methodological architecture that combines deep learning and state-space modelling, and apply it to rs-fMRI based Mild Cognitive Impairment (MCI) diagnosis. We first devise a Deep Auto-Encoder (DAE) to discover hierarchical non-linear functional relations among regions, by which we transform the regional features into an embedding space, whose bases are complex functional networks. Given the embedded functional features, we then use a Hidden Markov Model (HMM) to estimate dynamic characteristics of functional networks inherent in rs-fMRI via internal states, which are unobservable but can be inferred from observations statistically. By building a generative model with an HMM, we estimate the likelihood of the input features of rs-fMRI as belonging to the corresponding status, i.e., MCI or normal healthy control, based on which we identify the clinical label of a testing subject. In order to validate the effectiveness of the proposed method, we performed experiments on two different datasets and compared with state-of-the-art methods in the literature. We also analyzed the functional networks learned by DAE, estimated the functional connectivities by decoding hidden states in HMM, and investigated the estimated functional connectivities by means of a graph-theoretic approach.