Natasa Kovacevic
My current research interests are focused on brain “noise”,
variability, and connectivity and how these properties relate to behavior,
development, and pathology. I work with functional neuroimaging data such as
fMRI, EEG, and MEG. Unlike traditional mean-based analyses, where brain
responses are estimated by averaging over multiple trials,
“noise”/variability/connectivity based analyses must utilize single trial brain
signals. It is therefore critical to
preprocess the data using the best possible methods in order to remove numerous
artefacts and confounds. I have developed sophisticated automated pipelines for
preprocessing EEG and fMRI data. Both pipelines utilize independent component
analysis (
Some of the current projects are investigating changes in brain variability due to aging, maturation, task demands, consciousness levels (minimally conscious, locked-in, fully conscious), learning, and pathology (TBI, schizophrenia, epilepsy).
Vasily Vakorin
My research focuses on developing and testing mathematical and
computational techniques for analyzing functional MRI, EEG, and MEG
data. Specifically, I am interested in methods for studying functional
integration in the human brain. Currently, I am working on inferring
causal relations between neural activity in different brain regions. Jennifer Heisz
My research focus is at the interface of visual perception and cognition, detailing how visual processes change as a function of learning and prior knowledge. I primarily study this within the framework of the face recognition system: learning faces and getting to know a person. I use eye-tracking techniques to monitor learning-related changes in the type of visual information sought for processing. Notably, face learning leads to more efficient visual scanning (Heisz & Shore, 2008). I use EEG and MEG neuroimaging techniques to assess the timecourse of brain activity. As we get to know someone, and build up a rich associated semantic network, our early visual processing of his or her face is modified (Heisz & Shedden, 2009).
I am also interested in age-related changes in learning and memory, with an emphasis on dissociating processes that are spared from those that are impaired. My recent work suggests that older adults can improve their memory for newly learned faces through simple repetition (Heisz & Ryan, submitted). Incorporating this type of information into practice may help older adults maintain a higher quality of life.
I currently work as a postdoctoral fellow with Dr. Randy McIntosh. Together, we are exploring the idea that brain signal variability may convey useful information about underlying neural network properties. I am particularly interested in whether brain signal variability changes with learning and how such changes relate to the richness of the underlying memory representation. These two lines of research may be especially informative in the context of an aging brain, as both brain signal variability and memory change. To address these questions, I use a variety of multivariate statistical techniques, including a measure of multiscale entropy.
Andreea Diaconescu
My
research focuses on two main areas: multisensory integration and neural
network reorganization. In the work on multisensory integration, I
addressed the question of how stimulus characteristics, that link up to
motor behaviour, are processed across visual and auditory sensory
modalities. Using event-related potentials, I also investigated the
neural processes underlying spatial localization and object
identification in visual and auditory domains. The aim of this work is
to explore whether cognitive operations performed across multiple
modalities, visual, auditory, and tactile, can induce changes in
cortical plasticity and improvements in performance. In the work on
neural network reorganization, I collaborated with Dr. S. Kapur at the
Centre for Addiction and Mental Health, to study the effects of
dopamine agents on classical conditioning using fMRI. Our work showed
that as learning ensued, distinct neural networks activated, and their
activation patterns were modulated by increases and decreases in
dopamine levels. This project served as a framework for studying the
cortical changes observed in patients with schizophrenia. Next, we will
investigate whether changes in functional connectivity can predict
recovery from brain damage.Zainab Fatima
My
primary research interests revolve around studying the spatial and
temporal dynamics of large-scale interactions that occur in the brain
due to convergence of top-down (attentional) and bottom-up
(stimulus-driven) processing. I use functional neuroimaging (fMRI) data
to investigate how sensorimotor and cognitive systems are organized. By
changing task demands and cognitive load, sensorimotor systems can
change their interactions with cognitive systems and vice versa. Im
interested in studying the neural network properties of such dynamic
interactions using multivariate statistical techniques such as partial
least squares (PLS) and structural equation modeling (SEM).
My
secondary interests lie in examining how the brain organization changes
as a result of learning. Im specifically interested in how the
prefrontal cortex, parts of the basal ganglia and the hippocampus
interact with each other at the beginning of learning a task and as
learning progresses. Im also fascinated by changing network dynamics
in learning due to the type of strategy individuals use. I will be
exploring this area in more detail in further graduate work.
Michele Korostil
My research program uses functional MRI to examine the neural systems underlying practice-related learning effects in persons with schizophrenia.
My research interests focus on the application of mathematics to the study of interactions in the human brain that give rise to cognition and behavior, as measured by functional neuroimaging and event-related potentials. One aspect is the analysis of these signals using multivariate statistical techniques developed in our lab. Though these methods are mostly employed in the study of higher processes such as memory, I am interested in using them to study the functional connectivity underlying lower processes such as motion perception. Another aspect is the characterization of brain signals from a dynamical systems perspective, where perception, cognition and behavior are treated as stable yet fluid emergent states of nonlinear neural interactions. I am interested in applying these ideas to neuroimaging methodology, both in terms of experimental design (which changes considerably when the objective is to cause qualitative changes in the system such as bifurcations) and analysis (where individual subject trends are often more telling than group trends). I am also interested in the related concept of signal complexity: how complexity varies as a function of task demand, and how task success varies as a function of complexity.
Jordan Poppenk
My
past and current research in memory focuses on the cognitive and neural
basis of how humans detect and encode novel information. To determine
whether something has changed in our environment, we must somehow
contrast its features against a representation stored in memory. This
process of novelty detection is ongoing and apparently automatic, and
recent (controversial) evidence suggests it is also a prerequisite for
memory encoding.
In my undergraduate thesis, I investigated neural
sensitivity to, and subsequent memory for, various types of novelty.
Functional imaging data revealed that prefrontal and medial temporal
lobe regions linked with long-term memory encoding were differentially
sensitive to various novelty types, reflecting subsequent memory for
those novelty types. My current M.A. dissertation will include a
functional network analysis of that data set and a cognitive experiment
to test claims that novelty-detection is a pre-requisite for encoding.
This work will also prepare a paradigm for neural investigation of
novelty-detection and encoding.
Personal website: http://individual.utoronto.ca/poppenk/
My research involves investigating changes in neural networks associated with recovery after brain damage. I am interested in measuring integration and differentiation of brain activity in patients with traumatic brain injury (TBI). The nature of TBI pathology is diffuse, which suggests a more distributed network dysfunction. Therefore, in order to capture brain-behavior relationships in TBI, I am currently examining the interactions between neuronal groups rather than focusing on isolated brain regions. For my M.A. thesis, I am analyzing complexity in MEG signals obtained during a visual feature-matching task which is sensitive to impairments in attention after TBI. High values of complexity correspond to an optimal balance of functional specialization and functional integration. My initial results show lower values of complexity in some neural loci after TBI. This may indicate less information processing capacity in these regions. In addition, TBI patients with higher complexity values show less variability in performance, suggesting higher complexity may relate to improved function. Future directions of my research in this area will involve examining how complexity relates to processes involving large-scale neural interactions such as conscious awareness. Outside of this vein of research, I will also be conducting a project in collaboration with Mary Pat McAndrews at Toronto Western and Steven Small at the University of Chicago that seeks to understand changes in neural networks after childhood stroke.
Grigori Yourganov
My
work is concerned with evaluation of performance of different
statistical methods of fMRI signal detection. The evaluation is done on
simple simulated data: this allows us to use well-known
signal-detection metrics like ROC area and to study how performance is
affected by such factors as contrast-to-noise ratio and dynamic range
of the signal, network correlation, and sample size. Also, I am using
metrics that don’t require the knowledge of “ground truth”, such as
reproducibility of spatial maps and classification accuracy: I apply
these metrics to real as well as simulated data sets. My emphasis is on
multivariate methods, especially on Fisher’s linear discriminant and
quadratic discriminant; these well-known algorithms often outperform
such popular methods as Independent Component Analysis and Support
Vector Machines. Also, they usually perform better than univariate
General Linear Model, especially when the network structure of the
signal is manifest. I pay special interest to data sets of small size
and low signal-to-noise ratio, which are common in fMRI studies.
A big part of my work is estimation of intrinsic dimensionality of fMRI
data, that is, the number of principal components that defines the
optimal approximation of the data set. My work shows that it is often
better to select the subspace size that directly optimizes the
data-driven metrics (e.g. spatial map reproducibility), than to use
popular analytical dimensionality estimators like Minimum Description
Length or optimal Bayesian evidence. Even though such analytical
estimators have good asymptotic performance, they frequently produce
sub-optimal results in data sets of small size.
Maria Tassopoulos
I graduated from the University of Toronto with an HBSc specializing in Psychology. Currently I am working on various projects. An example of an ongoing study includes investigating parts of the brain that are important for integrating sensory and motor information. The spatiotemporal dynamics of these interactions are examined using fMRI. We are also using fMRI to explore the brain-behaviour relationship of learning through practice in schizophrenia. An understanding of the neural underpinnings of certain stages of learning is needed to enhance our knowledge of the illness and subsequent treatment protocols. Another aspect of our research consists of acquiring conditioned responses to subliminal stimuli using physiological measures like GSR and EMG.
I am involved with managing our lab's subject database as well as our websites. I contribute to programming experiments, recruiting and running subjects (behavioural and neuroimaging studies), and assist with grant and manuscript submissions. I also provide support for all stages of brain imaging data processing and analyses: I preprocess fMRI data using AFNI and SPM, which I then analyze using PLS.
Tanya Brown
I recently graduated from the University of Toronto
Hongye Wang
My background is in Electrical Engineering with a focus on signal processing. I am generally interested in various methods for
analyzing neuroimaging data, such as data from fMRI, EEG, and MEG. Currently I am working on a project that investigates the functional connectivity of the human brain by using seed-based/ROI-based connectivity analysis.
I also assist with running subjects and collecting EEG data, as well as preprocessing fMRI data using PLS and SPM.
Michele Korostil
My research program uses functional MRI to examine the neural systems underlying practice-related learning effects in persons with schizophrenia.My research interests focus on the application of mathematics to the study of interactions in the human brain that give rise to cognition and behavior, as measured by functional neuroimaging and event-related potentials. One aspect is the analysis of these signals using multivariate statistical techniques developed in our lab. Though these methods are mostly employed in the study of higher processes such as memory, I am interested in using them to study the functional connectivity underlying lower processes such as motion perception. Another aspect is the characterization of brain signals from a dynamical systems perspective, where perception, cognition and behavior are treated as stable yet fluid emergent states of nonlinear neural interactions. I am interested in applying these ideas to neuroimaging methodology, both in terms of experimental design (which changes considerably when the objective is to cause qualitative changes in the system such as bifurcations) and analysis (where individual subject trends are often more telling than group trends). I am also interested in the related concept of signal complexity: how complexity varies as a function of task demand, and how task success varies as a function of complexity.Jordan Poppenk
My
past and current research in memory focuses on the cognitive and neural
basis of how humans detect and encode novel information. To determine
whether something has changed in our environment, we must somehow
contrast its features against a representation stored in memory. This
process of novelty detection is ongoing and apparently automatic, and
recent (controversial) evidence suggests it is also a prerequisite for
memory encoding.In my undergraduate thesis, I investigated neural sensitivity to, and subsequent memory for, various types of novelty. Functional imaging data revealed that prefrontal and medial temporal lobe regions linked with long-term memory encoding were differentially sensitive to various novelty types, reflecting subsequent memory for those novelty types. My current M.A. dissertation will include a functional network analysis of that data set and a cognitive experiment to test claims that novelty-detection is a pre-requisite for encoding. This work will also prepare a paradigm for neural investigation of novelty-detection and encoding.
Personal website: http://individual.utoronto.ca/poppenk/
My research involves investigating changes in neural networks associated with recovery after brain damage. I am interested in measuring integration and differentiation of brain activity in patients with traumatic brain injury (TBI). The nature of TBI pathology is diffuse, which suggests a more distributed network dysfunction. Therefore, in order to capture brain-behavior relationships in TBI, I am currently examining the interactions between neuronal groups rather than focusing on isolated brain regions. For my M.A. thesis, I am analyzing complexity in MEG signals obtained during a visual feature-matching task which is sensitive to impairments in attention after TBI. High values of complexity correspond to an optimal balance of functional specialization and functional integration. My initial results show lower values of complexity in some neural loci after TBI. This may indicate less information processing capacity in these regions. In addition, TBI patients with higher complexity values show less variability in performance, suggesting higher complexity may relate to improved function. Future directions of my research in this area will involve examining how complexity relates to processes involving large-scale neural interactions such as conscious awareness. Outside of this vein of research, I will also be conducting a project in collaboration with Mary Pat McAndrews at Toronto Western and Steven Small at the University of Chicago that seeks to understand changes in neural networks after childhood stroke.Grigori Yourganov
My
work is concerned with evaluation of performance of different
statistical methods of fMRI signal detection. The evaluation is done on
simple simulated data: this allows us to use well-known
signal-detection metrics like ROC area and to study how performance is
affected by such factors as contrast-to-noise ratio and dynamic range
of the signal, network correlation, and sample size. Also, I am using
metrics that don’t require the knowledge of “ground truth”, such as
reproducibility of spatial maps and classification accuracy: I apply
these metrics to real as well as simulated data sets. My emphasis is on
multivariate methods, especially on Fisher’s linear discriminant and
quadratic discriminant; these well-known algorithms often outperform
such popular methods as Independent Component Analysis and Support
Vector Machines. Also, they usually perform better than univariate
General Linear Model, especially when the network structure of the
signal is manifest. I pay special interest to data sets of small size
and low signal-to-noise ratio, which are common in fMRI studies.A big part of my work is estimation of intrinsic dimensionality of fMRI data, that is, the number of principal components that defines the optimal approximation of the data set. My work shows that it is often better to select the subspace size that directly optimizes the data-driven metrics (e.g. spatial map reproducibility), than to use popular analytical dimensionality estimators like Minimum Description Length or optimal Bayesian evidence. Even though such analytical estimators have good asymptotic performance, they frequently produce sub-optimal results in data sets of small size.
Maria Tassopoulos
I graduated from the University of Toronto with an HBSc specializing in Psychology. Currently I am working on various projects. An example of an ongoing study includes investigating parts of the brain that are important for integrating sensory and motor information. The spatiotemporal dynamics of these interactions are examined using fMRI. We are also using fMRI to explore the brain-behaviour relationship of learning through practice in schizophrenia. An understanding of the neural underpinnings of certain stages of learning is needed to enhance our knowledge of the illness and subsequent treatment protocols. Another aspect of our research consists of acquiring conditioned responses to subliminal stimuli using physiological measures like GSR and EMG.I am involved with managing our lab's subject database as well as our websites. I contribute to programming experiments, recruiting and running subjects (behavioural and neuroimaging studies), and assist with grant and manuscript submissions. I also provide support for all stages of brain imaging data processing and analyses: I preprocess fMRI data using AFNI and SPM, which I then analyze using PLS.
Tanya Brown
I recently graduated from the University of Toronto
Hongye Wang
My background is in Electrical Engineering with a focus on signal processing. I am generally interested in various methods for
analyzing neuroimaging data, such as data from fMRI, EEG, and MEG. Currently I am working on a project that investigates the functional connectivity of the human brain by using seed-based/ROI-based connectivity analysis.I also assist with running subjects and collecting EEG data, as well as preprocessing fMRI data using PLS and SPM.