Arrow Research search

Author name cluster

Mark S. Cohen

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

18 papers
1 author row

Possible papers

18

YNICL Journal 2017 Journal Article

Assessing neural tuning for object perception in schizophrenia and bipolar disorder with multivariate pattern analysis of fMRI data

  • Eric A. Reavis
  • Junghee Lee
  • Jonathan K. Wynn
  • Stephen A. Engel
  • Mark S. Cohen
  • Keith H. Nuechterlein
  • David C. Glahn
  • Lori L. Altshuler

INTRODUCTION: Deficits in visual perception are well-established in schizophrenia and are linked to abnormal activity in the lateral occipital complex (LOC). Related deficits may exist in bipolar disorder. LOC contains neurons tuned to object features. It is unknown whether neural tuning in LOC or other visual areas is abnormal in patients, contributing to abnormal perception during visual tasks. This study used multivariate pattern analysis (MVPA) to investigate perceptual tuning for objects in schizophrenia and bipolar disorder. METHODS: Fifty schizophrenia participants, 51 bipolar disorder participants, and 47 matched healthy controls completed five functional magnetic resonance imaging (fMRI) runs of a perceptual task in which they viewed pictures of four different objects and an outdoor scene. We performed classification analyses designed to assess the distinctiveness of activity corresponding to perception of each stimulus in LOC (a functionally localized region of interest). We also performed similar classification analyses throughout the brain using a searchlight technique. We compared classification accuracy and patterns of classification errors across groups. RESULTS: Stimulus classification accuracy was significantly above chance in all groups in LOC and throughout visual cortex. Classification errors were mostly within-category confusions (e.g., misclassifying one chair as another chair). There were no group differences in classification accuracy or patterns of confusion. CONCLUSIONS: The results show for the first time MVPA can be used successfully to classify individual perceptual stimuli in schizophrenia and bipolar disorder. However, the results do not provide evidence of abnormal neural tuning in schizophrenia and bipolar disorder.

Details DOI

YNICL Journal 2016 Journal Article

Alpha desynchronization and fronto­parietal connectivity during spatial working memory encoding deficits in ADHD: A simultaneous EEG­fMRI study

  • Agatha Lenartowicz
  • Steven Lu
  • Cameron Rodriguez
  • Edward P. Lau
  • Patricia D. Walshaw
  • James T. McCracken
  • Mark S. Cohen
  • Sandra K. Loo

The underlying mechanisms of alpha band (8–12Hz) neural oscillations are of importance to the functioning of attention control systems as well as to neuropsychiatric conditions that are characterized by deficits of that system, such as attention deficit hyperactivity disorder (ADHD). The objectives of the present study were to test if visual encoding-related alpha event-related desynchronization (ERD) correlates with fronto-parieto-occipital connectivity, and whether this is disrupted in ADHD during spatial working memory (SWM) performance. We acquired EEG concurrently with fMRI in thirty boys (12–16yrs. old, 15 with ADHD), during SWM encoding. Psychophysiological connectivity analyses indicated that alpha ERD during SWM encoding was associated with both occipital activation and fronto-parieto-occipital functional connectivity, a finding that expands on prior associations between alpha ERD and occipital activation. This finding provides novel support for the interpretation of alpha ERD (and the associated changes in occipital activation) as a phenomenon that involves, and perhaps arises as a result of, top-down network interactions. Alpha ERD was associated less strongly with occipital activity, but associated more strongly with fronto-parieto-occipital connectivity in ADHD, consistent with a compensatory attentional response. Additionally, we illustrate that degradation of EEG data quality by MRI-amplified motion artifacts is robust to existing cleaning algorithms and is significantly correlated with hyperactivity symptoms and the ADHD Combined Type diagnosis. We conclude that persistent motion-related MR artifacts in EEG data can increase variance and introduce bias in interpretation of group differences in populations characterized by hypermobility — a clear limitation of current-state EEG-fMRI methodology.

Details DOI

YNIMG Journal 2014 Journal Article

Non-negative matrix factorization of multimodal MRI, fMRI and phenotypic data reveals differential changes in default mode subnetworks in ADHD

  • Ariana Anderson
  • Pamela K. Douglas
  • Wesley T. Kerr
  • Virginia S. Haynes
  • Alan L. Yuille
  • Jianwen Xie
  • Ying Nian Wu
  • Jesse A. Brown

In the multimodal neuroimaging framework, data on a single subject are collected from inherently different sources such as functional MRI, structural MRI, behavioral and/or phenotypic information. The information each source provides is not independent; a subset of features from each modality maps to one or more common latent dimensions, which can be interpreted using generative models. These latent dimensions, or “topics, ” provide a sparse summary of the generative process behind the features for each individual. Topic modeling, an unsupervised generative model, has been used to map seemingly disparate features to a common domain. We use Non-Negative Matrix Factorization (NMF) to infer the latent structure of multimodal ADHD data containing fMRI, MRI, phenotypic and behavioral measurements. We compare four different NMF algorithms and find that the sparsest decomposition is also the most differentiating between ADHD and healthy patients. We identify dimensions that map to interpretable, recognizable dimensions such as motion, default mode network activity, and other such features of the input data. For example, structural and functional graph theory features related to default mode subnetworks clustered with the ADHD-Inattentive diagnosis. Structural measurements of the default mode network (DMN) regions such as the posterior cingulate, precuneus, and parahippocampal regions were all related to the ADHD-Inattentive diagnosis. Ventral DMN subnetworks may have more functional connections in ADHD-I, while dorsal DMN may have less. ADHD topics are dependent upon diagnostic site, suggesting diagnostic differences across geographic locations. We assess our findings in light of the ADHD-200 classification competition, and contrast our unsupervised, nominated topics with previously published supervised learning methods. Finally, we demonstrate the validity of these latent variables as biomarkers by using them for classification of ADHD in 730 patients. Cumulatively, this manuscript addresses how multimodal data in ADHD can be interpreted by latent dimensions.

Details DOI

YNIMG Journal 2014 Journal Article

The utility of data-driven feature selection: Re: Chu et al. 2012

  • Wesley T. Kerr
  • Pamela K. Douglas
  • Ariana Anderson
  • Mark S. Cohen

The recent Chu et al. (2012) manuscript discusses two key findings regarding feature selection (FS): (1) data driven FS was no better than using whole brain voxel data and (2) a priori biological knowledge was effective to guide FS. Use of FS is highly relevant in neuroimaging-based machine learning, as the number of attributes can greatly exceed the number of exemplars. We strongly endorse their demonstration of both of these findings, and we provide additional important practical and theoretical arguments as to why, in their case, the data-driven FS methods they implemented did not result in improved accuracy. Further, we emphasize that the data-driven FS methods they tested performed approximately as well as the all-voxel case. We discuss why a sparse model may be favored over a complex one with similar performance. We caution readers that the findings in the Chu et al. report should not be generalized to all data-driven FS methods.

Details DOI

YNIMG Journal 2012 Journal Article

Echo planar imaging before and after fMRI: A personal history

  • Mark S. Cohen
  • Franz Schmitt

Echo-planar imaging (EPI) plays a crucial role in functional MRI. Focusing especially on the period from 1988 to 1992, the authors offer personal recollections, on the development of practical means of deploying EPI, the people that participated, and its impact on MRI in general.

Details DOI

YNIMG Journal 2011 Journal Article

Altered dynamic coupling of lateral occipital complex during visual perception in schizophrenia

  • Philippe-Olivier Harvey
  • Junghee Lee
  • Mark S. Cohen
  • Stephen A. Engel
  • David C. Glahn
  • Keith H. Nuechterlein
  • Jonathan K. Wynn
  • Michael F. Green

Introduction There is mounting evidence that visual perception abnormalities in schizophrenia are partly explained by a dysfunction of the lateral occipital complex (LO). We previously demonstrated that schizophrenia patients had broader topography and reduced magnitude of activity of LO. However, the functional connectivity of LO with other brain regions during visual perception has not been directly investigated in schizophrenia. Materials and methods Eighteen patients with schizophrenia and eighteen matched controls performed a backward masking task during functional magnetic resonance imaging (fMRI). Stimulus onset asynchronies were manipulated to change the level of target visibility. To examine connectivity with LO function we conducted psychophysiological interactions (PPI) analyses using: 1) a region of interest (ROI) approach and 2) a whole brain analysis. ROIs were defined based on a contrast of trials on which a target was presented versus null trials in which no stimuli were presented. Results Eleven ROIs were identified. Both groups showed similar strength of coupling between LO and the 11 ROIs when visibility was not taken into account. Healthy controls showed clear changes in coupling between LO and prefrontal and parietal regions as a function of target visibility (higher coupling with more visible targets). In comparison, patients showed reduced dynamic coupling with LO in the right superior frontal gyrus (significant after correcting for multiple comparisons) and a trend for reduced coupling in the left precuneus and left inferior frontal regions. Whole brain analysis identified additional regions that showed dynamic coupling with LO in healthy controls, but not in patients. Discussion The increased coupling between LO and higher-level parietal and prefrontal regions during visual awareness in healthy controls likely reflects visual reentrant processing. The lack of modulation of coupling between LO and key prefrontal and parietal regions found in schizophrenia may partly reflect abnormalities in LO tuning. The altered LO coupling may contribute to visual perception abnormalities in schizophrenia.

Details DOI

YNIMG Journal 2011 Journal Article

Common component classification: What can we learn from machine learning?

  • Ariana Anderson
  • Jennifer S. Labus
  • Eduardo P. Vianna
  • Emeran A. Mayer
  • Mark S. Cohen

Machine learning methods have been applied to classifying fMRI scans by studying locations in the brain that exhibit temporal intensity variation between groups, frequently reporting classification accuracy of 90% or better. Although empirical results are quite favorable, one might doubt the ability of classification methods to withstand changes in task ordering and the reproducibility of activation patterns over runs, and question how much of the classification machines' power is due to artifactual noise versus genuine neurological signal. To examine the true strength and power of machine learning classifiers we create and then deconstruct a classifier to examine its sensitivity to physiological noise, task reordering, and across-scan classification ability. The models are trained and tested both within and across runs to assess stability and reproducibility across conditions. We demonstrate the use of independent components analysis for both feature extraction and artifact removal and show that removal of such artifacts can reduce predictive accuracy even when data has been cleaned in the preprocessing stages. We demonstrate how mistakes in the feature selection process can cause the cross-validation error seen in publication to be a biased estimate of the testing error seen in practice and measure this bias by purposefully making flawed models. We discuss other ways to introduce bias and the statistical assumptions lying behind the data and model themselves. Finally we discuss the complications in drawing inference from the smaller sample sizes typically seen in fMRI studies, the effects of small or unbalanced samples on the Type 1 and Type 2 error rates, and how publication bias can give a false confidence of the power of such methods. Collectively this work identifies challenges specific to fMRI classification and methods affecting the stability of models.

Details DOI

YNIMG Journal 2011 Journal Article

Performance comparison of machine learning algorithms and number of independent components used in fMRI decoding of belief vs. disbelief

  • P.K. Douglas
  • Sam Harris
  • Alan Yuille
  • Mark S. Cohen

Machine learning (ML) has become a popular tool for mining functional neuroimaging data, and there are now hopes of performing such analyses efficiently in real-time. Towards this goal, we compared accuracy of six different ML algorithms applied to neuroimaging data of persons engaged in a bivariate task, asserting their belief or disbelief of a variety of propositional statements. We performed unsupervised dimension reduction and automated feature extraction using independent component (IC) analysis and extracted IC time courses. Optimization of classification hyperparameters across each classifier occurred prior to assessment. Maximum accuracy was achieved at 92% for Random Forest, followed by 91% for AdaBoost, 89% for Naïve Bayes, 87% for a J48 decision tree, 86% for K*, and 84% for support vector machine. For real-time decoding applications, finding a parsimonious subset of diagnostic ICs might be useful. We used a forward search technique to sequentially add ranked ICs to the feature subspace. For the current data set, we determined that approximately six ICs represented a meaningful basis set for classification. We then projected these six IC spatial maps forward onto a later scanning session within subject. We then applied the optimized ML algorithms to these new data instances, and found that classification accuracy results were reproducible. Additionally, we compared our classification method to our previously published general linear model results on this same data set. The highest ranked IC spatial maps show similarity to brain regions associated with contrasts for belief > disbelief, and disbelief < belief.

Details DOI

YNIMG Journal 2010 Journal Article

Classification of spatially unaligned fMRI scans

  • Ariana Anderson
  • Ivo D. Dinov
  • Jonathan E. Sherin
  • Javier Quintana
  • A.L. Yuille
  • Mark S. Cohen

The analysis of fMRI data is challenging because they consist generally of a relatively modest signal contained in a high-dimensional space: a single scan can contain millions of voxel recordings over space and time. We present a method for classification and discrimination among fMRI that is based on modeling the scans as distance matrices, where each matrix measures the divergence of spatial network signals that fluctuate over time. We used single-subject independent components analysis (ICA), decomposing an fMRI scan into a set of statistically independent spatial networks, to extract spatial networks and time courses from each subject that have unique relationship with the other components within that subject. Mathematical properties of these relationships reveal information about the infrastructure of the brain by measuring the interaction between and strength of the components. Our technique is unique, in that it does not require spatial alignment of the scans across subjects. Instead, the classifications are made solely on the temporal activity taken by the subject's unique ICs. Multiple scans are not required and multivariate classification is implementable, and the algorithm is effectively blind to the subject-uniform underlying task paradigm. Classification accuracy of up to 90% was realized on a resting-scanned schizophrenia/normal dataset and a tasked multivariate Alzheimer's/old/young dataset. We propose that the ICs represent a plausible set of imaging basis functions consistent with network-driven theories of neural activity in which the observed signal is an aggregate of independent spatial networks having possibly dependent temporal activity.

Details DOI

YNIMG Journal 2005 Journal Article

Hippocampal activations during encoding and retrieval in a verbal working memory paradigm

  • Katherine H. Karlsgodt
  • David Shirinyan
  • Theo G.M. van Erp
  • Mark S. Cohen
  • Tyrone D. Cannon

Though the hippocampus has been associated with encoding and retrieval processes in episodic memory, the precise nature of its involvement in working memory has yet to be determined. This functional magnetic resonance imaging (fMRI) study employed a verbal working memory paradigm that allows for the within-subject comparison of functional activations during encoding, maintenance, and retrieval. In each trial, participants were shown 5 target words and, after an 8 s delay, a series of probe words. Probe words consisted of target matches, phonetically or semantically related foils, or foils unrelated to the target words. Both the left and right hippocampi showed higher mean activation amplitudes during encoding than maintenance. In contrast, the right dorsolateral prefrontal cortex (DLPFC) showed greater activation during maintenance than encoding. Both hippocampal and DLPFC regions were more active during retrieval than maintenance. Furthermore, an analysis of retrieval activation separated by probe type showed a trend toward greater bilateral hippocampal activation for probes related (both semantically and phonetically) to the target than for unrelated probes and still greater activation for target matches. This pattern suggests that there may be roles for the hippocampus and DLPFC in working memory that change as function of information processing stage. Additionally, the trend towards increased involvement of the hippocampus with the increase in relatedness of the probe stimuli to the information maintained is interpreted to be consistent with the role of the hippocampus in recollection-based retrieval in long-term memory and may indicate that this role extends to working memory processes.

Details DOI

YNIMG Journal 2004 Journal Article

Concurrent EEG/fMRI analysis by multiway Partial Least Squares

  • Eduardo Martı́nez-Montes
  • Pedro A. Valdés-Sosa
  • Fumikazu Miwakeichi
  • Robin I. Goldman
  • Mark S. Cohen

Data may now be recorded concurrently from EEG and functional MRI, using the Simultaneous Imaging for Tomographic Electrophysiology (SITE) method. As yet, there is no established means to integrate the analysis of the combined data set. Recognizing that the hemodynamically convolved time-varying EEG spectrum, S, is intrinsically multidimensional in space, frequency, and time motivated us to use multiway Partial Least-Squares (N-PLS) analysis to decompose EEG (independent variable) and fMRI (dependent variable) data uniquely as a sum of “atoms”. Each EEG atom is the outer product of spatial, spectral, and temporal signatures and each fMRI atom the product of spatial and temporal signatures. The decomposition was constrained to maximize the covariance between corresponding temporal signatures of the EEG and fMRI. On all data sets, three components whose spectral peaks were in the theta, alpha, and gamma bands appeared; only the alpha atom had a significant temporal correlation with the fMRI signal. The spatial distribution of the alpha-band atom included parieto-occipital cortex, thalamus, and insula, and corresponded closely to that reported by Goldman et al. [NeuroReport 13(18) (2002) 2487] using a more conventional analysis. The source reconstruction from EEG spatial signature showed only the parieto-occipital sources. We interpret these results to indicate that some electrical sources may be intrinsically invisible to scalp EEG, yet may be revealed through conjoint analysis of EEG and fMRI data. These results may also expose brain regions that participate in the control of brain rhythms but may not themselves be generators. As of yet, no single neuroimaging method offers the optimal combination of spatial and temporal resolution; fusing fMRI and EEG meaningfully extends the spatio-temporal resolution and sensitivity of each method.

Details DOI

YNIMG Journal 2001 Journal Article

Qualitative and Quantitative Evaluation of Six Algorithms for Correcting Intensity Nonuniformity Effects

  • James B. Arnold
  • Jeih-San Liow
  • Kirt A. Schaper
  • Joshua J. Stern
  • John G. Sled
  • David W. Shattuck
  • Andrew J. Worth
  • Mark S. Cohen

The desire to correct intensity nonuniformity in magnetic resonance images has led to the proliferation of nonuniformity-correction (NUC) algorithms with different theoretical underpinnings. In order to provide end users with a rational basis for selecting a given algorithm for a specific neuroscientific application, we evaluated the performance of six NUC algorithms. We used simulated and real MRI data volumes, including six repeat scans of the same subject, in order to rank the accuracy, precision, and stability of the nonuniformity corrections. We also compared algorithms using data volumes from different subjects and different (1. 5T and 3. 0T) MRI scanners in order to relate differences in algorithmic performance to intersubject variability and/or differences in scanner performance. In phantom studies, the correlation of the extracted with the applied nonuniformity was highest in the transaxial (left-to-right) direction and lowest in the axial (top-to-bottom) direction. Two of the six algorithms demonstrated a high degree of stability, as measured by the iterative application of the algorithm to its corrected output. While none of the algorithms performed ideally under all circumstances, locally adaptive methods generally outperformed nonadaptive methods.

Details DOI

YNIMG Journal 2000 Journal Article

Spatiotopic Organization in Human Superior Colliculus Observed with fMRI

  • Richard M. DuBois
  • Mark S. Cohen

Spatiotopy is a fundamental organizing principle of the visual brain. Using functional magnetic resonance imaging, we have demonstrated reliable data, consistent with spatiotopic organization in the human superior colliculi. Five subjects underwent cardiac-triggered echo-planar image acquisition, during which they viewed alternating left and right visual hemifield stimulation. Intensity variations from the variable TR were removed, and the data were evaluated for correlation with the lateralized stimulus. The data indicate a strongly preferential response of the left superior colliculus to the right side of visual space, and vice versa. This is consistent with previous findings in animal systems and confirms the existence of spatiotopy in the human superior colliculus.

Details DOI

YNIMG Journal 1997 Journal Article

Parametric Analysis of fMRI Data Using Linear Systems Methods

  • Mark S. Cohen

Using a model of the functional MRI (fMRI) impulse response based on published data, we have demonstrated that the form of the fMRI response to stimuli of freely varied timing can be modeled well by convolution of the impulse response with the behavioral stimulus. The amplitudes of the responses as a function of parametrically varied behavioral conditions are fitted well using a piecewise linear approximation. Use of the combined model, in conjunction with correlation analysis, results in an increase in sensitivity for the MRI study. This approach, based on the well-established methods of linear systems analysis, also allows a quantitative comparison of the response amplitudes across subjects to a broad range of behavioral conditions. Fit parameters, derived from the amplitude data, are relatively insensitive to a variety of MRI-related artifacts and yield results that are compared readily across subjects.

Details DOI
v2026.09.13