Classification of cocaine-dependent participants with dynamic functional connectivity from functional magnetic resonance imaging data.

Static functional connectivity (FC) analyses based on functional magnetic resonance imaging (fMRI) data have been extensively explored for studying various psychiatric conditions in the brain, including cocaine addiction. A recently emerging, more powerful technique, dynamic functional connectivity (DFC), studies how the FC dynamics change during the course of the fMRI experiments. The aim in this paper was to develop a computational approach, using a machine learning framework, to determine if DFC features were more successful than FC features in the classification of cocaine-dependent patients and healthy controls. fMRI data were obtained from of 25 healthy and 58 cocaine-dependent participants while performing a motor response inhibition task, stop signal task. Group independent component analysis was carried out on all participant data to compute spatially independent components (ICs). Eight ICs were selected manually as relevant brain networks, which were used to classify healthy versus cocaine-dependent participants. FC and DFC measures of the chosen IC pairs were used as features for the classification algorithm. Support Vector Machines were used for both feature selection/reduction and participant classification. Based on DFC with only seven IC pairs, participants were successfully classified with 95% accuracy (and with 90% accuracy with three IC pairs), whereas static FC yielded only 81% accuracy. Visual, sensorimotor, default mode, and executive control networks, amygdala, and insula played the most significant role in the DFC-based classification. These findings support the use of DFC-based classification of fMRI data as a potential biomarker for the identification of cocaine dependence.

Exploring brain mechanisms underlying Gulf War Illness with group ICA based analysis of fMRI resting state networks.

Around 200,000 veterans (up to 32% of those deployed) of the 1991 Gulf War (GW) suffer from GW illness (GWI), which is characterized by multiple deficits in cognitive, affective, sensory and nociception domains. In this study we employed resting state fMRI (rsfMRI) to map impairments in brain function in GWI with advanced network analysis. RsfMRI data was obtained from 60 GWI veterans and 30 age-matched military controls. Group independent component analysis (GICA) was conducted to probe the functional connectivity networks in all 90 subjects. GICA revealed impaired functional connectivity (FC) in GWI veterans between a number of brain function networks consistent with their self-reported symptoms. GWI veterans exhibited impaired FC between language networks, and sensory input networks of all modalities as well as motor output networks. GWI veterans also exhibited impaired FC between different sensory perception and motor networks, and between different networks in the sensorimotor domain. These FC impairments provide putative mechanism of central nervous system dysfunction in GWI.

Successful classification of cocaine dependence using brain imaging: a generalizable machine learning approach.

BACKGROUND: Neuroimaging studies have yielded significant advances in the understanding of neural processes relevant to the development and persistence of addiction. However, these advances have not explored extensively for diagnostic accuracy in human subjects. The aim of this study was to develop a statistical approach, using a machine learning framework, to correctly classify brain images of cocaine-dependent participants and healthy controls. In this study, a framework suitable for educing potential brain regions that differed between the two groups was developed and implemented. Single Photon Emission Computerized Tomography (SPECT) images obtained during rest or a saline infusion in three cohorts of 2-4 week abstinent cocaine-dependent participants (n = 93) and healthy controls (n = 69) were used to develop a classification model. An information theoretic-based feature selection algorithm was first conducted to reduce the number of voxels. A density-based clustering algorithm was then used to form spatially connected voxel clouds in three-dimensional space. A statistical classifier, Support Vectors Machine (SVM), was then used for participant classification. Statistically insignificant voxels of spatially connected brain regions were removed iteratively and classification accuracy was reported through the iterations. RESULTS: The voxel-based analysis identified 1,500 spatially connected voxels in 30 distinct clusters after a grid search in SVM parameters. Participants were successfully classified with 0.88 and 0.89 F-measure accuracies in 10-fold cross validation (10xCV) and leave-one-out (LOO) approaches, respectively. Sensitivity and specificity were 0.90 and 0.89 for LOO; 0.83 and 0.83 for 10xCV. Many of the 30 selected clusters are highly relevant to the addictive process, including regions relevant to cognitive control, default mode network related self-referential thought, behavioral inhibition, and contextual memories. Relative hyperactivity and hypoactivity of regional cerebral blood flow in brain regions in cocaine-dependent participants are presented with corresponding level of significance. CONCLUSIONS: The SVM-based approach successfully classified cocaine-dependent and healthy control participants using voxels selected with information theoretic-based and statistical methods from participants' SPECT data. The regions found in this study align with brain regions reported in the literature. These findings support the future use of brain imaging and SVM-based classifier in the diagnosis of substance use disorders and furthering an understanding of their underlying pathology.

Design of a novel biomedical signal processing and analysis tool for functional neuroimaging.

In this paper, a MATLAB-based graphical user interface (GUI) software tool for general biomedical signal processing and analysis of functional neuroimaging data is introduced. Specifically, electroencephalography (EEG) and electrocardiography (ECG) signals can be processed and analyzed by the developed tool, which incorporates commonly used temporal and frequency analysis methods. In addition to common methods, the tool also provides non-linear chaos analysis with Lyapunov exponents and entropies; multivariate analysis with principal and independent component analyses; and pattern classification with discriminant analysis. This tool can also be utilized for training in biomedical engineering education. This easy-to-use and easy-to-learn, intuitive tool is described in detail in this paper.

GPU accelerated dynamic functional connectivity analysis for functional MRI data.

Recent advances in multi-core processors and graphics card based computational technologies have paved the way for an improved and dynamic utilization of parallel computing techniques. Numerous applications have been implemented for the acceleration of computationally-intensive problems in various computational science fields including bioinformatics, in which big data problems are prevalent. In neuroimaging, dynamic functional connectivity (DFC) analysis is a computationally demanding method used to investigate dynamic functional interactions among different brain regions or networks identified with functional magnetic resonance imaging (fMRI) data. In this study, we implemented and analyzed a parallel DFC algorithm based on thread-based and block-based approaches. The thread-based approach was designed to parallelize DFC computations and was implemented in both Open Multi-Processing (OpenMP) and Compute Unified Device Architecture (CUDA) programming platforms. Another approach developed in this study to better utilize CUDA architecture is the block-based approach, where parallelization involves smaller parts of fMRI time-courses obtained by sliding-windows. Experimental results showed that the proposed parallel design solutions enabled by the GPUs significantly reduce the computation time for DFC analysis. Multicore implementation using OpenMP on 8-core processor provides up to 7.7× speed-up. GPU implementation using CUDA yielded substantial accelerations ranging from 18.5× to 157× speed-up once thread-based and block-based approaches were combined in the analysis. Proposed parallel programming solutions showed that multi-core processor and CUDA-supported GPU implementations accelerated the DFC analyses significantly. Developed algorithms make the DFC analyses more practical for multi-subject studies with more dynamic analyses.

Default-mode-like network activation in awake rodents.

During wakefulness and in absence of performing tasks or sensory processing, the default-mode network (DMN), an intrinsic central nervous system (CNS) network, is in an active state. Non-human primate and human CNS imaging studies have identified the DMN in these two species. Clinical imaging studies have shown that the pattern of activity within the DMN is often modulated in various disease states (e.g., Alzheimer's, schizophrenia or chronic pain). However, whether the DMN exists in awake rodents has not been characterized. The current data provides evidence that awake rodents also possess 'DMN-like' functional connectivity, but only subsequent to habituation to what is initially a novel magnetic resonance imaging (MRI) environment as well as physical restraint. Specifically, the habituation process spanned across four separate scanning sessions (Day 2, 4, 6 and 8). At Day 8, significant (p<0.05) functional connectivity was observed amongst structures such as the anterior cingulate (seed region), retrosplenial, parietal, and hippocampal cortices. Prior to habituation (Day 2), functional connectivity was only detected (p<0.05) amongst CNS structures known to mediate anxiety (i.e., anterior cingulate (seed region), posterior hypothalamic area, amygdala and parabracial nucleus). In relating functional connectivity between cingulate-default-mode and cingulate-anxiety structures across Days 2-8, a significant inverse relationship (r = -0.65, p = 0.0004) was observed between these two functional interactions such that increased cingulate-DMN connectivity corresponded to decreased cingulate anxiety network connectivity. This investigation demonstrates that the cingulate is an important component of both the rodent DMN-like and anxiety networks.

A Data-Driven Investigation of Gray Matter-Function Correlations in Schizophrenia during a Working Memory Task.

The brain is a vastly interconnected organ and methods are needed to investigate its long range structure(S)-function(F) associations to better understand disorders such as schizophrenia that are hypothesized to be due to distributed disconnected brain regions. In previous work we introduced a methodology to reduce the whole brain S-F correlations to a histogram and here we reduce the correlations to brain clusters. The application of our approach to sMRI [gray matter (GM) concentration maps] and functional magnetic resonance imaging data (general linear model activation maps during Encode and Probe epochs of a working memory task) from patients with schizophrenia (SZ, n = 100) and healthy controls (HC, n = 100) presented the following results. In HC the whole brain correlation histograms for GM-Encode and GM-Probe overlap for Low and Medium loads and at High the histograms separate, but in SZ the histograms do not overlap for any of the load levels and Medium load shows the maximum difference. We computed GM-F differential correlation clusters using activation for Probe Medium, and they included regions in the left and right superior temporal gyri, anterior cingulate, cuneus, middle temporal gyrus, and the cerebellum. Inter-cluster GM-Probe correlations for Medium load were positive in HC but negative in SZ. Within group inter-cluster GM-Encode and GM-Probe correlation comparisons show no differences in HC but in SZ differences are evident in the same clusters where HC vs. SZ differences occurred for Probe Medium, indicating that the S-F integrity during Probe is aberrant in SZ. Through a data-driven whole brain analysis approach we find novel brain clusters and show how the S-F differential correlation changes during Probe and Encode at three memory load levels. Structural and functional anomalies have been extensively reported in schizophrenia and here we provide evidences to suggest that evaluating S-F associations can provide important additional information.

Paradigm shift in translational neuroimaging of CNS disorders.

During the last two decades, functional neuroimaging technology, especially functional magnetic resonance imaging (fMRI), has improved tremendously, with new attention towards resting-state functional connectivity of the brain. This development has allowed scientists to study changes in brain structure and function, and probe these two properties under conditions of evoked stimulation, disease and drug administration. In the domain of functional imaging, the identification and characterization of central nervous system (CNS) functional networks have emerged as potential biomarkers for CNS disorders in humans. Recent attempts to translate clinical neuroimaging methodology to preclinical studies have also been carried out, which offer new opportunities in translational neuroscience research. In this paper, we review recent developments in structural and functional MRI and their use to probe functional connectivity in various CNS disorders such as schizophrenia, mood disorders, Alzheimer's disease (AD) and pain.

A method for evaluating dynamic functional network connectivity and task-modulation: application to schizophrenia.

OBJECTIVE: In this paper, we develop a dynamic functional network connectivity (FNC) analysis approach using correlations between windowed time-courses of different brain networks (components) estimated via spatial independent component analysis (sICA). We apply the developed method to fMRI data to evaluate it and to study task-modulation of functional connections. MATERIALS AND METHODS: We study the theoretical basis of the approach, perform a simulation analysis and apply it to fMRI data from schizophrenia patients (SP) and healthy controls (HC). Analyses on the fMRI data include: (a) group sICA to determine regions of significant task-related activity, (b) static and dynamic FNC analysis among these networks by using maximal lagged-correlation and time-frequency analysis, and (c) HC-SP group differences in functional network connections and in task-modulation of these connections. RESULTS: This new approach enables an assessment of task-modulation of connectivity and identifies meaningful inter-component linkages and differences between the two study groups during performance of an auditory oddball task (AOT). The static FNC results revealed that connectivities involving medial visual-frontal, medial temporal-medial visual, parietal-medial temporal, parietal-medial visual and medial temporal-anterior temporal were significantly greater in HC, whereas only the right lateral fronto-parietal (RLFP)-orbitofrontal connection was significantly greater in SP. The dynamic FNC revealed that task-modulation of motor-frontal, RLFP-medial temporal and posterior default mode (pDM)-parietal connections were significantly greater in SP, and task modulation of orbitofrontal-pDM and medial temporal-frontal connections were significantly greater in HC (all P < 0.05). CONCLUSION: The task-modulation of dynamic FNC provided findings and differences between the two groups that are consistent with the existing hypothesis that schizophrenia patients show less segregated motor, sensory, cognitive functions and less segregated default mode network activity when engaged with a task. Dynamic FNC, based on sICA, provided additional results which are different than, but complementary to, those of static FNC. For example, it revealed dynamic changes in default mode network connectivities with other regions which were significantly different in schizophrenia in terms of task-modulation, findings which were not possible to discover by static FNC.

Application of FT-based MMSE deconvolution method for cerebral blood flow measurement in patients with leukoaraiosis.

INTRODUCTION: The bolus-tracking (BT) technique is the most popular perfusion-weighted (PW) dynamic susceptibility contrast MRI method used for estimating cerebral blood flow (CBF), cerebral blood volume and mean transit time. The BT technique uses a convolution model that establishes the input-output relationship between blood flow and the vascular tracer concentration. Singular value decomposition (SVD)- and Fourier transform (FT)-based deconvolution methods are popular and widely used for estimating PW MRI parameters. However, from the published literature, it appears that SVD is more widely accepted than other methods. In a previous article, an FT-based minimum mean-squared error (MMSE) technique was proposed and simulation experiments were performed to compare it with the well-established circular SVD (oSVD) method. In this study, the FT-based MMSE method has been used to estimate relative CBF (rCBF) in 13 patients with white matter lesions (WMLs) (leukoaraiosis), and results are compared with the widely used oSVD method. MATERIALS AND METHODS: Thirteen patients with leukoaraiosis were imaged on a 1.5-T Siemens whole-body scanner. After acquiring the localizer and structural scans consisting of FLAIR (fluid attenuated with inversion recovery), T(1)-weighted and T(2)-weighted images, perfusion study was implemented as part of the MRI protocol. For each patient and method, two values were calculated: (a) rCBF for normal white matter (NWM) ROI, obtained by dividing the average CBF value in NWM ROI with average CBF in gray matter (GM) ROI, and (b) rCBF for WML ROI, obtained by dividing the average CBF value in WML ROI with average CBF in GM ROI. Results for the two deconvolution methods were computed. RESULTS AND DISCUSSION: A significant (P<.05) decrease in estimated rCBF was observed in the WML in all the patients using the MMSE method, while for the oSVD method, the decrease was observed in all but one patient. Initial results suggest that the MMSE method is comparable to the oSVD method for estimating rCBF in NMW while it may be better than oSVD for estimating rCBF in lesions of low flow. Studies involving a larger patient population may be required to further validate the findings of this work.

Characterization of BBB permeability in a preclinical model of cryptococcal meningoencephalitis using magnetic resonance imaging.

Cryptococcus neoformans meningoencephalitis (CNME) is a leading fungal cause of death among acquired immunodeficiency syndrome patients. Innovative preclinical systems that permit high throughput in vivo evaluation of novel agents are desperately needed. Magnetic resonance imaging (MRI) was evaluated as a tool to develop a rat model of CNME and to quantify noninvasively blood-brain barrier (BBB) disruption secondary to this disease. The aim of this study was to identify MRI changes compared with histopathology and fungal burden measurements as potential biomarkers. A well-characterized strain of C neoformans (CN) var grubii was used to infect rats using intravenous inoculation. An inoculum-finding study was performed by infecting rats with 10(3), 10(5), and 10(7) colony-forming units (CFUs). Animals underwent dynamic MRI on days 4 and 7 after inoculation. An inoculum-confirming study was performed by infecting rats with 10(5), 10(6), and 10(7) CFU and fungal burden was determined in the brain, lung, and spleen. Animals infected with 10(7) CFU of CN developed lesions that appeared hyperintense on T2-weighted images on day 4. The histopathology results correlated well with MRI data. Diffusion weighted and permeability estimates were 1.4 and 6.1-fold higher, respectively, in lesions compared with healthy tissue. Magnetic resonance imaging is a promising preclinical tool to evaluate effects of antifungal and adjunctive agents.

Normobaric hyperoxia inhibits NADPH oxidase-mediated matrix metalloproteinase-9 induction in cerebral microvessels in experimental stroke.

Matrix metalloproteinase-9 (MMP-9) and NADPH oxidase contribute to blood-brain barrier (BBB) disruption after ischemic stroke. We have previously shown that normobaric hyperoxia (NBO) treatment reduces MMP-9 and oxygen free radical generation in ischemic brain. In this study, we tested the hypothesis that NBO protects the BBB through inhibiting NADPH oxidase-mediated MMP-9 induction in transient focal cerebral ischemia. Male Sprague-Dawley rats (n = 69) were given NBO (95% O2) or normoxia (21% O2) during 90-min filament occlusion of the middle cerebral artery. Cerebral microvessels were isolated for analyzing MMP-9 and NADPH oxidase. BBB damage was non-invasively quantified with magnetic resonance imaging. In normoxic rats, both NADPH oxidase catalytic subunit gp91(phox) and MMP-9 expression were up-regulated in ischemic hemispheric microvessels after 90-min middle cerebral artery occlusion with 22.5 h reperfusion. Inhibition of NADPH oxidase with apocynin reduced the MMP-9 increase, indicating a causal link between NADPH oxidase-derived superoxide and MMP-9 induction. NBO treatment inhibited gp91(phox) expression, NADPH oxidase activity, and MMP-9 induction, which led to significantly less BBB damage and brain edema in the ischemic brain. These results suggest that gp91(phox) containing NADPH oxidase plays an important role in MMP-9 induction in ischemic BBB microvasculature, and that NBO treatment may attenuate MMP-9 induction and brain edema through inhibiting NADPH oxidase after transient cerebral ischemia.

Cerebral blood flow estimation from perfusion-weighted MRI using FT-based MMSE filtering method.

INTRODUCTION: Perfusion-weighted MRI can be used for estimating blood flow parameters using bolus tracking technique based on dynamic susceptibility contrast MRI. In order to extract flow parameters, several deconvolution techniques have been proposed, of which the singular value decomposition (SVD) and Fourier transform (FT)-based techniques are more popular and widely used. In this work, an FT-based method has been proposed that involves derivation of an optimal shaped filter (defined as a filter function) estimated using minimum mean-squared error (MMSE) technique in the frequency domain. The proposed technique has been compared with the well-established SVD technique using simulation experiments. SIMULATION METHODS: Simulation was performed in multiple steps. An arterial input function (AIF) was first defined based on a certain blood flow value. The T2* signal change was then derived from this AIF, and noise was added to the signal. Then, a unique and optimal shaped filter function Phi(f) was derived in order to obtain the best estimate of scaled residue function. One way is by minimizing the mean-squared error between the noiseless and noisy scaled residue function, i.e., using an MMSE method. The effect of low and moderate noise and distorted AIF on cerebral blood flow (CBF) estimates was obtained by using FT-based MMSE method. Results were compared with the SVD technique. In this work, SVD technique was assumed to be the standard reference deconvolution technique. RESULTS AND DISCUSSION: For low-noise condition, the FT-based technique was more stable than the SVD technique, while for moderate noise, both techniques consistently underestimated CBF. SVD technique was found to be more stable in presence of AIF distortions. However, SVD technique was found to be unstable due to AIF delay compared to the FT-based MMSE method. The shaped filter function was found to be sensitive to effect of AIF distortions.

Statistical adaptive sensing by detectors with spectrally overlapping bands.

It has recently been reported that by using a spectral-tuning algorithm, the photocurrents of multiple detectors with spectrally overlapping responsivities can be optimally combined to synthesize, within certain limits, the response of a detector with an arbitrary responsivity. However, it is known that the presence of noise in the photocurrent can degrade the performance of this algorithm significantly, depending on the choice of the responsivity spectrum to be synthesized. We generalize this algorithm to accommodate noise. The results are applied to quantum-dot mid-infrared detectors with bias-dependent spectral responses. Simulation and experiment are used to show the ability of the algorithm to reduce the adverse effect of noise on its spectral-tuning capability.