Seizure onset zone (SOZ) identification using effective brain connectivity of epileptogenic networks.

Objective. To demonstrate the capability of utilizing graph feature-based supervised machine learning (ML) algorithm on intracranial electroencephalogram recordings for the identification of seizure onset zones (SOZs) in individuals with drug-resistant epilepsy.Approach. Utilizing three model-free measures of effective connectivity (EC)-directed information, mutual information-guided Granger causality index (MI-GCI), and frequency-domain convergent cross-mapping (FD-CCM) - directed graphs are generated. Graph centrality measures at different sparsity are used as the classifier's features.Main results. The centrality features achieve high accuracies exceeding 90% in distinguishing SOZ electrodes from non-SOZ electrodes. Notably, a sparse graph representation with just ten features and simple ML models effectively achieves such performance. The study identifies FD-CCM centrality measures as particularly significant, with a mean AUC of 0.93, outperforming prior literature. The FD-CCM-based graph modeling also highlights elevated centrality measures among SOZ electrodes, emphasizing heightened activity relative to non-SOZ electrodes during ictogenesis.Significance. This research not only underscores the efficacy of automated SOZ identification but also illuminates the potential of specific EC measures in enhancing discriminative power within the context of epilepsy research.

Reservoir Computing With Dynamic Reservoir using Cascaded DNA Memristors.

This article proposes molecular and DNA memristors where the state is defined by a single output variable. In past molecular and DNA memristors, the state of the memristor was defined based on two output variables. These memristors cannot be cascaded because their input and output sizes are different. We introduce a different definition of state for the molecular and DNA memristors. This change allows cascading of memristors. The proposed memristors are used to build reservoir computing (RC) models that can process temporal inputs. An RC system consists of two parts: reservoir and readout layer. The first part projects the information from the input space into a high-dimensional feature space. We also study the input-state characteristics of the cascaded memristors and show that the cascaded memristors retain the memristive behavior. The cascade connections in a reservoir can change dynamically; this allows the synthesis of a dynamic reservoir as opposed to a static one in the prior work. This reduces the number of memristors significantly compared to a static reservoir. The inputs to the readout layer correspond to one molecule per state instead of two; this significantly reduces the number of molecular and DSD reactions for the readout layer. A DNA RC system consisting of DNA memristors and a DNA readout layer is used to detect seizures from intra-cranial electroencephalogram (iEEG). We also demonstrate that a DNA RC system consisting of three cascaded DNA memristors and a DNA readout layer can be used to solve the time-series prediction task. The proposed approach can reduce the number of DNA strand displacement (DSD) reactions by three to five times compared to prior approaches.

Effective Brain Connectivity Extraction by Frequency-Domain Convergent Cross-Mapping (FDCCM) and Its Application in Parkinson's Disease Classification.

OBJECTIVE: Inferring causal or effective connectivity between measured timeseries is crucial to understanding directed interactions in complex systems. This task is especially challenging in the brain as the underlying dynamics are not well-understood. This paper aims to introduce a novel causality measure called frequency-domain convergent cross-mapping (FDCCM) that utilizes frequency-domain dynamics through nonlinear state-space reconstruction. METHOD: Using synthesized chaotic timeseries, we investigate general applicability of FDCCM at different causal strengths and noise levels. We also apply our method on two resting-state Parkinson's datasets with 31 and 54 subjects, respectively. To this end, we construct causal networks, extract network features, and perform machine learning analysis to distinguish Parkinson's disease patients (PD) from age and gender-matched healthy controls (HC). Specifically, we use the FDCCM networks to compute the betweenness centrality of the network nodes, which act as features for the classification models. RESULT: The analysis on simulated data showed that FDCCM is resilient to additive Gaussian noise, making it suitable for real-world applications. Our proposed method also decodes scalp-EEG signals to classify the PD and HC groups with approximately 97% leave-one-subject-out cross-validation accuracy. We compared decoders from six cortical regions to find that features derived from the left temporal lobe lead to a higher classification accuracy of 84.5% compared to other regions. Moreover, when the classifier trained using FDCCM networks from one dataset was tested on an independent out-of-sample dataset, it attained an accuracy of 84%. This accuracy is significantly higher than correlational networks (45.2%) and CCM networks (54.84%). SIGNIFICANCE: These findings suggest that our spectral-based causality measure can improve classification performance and reveal useful network biomarkers of Parkinson's disease.

Betweenness Centrality in Resting-State Functional Networks Distinguishes Parkinson's Disease.

The goal of this paper is to use graph theory network measures derived from non-invasive electroencephalography (EEG) to develop neural decoders that can differentiate Parkinson's disease (PD) patients from healthy controls (HC). EEG signals from 27 patients and 27 demographically matched controls from New Mexico were analyzed by estimating their functional networks. Data recorded from the patients during ON and OFF levodopa sessions were included in the analysis for comparison. We used betweenness centrality of estimated functional networks to classify the HC and PD groups. The classifiers were evaluated using leave-one-out cross-validation. We observed that the PD patients (on and off medication) could be distinguished from healthy controls with 89% accuracy - approximately 4% higher than the state-of-the-art on the same dataset. This work shows that brain network analysis using extracranial resting-state EEG can discover patterns of interactions indicative of PD. This approach can also be extended to other neurological disorders.

DNA Memristors and Their Application to Reservoir Computing.

This paper introduces memristors realized by molecular and DNA reactions. Molecular memristors process one input molecule, generate two output molecules, and are realized using two molecular reactions with two different rate constants. The DNA memristors are realized using five DNA strand displacement (DSD) reactions with two effective rate constants. The hysteresis behavior is preserved in the proposed memristors, and this behavior can be altered by changing the ratios of the rate constants. The state of the memristor can be computed from the concentrations of the two output molecules using bipolar fractional coding. We describe how the proposed memristors can be used to learn the spatial and temporal properties of data via the reservoir computing (RC) model. An RC system can be divided into two parts: reservoir and readout layer. The first part transfers the information from the input space to a high-dimensional spatiotemporal feature space represented by the state of reservoirs. The connectivity structure of the reservoir will remain fixed through the dynamical evaluations. The readout layer effectively maps the projected features to the target output. A dynamical memristor array is used to implement an RC system that exploits the internal dynamical processes of the memristors. The readout layer implements a matrix-vector multiplication using molecular reactions, also based on bipolar fractional coding. All molecular reactions are mapped to DSD reactions. The RC system based on the DNA reservoir and the DNA readout layer is used to solve a handwritten digit recognition task and a second-order time series prediction task. The performance of the DNA RC system is comparable to that of an electronic memristor RC system for both tasks.

Reservoir Computing Using DNA Oscillators.

This paper presents novel implementations for reservoir computing (RC) using DNA oscillators. An RC system consists of two parts: reservoir and readout layer. The reservoir projects input signals into a high-dimensional feature space which is formed by the state of the reservoir. The internal connectivity structure of the reservoir remains unchanged throughout computation. After training, the readout layer maps the projected features into the desired output. It has been shown in prior work that coupled deoxyribozyme oscillators can be used as the reservoir. In this paper, we utilize the n-phase molecular oscillator (n ≥ 3) presented in our prior work. The readout layer implements a matrix-vector multiplication using molecular reactions based on molecular analog multiplication. All molecular reactions are mapped to DNA strand displacement (DSD) reactions. We also introduce a novel encoding method that can significantly reduce the reaction time. The feasibility of the proposed RC systems based on the DNA oscillator is demonstrated for the handwritten digit recognition task and a second-order nonlinear prediction task.

Decoding Human Cognitive Control Using Functional Connectivity of Local Field Potentials.

Many patients with mental illnesses characterized by impaired cognitive control have no relief from gold-standard clinical treatments resulting in a pressing need for new alternatives. This paper develops a neural decoder to detect task engagement in ten human subjects during a conflict-based behavioral task known as the multi-source interference task (MSIT). Task engagement is of particular interest here because closed-loop brain stimulation during those states can augment decision-making. The functional connectivity patterns of the electrodes are extracted. A principal component analysis of these patterns is carried out and the ranked principal components are used as inputs to train subject-specific linear support vector machine classifiers. In this paper, we show that task engagement can be differentiated from background brain activity with a median accuracy of 89.7%. This was accomplished by constructing distributed functional networks from local field potentials recording during the task performance. A further challenge is that goal-directed efforts take place over higher temporal resolution. Task engagement must thus be detected at a similar rate for proactive intervention. We show that our algorithms can detect task engagement from neural recordings in less than 2 seconds; this can be further improved using an application-specific device.

Spectral Features Based Decoding of Task Engagement: The Role of Theta and High Gamma Bands in Cognitive Control.

This paper analyzes local field potentials (LFP) from 10 human subjects to discover frequency-dependent biomarkers of cognitive conflict. We utilize cortical and sub-cortical LFP recordings from the subjects during a cognitive task known as the Multi-Source Interference Task (MSIT). We decode the task engagement and discover biomarkers that may facilitate closed-loop neuromodulation to enhance cognitive control. First, we show that spectral power features in predefined frequency bands can be used to classify task and non-task segments with a median accuracy of 88.1%. Here the features are first ranked using the Bayes Factor and then used as inputs to subject-specific linear support vector machine classifiers. Second, we show that theta (4-8 Hz) band, and high gamma (65-200 Hz) band oscillations are modulated during the task performance. Third, by isolating time-series from specific brain regions of interest, we observe that a subset of the dorsolateral prefrontal cortex features is sufficient to decode the task states. The paper shows that cognitive control evokes robust neurological signatures, especially in the prefrontal cortex (PFC).

Effect of Modulating fMRI Time-Series on Fluid Ability and Fluid Intelligence for Healthy Humans.

This paper investigates the effect of filtering (or modulating) the functional magnetic resonance imaging (fMRI) time-series on intelligence metrics predicted using dynamic functional connectivity (dFC). Thirteen brain regions that have highest correlation with intelligence are selected and their corresponding time-series are filtered. Using filtered time-series, the modified intelligence metrics are predicted. This experiment investigates whether modulating the time-series of one or two regions of the brain will increase or decrease the fluid ability and fluid intelligence among healthy humans. Two sets of experiments are performed. In the first case, each of the thirteen regions is separately filtered using four different digital filters with passbands: i) 0 - 0.25π, ii) 0.25π - 0.5π, iii) 0.5π - 0.75π, and iv) 0.75π - π, respectively. In the second case, two of the thirteen regions are filtered simultaneously using a low-pass filter of passband 0 - 0.25π. In both cases, the predicted intelligence declined for 45-65% of the subjects after filtering in comparison with the ground truths. In the first case, the low-pass filtering process had the highest predicted intelligence among the four filters. In the second case, it was noticed that the filtering of two regions simultaneously resulted in a higher prediction of intelligence for over 80% of the subjects compared to low-pass filtering of a single region.

Seizure Prediction using Convolutional Neural Networks and Sequence Transformer Networks.

Accurate seizure prediction is important for design of wearable and implantable devices that can improve the lives of subjects with epilepsy. Such implantable devices can be used for closed-loop neuromodulation. However, there are many challenges that inhibit the performance of prediction models. One challenge in accurately predicting seizures is the nonstationarity of the EEG signals. This paper presents a patient-specific deep learning approach to improve predictive performance by transforming EEG data before extracting features for seizure prediction. In the proposed approach, a Sequence Transformer Network (STN) is first used to learn temporal and magnitude invariances in EEG data. The proposed method further computes the short-time Fourier transform (STFT) of the transformed EEG signals as input features to a convolutional neural network (CNN). A k-out-of-n post-processing method is used to reduce the significance of isolated false positives. The approach is tested using intracranial EEG from the American Epilepsy Society Seizure Prediction Challenge dataset. Leave-one-out cross validation is used to evaluate the model. The proposed model achieves an overall sensitivity of 82%, false prediction rate of 0.38/hour, and average AUC of 0.746.

Effect of SSRIs on Resting-State Functional Brain Networks in Adolescents with Major Depressive Disorder.

Investigation of brain changes in functional connectivity and functional network topology from receiving 8-week selective serotonin reuptake inhibitor (SSRI) treatments is conducted in 12 unmedicated adolescents with major depressive disorder (MDD) by using wavelet-filtered resting-state functional magnetic resonance imaging (fMRI). Changes are observed in frontal-limbic, temporal, and default mode networks. In particular, topological analysis shows, at the global scale and in the 0.12-0.25 Hz band, that the normalized clustering coefficient and smallworldness of brain networks decreased after treatment. Regional changes in clustering coefficient and efficiency were observed in the bilateral caudal middle frontal gyrus, rostral middle frontal gyrus, superior temporal gyrus, left pars triangularis, putamen, and right superior frontal gyrus. Furthermore, changes of nodal centrality and changes of connectivity associated with these frontal and temporal regions confirm the global topological alternations. Moreover, frequency dependence is observed from FDR-controlled subnetworks for the limbic-cortical connectivity change. In the high-frequency band, the altered connections involve mostly frontal regions, while the altered connections in the low-frequency bands spread to parietal and temporal areas. Due to the limitation of small sample sizes and lack of placebo control, these preliminary findings require confirmation with future work using larger samples. Confirmation of biomarkers associated with treatment could suggest potential avenues for clinical applications such as tracking treatment response and neurobiologically informed treatment optimization.

Region-Level Functional and Effective Network Analysis of Human Brain During Cognitive Task Engagement.

Mental disorders are a major source of disability, with few effective treatments. It has recently been argued that these diseases might be effectively treated by focusing on decision-making, and specifically remediating decision-making deficits that act as "ingredients" in these disorders. Prior work showed that direct electrical brain stimulation can enhance human cognitive control, and consequently decision-making. This raises a challenge of detecting cognitive control lapses directly from electrical brain activity. Here, we demonstrate approaches to overcome that challenge. We propose a novel method, referred to as maximal variance node merging (MVNM), that merges nodes within a brain region to construct informative inter-region brain networks. We employ this method to estimate functional (correlational) and effective (causal) networks using local field potentials (LFP) during a cognitive behavioral task. The effective networks computed using convergent cross mapping differentiate task engagement from background neural activity with 85% median classification accuracy. We also derive task engagement networks (TENs): networks that constitute the most discriminative inter-region connections. Subsequent graph analysis illustrates the crucial role of the dorsolateral prefrontal cortex (dlPFC) in task engagement, consistent with a widely accepted model for cognition. We also show that task engagement is linked to prefrontal cortex theta (4-8 Hz) oscillations. We, therefore, identify objective biomarkers associated with task engagement. These approaches may generalize to other cognitive functions, forming the basis of a network-based approach to detecting and rectifying decision deficits.

Classification of Adolescent Major Depressive Disorder Via Static and Dynamic Connectivity.

This paper introduces an approach for classifying adolescents suffering from MDD using resting-state fMRI. Accurate diagnosis of MDD involves interviews with adolescent patients and their parents, symptom rating scales based on Diagnostic and Statistical Manual of Mental Disorders (DSM), behavioral observation as well as the experience of a clinician. Discovering predictive biomarkers for diagnosing MDD patients using functional magnetic resonance imaging (fMRI) scans can assist the clinicians in their diagnostic assessments. This paper investigates various static and dynamic connectivity measures extracted from resting-state fMRI for assisting with MDD diagnosis. First, absolute Pearson correlation matrices from 85 brain regions are computed and they are used to calculate static features for predicting MDD. A predictive sub-network extracted using sub-graph entropy classifies adolescent MDD vs. typical healthy controls with high accuracy, sensitivity and specificity. Next, approaches utilizing dynamic connectivity are employed to extract tensor based, independent component based and principal component based subject specific attributes. Finally, features from static and dynamic approaches are combined to create a feature vector for classification. A leave-one-out cross-validation method is used for the final predictor performance. Out of 49 adolescents with MDD and 33 matched healthy controls, a support vector machine (SVM) classifier using a radial basis function (RBF) kernel using differential sub-graph entropy combined with dynamic connectivity features classifies MDD vs. healthy controls with an accuracy of 0.82 for leave-one-out cross-validation. This classifier has specificity and sensitivity of 0.79 and 0.84, respectively.

Predicting Biological Gender and Intelligence From fMRI via Dynamic Functional Connectivity.

OBJECTIVE: This paper explores the predictive capability of dynamic functional connectivity extracted from functional magnetic resonance imaging (fMRI) of the human brain, in contrast to static connectivity used in past research. METHODS: Several state-of-the-art features extracted from static functional connectivity of the brain are employed to predict biological gender and intelligence using publicly available Human Connectome Project (HCP) database. Next, a novel tensor parallel factor (PARAFAC) decomposition model is proposed to decompose sequence of dynamic connectivity matrices into common connectivity components that are orthonormal to each other, common time-courses, and corresponding distinct subject-wise weights. The subject-wise loading of the components are employed to predict biological gender and intelligence using a random forest classifier (respectively, regressor) using 5-fold cross-validation. RESULTS: The results demonstrate that dynamic functional connectivity can indeed classify biological gender with a high accuracy (0.94, where male identification accuracy was 0.87 and female identification accuracy was 0.97). It can also predict intelligence with less normalized mean square error (0.139 for fluid intelligence and 0.031 for fluid ability metrics) compared with other functional connectivity measures (the nearest mean square error were 0.147 and 0.037 for fluid intelligence and fluid ability metrics, respectively, using static connectivity approaches). CONCLUSION: Our work is an important milestone for the understanding of non-stationary behavior of hemodynamic blood-oxygen level dependent (BOLD) signal in brain and how they are associated with biological gender and intelligence. SIGNIFICANCE: The paper demonstrates that dynamic behavior of brain can contribute substantially towards forming a fingerprint of biological gender and intelligence.

Molecular and DNA Artificial Neural Networks via Fractional Coding.

This article considers implementation of artificial neural networks (ANNs) using molecular computing and DNA based on fractional coding. Prior work had addressed molecular two-layer ANNs with binary inputs and arbitrary weights. In prior work using fractional coding, a simple molecular perceptron that computes sigmoid of scaled weighted sum of the inputs was presented where the inputs and the weights lie between [-1,1]. Even for computing the perceptron, the prior approach suffers from two major limitations. First, it cannot compute the sigmoid of the weighted sum, but only the sigmoid of the scaled weighted sum. Second, many machine learning applications require the coefficients to be arbitrarily positive and negative numbers that are not bounded between [-1,1]; such numbers cannot be handled by the prior perceptron using fractional coding. This paper makes four contributions. First molecular perceptrons that can handle arbitrary weights and can compute sigmoid of the weighted sums are presented. Thus, these molecular perceptrons are ideal for regression applications and multi-layer ANNs. A new molecular divider is introduced and is used to compute sigmoid(ax) where . Second, based on fractional coding, a molecular artificial neural network (ANN) with one hidden layer is presented. Third, a trained ANN classifier with one hidden layer from seizure prediction application from electroencephalogram is mapped to molecular reactions and DNA and their performances are presented. Fourth, molecular activation functions for rectified linear unit (ReLU) and softmax are also presented.

Sub-graph entropy based network approaches for classifying adolescent obsessive-compulsive disorder from resting-state functional MRI.

This paper presents a novel approach for classifying obsessive-compulsive disorder (OCD) in adolescents from resting-state fMRI data. Currently, the state-of-the-art for diagnosing OCD in youth involves interviews with adolescent patients and their parents by an experienced clinician, symptom rating scales based on Diagnostic and Statistical Manual of Mental Disorders (DSM), and behavioral observation. Discovering signal processing and network-based biomarkers from functional magnetic resonance imaging (fMRI) scans of patients has the potential to assist clinicians in their diagnostic assessments of adolescents suffering from OCD. This paper investigates the clinical diagnostic utility of a set of univariate, bivariate and multivariate features extracted from resting-state fMRI using an information-theoretic approach in 15 adolescents with OCD and 13 matched healthy controls. Results indicate that an information-theoretic approach based on sub-graph entropy is capable of classifying OCD vs. healthy subjects with high accuracy. Mean time-series were extracted from 85 brain regions and were used to calculate Shannon wavelet entropy, Pearson correlation matrix, network features and sub-graph entropy. In addition, two special cases of sub-graph entropy, namely node and edge entropy, were investigated to identify important brain regions and edges from OCD patients. A leave-one-out cross-validation method was used for the final predictor performance. The proposed methodology using differential sub-graph (edge) entropy achieved an accuracy of 0.89 with specificity 1 and sensitivity 0.80 using leave-one-out cross-validation with in-fold feature ranking and selection. The high classification accuracy indicates the predictive power of the sub-network as well as edge entropy metric.

Discriminative Ratio of Spectral Power and Relative Power Features Derived via Frequency-Domain Model Ratio With Application to Seizure Prediction.

The ratio of spectral power in two different bands and relative band power have been shown to be sometimes more discriminative features than the spectral power in a specific band for binary classification of a time series for seizure prediction. However, why and which ratio of spectral power and relative power features are better discriminators than a band power have not been understood. While general answers to why and which are difficult, this paper partially addresses the answer to these questions. Using auto-regressive modeling, this paper, for the first time, theoretically explains that for high signal-to-noise ratio (SNR) cases, the ratio features may sometime amplify the discriminability of one of the two states in a time series, as compared with a band power. This paper, also for the first time, introduces a novel frequency-domain model ratio (FDMR) that can be used to select the two frequency bands. The FDMR computes the ratio of the frequency responses of the two auto-regressive model filters that correspond to two different states. It is shown that the ratio implicitly cancels the effect of change of variance of the white noise that is input to the auto-regressive model in a non-stationary environment for high SNR conditions. It is also shown that under certain sufficient but not necessary conditions, the ratio of the spectral power and the relative band power, i.e., the band power divided by the total power spectral density, can be better discriminators than band power. Synthesized data and scalp EEG data from the MIT Physionet for patient-specific seizure prediction are used to explain why the ratios of spectral power obtained by a ranking algorithm in the prior literature satisfy the sufficient conditions for amplification of the ratio feature derived in this paper.

Ranking Regions, Edges and Classifying Tasks in Functional Brain Graphs by Sub-Graph Entropy.

This paper considers analysis of human brain networks or graphs constructed from time-series collected from functional magnetic resonance imaging (fMRI). In the network of time-series, the nodes describe the regions and the edge weights correspond to the absolute values of correlation coefficients of the time-series of the two nodes associated with the edges. The paper introduces a novel information-theoretic metric, referred as sub-graph entropy, to measure uncertainty associated with a sub-graph. Nodes and edges constitute two special cases of sub-graph structures. Node and edge entropies are used in this paper to rank regions and edges in a functional brain network. The paper analyzes task-fMRI data collected from 475 subjects in the Human Connectome Project (HCP) study for gambling and emotion tasks. The proposed approach is used to rank regions and edges associated with these tasks. The differential node (edge) entropy metric is defined as the difference of the node (edge) entropy corresponding to two different networks belonging to two different classes. Differential entropy of nodes and edges are used to rank top regions and edges associated with the two classes of data. Using top node and edge entropy features separately, two-class classifiers are designed using support vector machine (SVM) with radial basis function (RBF) kernel and leave-one-out method to classify time-series for emotion task vs. no-task, gambling task vs. no-task and emotion task vs. gambling task. Using node entropies, the SVM classifier achieves classification accuracies of 0.96, 0.97 and 0.98, respectively. Using edge entropies, the classifier achieves classification accuracies of 0.91, 0.96 and 0.94, respectively.

RETOUCH: The Retinal OCT Fluid Detection and Segmentation Benchmark and Challenge.

Retinal swelling due to the accumulation of fluid is associated with the most vision-threatening retinal diseases. Optical coherence tomography (OCT) is the current standard of care in assessing the presence and quantity of retinal fluid and image-guided treatment management. Deep learning methods have made their impact across medical imaging, and many retinal OCT analysis methods have been proposed. However, it is currently not clear how successful they are in interpreting the retinal fluid on OCT, which is due to the lack of standardized benchmarks. To address this, we organized a challenge RETOUCH in conjunction with MICCAI 2017, with eight teams participating. The challenge consisted of two tasks: fluid detection and fluid segmentation. It featured for the first time: all three retinal fluid types, with annotated images provided by two clinical centers, which were acquired with the three most common OCT device vendors from patients with two different retinal diseases. The analysis revealed that in the detection task, the performance on the automated fluid detection was within the inter-grader variability. However, in the segmentation task, fusing the automated methods produced segmentations that were superior to all individual methods, indicating the need for further improvements in the segmentation performance.

Classification of Major Depressive Disorder from Resting-State fMRI.

Major Depressive Disorder (MDD) is a very serious mental illness that can affect the daily lives of patients. Accurate diagnosis of this disorder is necessary for planning individualized treatment. However, diagnosing MDD requires the clinicians to personally interview the subjects and rate the symptoms based on Diagnostic and Statistical Manual of Mental Disorders (DSM), which can be very time consuming. Discovering quantifiable signals and biomarkers associated with MDD using functional magnetic resonance imaging (fMRI) scans of patients have the potential to assist the clinicians in their assessment. This paper explores the use of resting-state functional connectivity and network features to classify MDD vs. healthy subjects. For each subject, mean time-series are extracted from 85 brain regions and they are decomposed to 4-frequency bands. Mean time-series for each of the frequency bands are utilized to compute the Pearson correlation and network characteristics. Features are selected separately from correlation and network characteristics using Minimum Redundancy Maximum Relevance (mRMR) to create the final classifier. The proposed scheme achieves 79% accuracy (65 out of 82 subjects classified correctly) with 86% sensitivity (42 out of 49 MDD subjects identified correctly) and 70% specificity (23 out of 33 controls identified correctly) using leave-one-out classification with in-fold feature selection. Pearson correlation had the highest discrimination in band 0.015-0.03 Hz and network based features had the highest discrimination in band 0.03-0.06 Hz for distinguishing MDD vs. healthy subjects.