Michigan Neuroimaging Initiative

Upcoming Events

There are currently no events scheduled for the spring/summer terms.

Past events

Understanding memory disorders: At the level of cognitive process representational content?

Morgan Barense, University of Toronto

Tue, 9 Apr, 2019

Time: 4:00–5:00 PM

Location: 4464 East Hall

This is an event in the fMRI Lab Speaker series.

How does perception of an object relate to subsequent memory for that object? A central assumption in most modern theories of memory is that memory and perception are functionally and anatomically segregated. For example, amnesia resulting from medial temporal lobe (MTL) lesions is traditionally considered to be a selective deficit in long-term declarative memory with no effect on perceptual processes. This view is consistent with a popular paradigm in cognitive neuroscience, in which the brain is understood in terms of a modular organization of function based on cognitive process. The work I will present offers a new perspective. Guided by computational modelling complemented with neuropsychology and neuroimaging, I will provide support for the notion that memory and perception are inextricably intertwined throughout the MTL, relying on shared neural representations and computational mechanisms. I will then describe how this new framework can improve basic understanding of cognitive impairments observed in Alzheimer’s disease, as well as guide development of new diagnostic procedures for those at risk for dementia.

The Global Signal, Vigilance Fluctuations, and Nuisance Regression in Resting State fMRI

Thomas Liu, University of California, San Diego

Tue, 2 Apr, 2019

Time: 4:00–5:00 PM

Location: 4464 East Hall

Resting-state fMRI (rsfMRI) is now a widely used method to assess the functional connectivity (FC) of the brain. However, the mechanisms underlying rsfMRI are still poorly understood. In this talk I will address several related aspects of the rsfMRI signal. The first is the global signal, which represents the whole brain average signal and has been widely used as a regressor for removing the effects of global variations in resting-state activity. I will discuss the controversy surrounding global signal regression and describe new approaches for minimizing global signal effects. A related topic concerns the origins of global activity in the brain. There is growing evidence that a considerable portion of this global activity arises from fluctuations in vigilance and arousal. I will discuss the recent findings in this area and discuss the implications for the analysis and interpretation of rsfMRI studies. Finally, I will describe recent empirical and theoretical work demonstrating the limitations of regression based methods that are widely used to minimize the effects of nuisance components in rsfMRI studies.

This is an event in the fMRI Lab Speaker series.


Slides from the talk

Functional Brain Network Organization and Dynamics in Health and Disease

Jessica Cohen, University of North Carolina, Chapel Hill

Tue, 12 Mar, 2019

Time: 4:00–5:00 PM

Location: 4464 East Hall

This is an event in the fMRI Lab Speaker series.

The brain’s ability to adaptively engage different functional networks in the face of a changing environment is an important characteristic that enables a wide variety of behaviors. The goal of my research program is to understand how distinct brain networks interact with each other and flexibly reconfigure when confronted with a dynamic environment, as well as how network integration contributes to individual differences in behavior in both health and disease. In my talk, I will first discuss adaptive reconfiguration of functional brain network organization in response to changes in cognitive demands, followed by a depiction of situations in which stable brain network organization is adaptive. I will end by describing how dysfunctional brain network organization in ADHD underlies symptoms and cognitive deficits. Together, this research provides evidence that the healthy brain systematically reconfigures to adapt to current demands, and that dysfunction in this dynamic network behavior underlies ADHD.

At the intersection: Computational science, cognitive neuroscience, and data science

A visit by James Haxby, Director of both the Center for Cognitive Neuroscience and the Dartmouth Brain Imaging Center, and Evans Family Fellow, Psychological and Brain Sciences, Dartmouth College

This visit was coordinated by the Michigan Neuroimaging Initiative, with primary support from the Michigan Institute for Computational Discovery and Engineering (MICDE) and additional support from the fMRI Laboratory and the Psychology department.

Professor Haxby pioneered the use of machine-learning techniques (MVPA) for neuroimaging data. His work integrates and pushes the boundaries in developing cognitive neuroscience, computational methods, and data management tools.

He is noted for his use of complex, naturalisitic stimuli. For example, using brain signals associated with viewing the film Raiders of the Lost Ark to begin to determine which aspects of those brain signals are individual-specific and which generalize across people. This and other scenarios have also been used to create more specific category mappings for the human visual system.

People from his lab have made major contributions to open source computing resources for neuroimaging analysis, including development of PyMVPA, NeuroDebian, and DataLad (by Yaroslav Halchenko, who now has his own lab).

Bridging the divide: fostering interdisciplinary collaborative research in computational cognitive neuroscience

Mon, 18 Feb, 2019
Time: 3:00–4:00 PM
Location: 1017 Herbert Dow Engineering Building, 2300 Hayward, UM North Campus

Computational cognitive neuroscience is a burgeoning field. Sensitive imaging methods can now measure changing patterns of brain activity noninvasively producing massive, rich datasets. With open neuroscience, vast amounts of functional brain imaging data are publicly available. Advances in computational methods for analyzing these data and modeling the underlying cognitive processes have produced a host of sophisticated algorithms that produce surprising new insights, and these algorithms are available in extensive repositories of open source code.

Building the interdisciplinary community for this type of collaborative research, however, presents challenges. Taking advantage of these resources requires integration of knowledge of cognitive neuroscience to direct projects to important questions and knowledge of rapidly evolving computational approaches that can tackle these questions in innovative ways. Building an interdisciplinary community will involve developing both productive interdisciplinary collaborative teams and a new breed of “bilingual” computational cognitive neuroscientist.


PowerPoint slides

Current issues in computational neuroimaging of brain function

Tue, 19 Feb, 2019
Time: 10:30–11:30 AM
Location: 4446 East Hall


PowerPoint slides

Hyperalignment: modeling the shared information encoded in idiosyncratic fine-scale cortical topographies

Tue, 19 Feb, 2019
Time: 4:00–5:00 PM
Location: 4448 East Hall
Multivariate pattern analysis reconceptualizes cortical functional architecture as high-dimensional information spaces that are encoded in fine-grained topographies of response and functional connectivity patterns. We have developed a high-dimensional computational model of the information encoded in response and connectivity topographies that is shared across brains. We derive this model with new algorithms called response hyperalignment and connectivity hyperalignment.

Our model captures shared information as basis functions for response tuning and functional connectivity that are common across brains. These response and connectivity basis functions are instantiated in individual brains as multiplexed topographic basis functions that are specific to each individual brain. We developed this model using fMRI data collected while subjects watched meaningful, dynamic naturalistic stimuli, namely movies, and during the resting state. We are now investigating how this model can be leveraged for more sensitive analyses of individual differences in cortical functional architecture that predict cognitive differences.

Slides and videos

PowerPoint slides (large, contains embedded video)

Videos from slide 18
Excerpt from Raiders of the Lost Ark
Brain activity animation

Videos from slide 19
Subject 1 brain activity
Subject 2 brain activity

Using the Flux cluster for neuroimaging

Bennet Fauber, University of Michigan

Tue, 11 Dec, 2018

Time: 4:00–5:30 PM

Location: B254 East Hall

This workshop will introduce participants to basic usage of the Flux cluster. Participants will log into the cluster, create a job submission script, learn the most commonly changed options for their jobs. Examples will run neuroimaging specific software and R, in advance of the workshop on Neuropointilist, Fri, Dec 14, 2018.

NOTE Participants MUST apply for a Flux login account by 5:00 PM Wednesday, Dec 5, 2018, and register for Duo two-factor authentication. We will check for valid Flux accounts Friday evening and disenroll anyone who has not got a valid Flux login account. This is necessary so we can authorize participants to use the Flux training account.

To register for the workshop and for links to apply for your Flux account and enroll for Duo, please go to the registration page at: Workshop: Using the Flux cluster for neuroimaging

Neuropointillist: Using R to Foster Innovative Modeling of MRI Data

Tara Madhyastha, University of Washington, Seattle

Fri, 14 Dec, 2018

Time: 1:00–2:00 PM

Location: 4448 East Hall

The human brain is constantly changing in response to the environment, development, aging and neurodegeneration. Although fMRI has been crucial in helping us understand brain function, modeling trajectories of change over time, examining the relationship of individual differences in growth to other variables, and analysis of data with related individuals has been challenging within existing statistical GLM frameworks. There have been advances in addressing limitations of the GLM approach. For example, mixed effects models as implemented in AFNI greatly improve longitudinal modeling ability, with the ability to handle missing data, compare models, and use the underlying power of mixed effects modeling packages available the R statistical language. However, structural equation models (SEM), popular in the social sciences, give us the ability to examine change in neural processes as outcomes, predictors, correlates of other change processes, or moderators or mediators. This flexibility is currently lacking in neuroimaging software. In this workshop we describe the limitations of existing neuroimaging software and motivate taking a more flexible modeling approach in R. In this talk we will motivate this approach, and describe some innovative analyses that we have been able to perform.

Workshop: Neuropointillist, an R Package to Flexibly Model MRI Data

Tara Madhyastha, University of Washington, Seattle

Fri, 14 Dec, 2018

Time: 9:00–11:00 AM

Location: B254 East Hall

NOTE The workshop will use the Flux cluster. You must have a Flux account and have submitted at least one job prior to this workshop to actively participate. Please sign up for the Using the Flux cluster for neuroimaging workshop listed above if you do not have Flux experience already.

fMRI is an invaluable tool in helping us to understand brain function. However, analysis of fMRI data is complex and involves removing numerous sources of noise to accurately measure relatively small changes in blood flow in the brain (the blood oxygen level-dependent, or BOLD signal). After denoising, a typical modeling approach involves using a GLM to identify where the BOLD signal in each three-dimensional pixel, or voxel, changes in response to task demands. This denoising and modeling process is typically performed by special purpose fMRI analysis software packages. However, the flexibility of models that these packages support is quite limited, in part because of computational limitations when these packages were developed in the early 1990s.

Characterizing the spectrum of task fMRI connectivity approaches

Patrick JA Pruitt, Wayne State University

Tue, 13 Nov, 2018

Time: 4:00–5:00 PM

Location: 4464 East Hall

Task-based functional connectivity (FC) approaches have typically sought to characterize the modulation of connectivity by task condition (e.g. PPI, beta-series correlation). However, other more “resting-state”-like approaches to task-based connectivity are gaining traction. These techniques examine FC over the entire task session and either leave in (AS Greene et al, Nat Comm 2018) or attempt to regress out (“background connectivity” DA Fair et al, Neuroimage 2007) the effects of task stimuli. Existing somewhere between task-modulated FC and resting-state FC, what do these approaches have to offer our understanding of functional connectivity and – more broadly – cognition and disease? In my talk, I will:

In sum, the talk as designed is methods-focused and built on a foundation of the concepts underlying the different approaches, but also tying in recent work actually using the approaches (including some of my work in Dr. Damoiseaux’s lab).

Slides from the talk

Brain sandwiches: Fast and accurate modeling of longitudinal and repeated measures neuroimaging data

Jillian Hardee, University of Michigan

Tue, 27 Oct, 2018

Time: 4:00–5:00 PM

Location: 4464 East Hall

This talk will be about why longitudinal is done, what it is suited for, some of the challenges in conducting longitudinal studies, and an overview of one statistical approach, the sandwich estimator.

Slides from the talk

States and Stability in Human Brain Networks

Caterina Gratton, Northwestern University

Tue, 9 Oct, 2018

Time: 4:00–5:00 PM

Location: 4464 East Hall

This is an event in the fMRI Lab Speaker series.

Humans can easily and flexibly accomplish a wide variety of tasks, with different perceptual and cognitive demands, depending on their goals. This ability appears to depend on the coordinated interactions between brain regions that are organized into large-scale networks. In my research, I am interested in characterizing how human brain networks are organized, how they contribute to the myriad goal-directed behaviors that are essential to our daily lives, and how these processes break down. In my talk, I will present an overview of my research, highlighting three recent projects centered on understanding brain network organization and how it changes over different timescales. The first section will use dense measurements of brain networks from 10 individual subjects to examine how networks vary across individuals, days, and task states. In the second part of the talk, I will discuss how brain networks are affected by progressive neurodegeneration in Parkinson’s Disease. Finally, I will describe findings from a recent initiative into characterizing trait-like variation in brain networks. Jointly, these projects add to our understanding of brain organization and function, and how this organization contributes to complex behaviors.

Suggested readings

The link to the article requires a UM login account for access

Functional brain networks are dominated by stable group and individual factors, not cognitive or daily variation,
Caterina Gratton, et al.,
Neuron, 98, 439–452

Slides from the talk

fMRI Symposium

Date: 21 Sep, 2018
Location: 4448 East Hall
Symposium web site

The 2018 Functional MRI Symposium will be held in the Colloquium Room, located in East Hall, 4th floor, room 4448. Lunch and beverages will be provided. The day will be devoted to talks that represent a range of research from methodology to bioengineering to biostatistics to neuroscience.

Speakers and talk titles

David Brang, Department of Psychology, University of Michigan,
“Electrocorticography (ECoG): Overview of Methods and Applications”

Molly Simmonite, Department of Psychology, University of Michigan,
“Exploring Neural Distinctiveness in Healthy Aging using MVPA”

Leigh Goetschius, Department of Psychology, University of Michigan,
“A Multimodal Examination of Prefrontal-Amygdala White Matter Microstructure and Functional Amygdala Activation in Adolescents”

John Plass, Department of Psychology, University of Michigan,
“Modern Methods for Diffusion MRI: Whole-Brain Analysis at the Fiber Segment Level”

Adriene Beltz, Department of Psychology, University of Michigan,
“Estimating Time-Varying Networks with GIMME: A Description and Application”

Andrew Jahn, Department of Psychology, University of Michigan,
Bennet Fauber, Advanced Research Computing, University of Michigan,
Michael Angstadt, Department of Psychiatry, University of Michigan,
“Neuroimaging QA: Developing Tools and Training for Michigan Researchers”

Agenda: Symposium agenda

Ways to improve behavioral predictions from functional connectivity data

Dustin Scheinost, Assistant Professor of Radiology and Biomedical Imaging, Yale School of Medicine

Tue, 17 Apr, 2018

Time: 4:00–5:00 PM

Location: 4464 East Hall

Assessment of neural networks at the level of large-scale systems has the potential to provide novel insight into brain-behavior relationships. However, how best to probe these brain-behavior relationships remains unclear. This talk will describe how factors such as data reliability, brain state, and individualized functional parcellations moderate behavioral prediction performance of a recently developed functional connectivity analysis approach, connectome-based predictive modeling (CPM). Similarly, I will present an extension to CPM, labeled multidimensional CPM, that incorporates information from multiple tasks/modalities into a single model to improve behavioral predictions.

Suggested readings

The links to the articles require a UM login account for access

Influences on the Test–Retest Reliability of Functional Connectivity MRI and its Relationship with Behavioral Utility”,
Stephanie Noble, Marisa N. Spann, et al,
Cerebral Cortex, November 2017 27:5415–5429
doi: 10.1093/cercor/bhx230

Connectome-based predictive modeling of attention: Comparing different functional connectivity features and prediction methods across datasets
Kwangsun Yoo, Monica D. Rosenberg, et al.”,
Neuroimage, 167 (2018) 11–22
doi: 10.1016/j.neuroimage.2017.11.010

Submodular Approach to Create Individualized Parcellations of the Human Brain”,
Mehraveh Salehi, Amin Karbasi, et al.,
Conference Paper, September 2017
doi: 10.1007/978-3-319-66182-7_55

Using connectome-based predictive modeling to predict individual behavior from brain connectivity”,
Xilin Shen, Emily S Finn, et al.,
Nature Protocols, volume 12, 2017, pp 506–518
doi: 10.1038/nprot.2016.178

Task Integration for Connectome-Based Prediction via Canonical Correlation Analysis”,
Siyuan Gao, Abigail S. Greene, et al.,
Conference paper, 2018 IEEE International Symposium on Biomedical Imaging
(To be presented)

Code examples

Slides from the talk

Functional MRI of the Dynamic Brain, Quasiperiodic Pattern, Brain States, and Trajectories

Shella Keilholz, Associate Professor, Biomedical Engineering, Georgia Tech

Tue, 20 Mar, 2018

Time: 4:00–5:00 PM

Location: 4464 East Hall

This is an event in the fMRI Lab Speaker series.

Resting state functional magnetic resonance imaging (rs-fMRI) can capture activity patterns throughout the whole brain as a function of time. The whole brain patterns can be characterized into functional networks of brain areas whose activity maintains a statistical dependence over the course of the scan, a feature described as functional connectivity. More recently, researchers have moved beyond measures of functional connectivity that average across an entire scan (typically 5–10 minutes) to methods that can describe the dynamic features of the brain over the course of the scan. These include point process analysis, windowed functional connectivity techniques, hidden Markov models, and more. This talk will focus on a prominent dynamic feature of rs-fMRI data, quasiperiodic spatiotemporal patterns of activity (QPPs). The QPPs contribute substantially to average measures of functional connectivity, are tied to infraslow activity and behavioral performance, and are altered in patient groups. Preliminary data suggests that they reflect neuromodulatory input from deep brain nuclei and may allow this neurophysiological signal to be isolated from standard rs-fMRI exams.

Slides PDF slides from the talk

Understanding and Teaching Neuroimaging Analysis: Common Difficulties and the Search for Answers

Andrew Jahn, Post-doctoral fellow, Yale University
Andy will be joining UM as an fMRI Analysis consultant in mid-June, 2018

Mon, 19 Mar, 2018

Time: 4:00–5:00 PM

Location: 4464 East Hall

As neuroimaging research has become more widespread, so has the demand for more effective ways of learning how to do it. This is compounded by the myriad needs of researchers: Some are completely new to the field; others have some experience and wish to deepen their understanding; and still others simply want to fix basic errors and solve problems as quickly as possible.

In this talk, I tell my story about how I have learned and taught neuroimaging techniques – especially fMRI – and how this led to my creating online tutorials designed to teach others neuroimaging analysis. I outline what I see as common difficulties that impede understanding, and what I have found to be the most effective ways to help the different groups listed above. I end by summarizing what I predict will be the dominant trends in neuroimaging analysis over the next few years.

How Neurotechnologies are Providing New Insights In Vivo Into the Treatment of Migraine and other Chronic Pain Disorders

Alexandre DaSilva, HOPE Lab, fNIRS Lab at CHGD, and Associate Profesor, Dentistry

Tue, 13 Mar, 2018

Time: 4:00–5:00 PM

Location: 4464 East Hall

This is an event in the fMRI Lab Speaker series.

While understanding brain mechanisms in chronic pain is important, equally important is applying these concepts in the clinical environment. For example, recent in vivo molecular imaging studies have demonstrated that there is a dysfunctional μ-opioid and dopamine neurotransmission in certain brain regions of migraineurs during spontaneous headache attacks and allodynia. In parallel, other studies have shown scientific evidence that novel non-invasive neuromodulation tools can change endogenous neurotransmission and also provide relatively lasting pain relief in some pain disorders, including chronic migraine, TMD and fibromyalgia. Our overall goal is to discuss novel advances in pain neuroimaging (e.g., PET, fNIRS), with a focus on clinical applications, even with their combination of augmented reality. We will also discuss an emerging neuroimaging technology, functional near-infrared spectroscopy (fNIRS), that can now provide better understating of the ongoing impact of affective and sensory experience in the brain before, during, and after clinical pain.

Slides PDF slides from the talk

Unix scripting and Make

Thad Polk, Professor, Psychology

Fri, Jan 26, 2018

Time: 12:00–1:00 PM

Location: 4464 East Hall

This is in the Psychology Methods Hour series. See the Methods hour web site.

This talk will introduce using Unix scripting and the make to create reproducible workflows (pipelines) for neuroimaging analyses.

Reading Mary K Askren, et al. Using Make for Reproducible and Parallel Neuroimaging Workflow and Quality-Assurance http://journal.frontiersin.org/article/10.3389/fninf.2016.00002/full

Slides PowerPoint slides from the talk

Example files A gzipped, tar archive containing a simple example using the FSL bet command

Considerations in Relating the Functional Connectome to Behavior: Connectome Based Predictive Modeling

Todd Constable, Director of the Magnetic Resonance Research Center, Yale University

Thu, Jan 18, 2018

Time: 4:00–5:00 PM

Location: 4448 East Hall

fMRI Laboratory speaker series

This talk will focus on recent work relating the individual connectome to behavior and/or clinical symptoms. The individual connectome is a connectivity based measure obtained from fMRI data, that reflects the functional organization of an individual's brain. Variations in this functional organization can tell us something about the individual whether this be their capabilities on a behavioral task or some clinical symptom measure. The approach to connectome based predictive modeling will be described and factors that influence model performance discussed. Examples will be shown in which we are able to predict in novel individuals both behavioral and clinical measures obtained outside the scanner. Stated based manipulations will be discussed in the context of revealing trait based features. This work holds tremendous promise for understanding the neurophysiological basis for a range of normal behaviors, developmental trajectories, and neurological diseases and disorders.

Light refreshments will be served.

Diffusion MRI: Introduction and Modern Methods

John Plass, Psychology

Date: Nov 14, 2017

Time: 4:00 PM

Location: 4464 East Hall

fMRI Laboratory speaker series

Diffusion MRI (dMRI) is a non-invasive imaging technique used to probe the microstructure and morphology of white matter structures in the brain. Whereas diffusion-weighted scans have become ubiquitous in recent years, many researchers are often unclear on how to model, analyze, and interpret dMRI data. In this talk, I will use a straightforward pictographic approach to introduce dMRI analysis techniques used to identify and quantify the features of white matter pathways.

After introducing the most commonly used models (diffusion tensor models), I will demonstrate their shortcomings and introduce recently-developed alternatives. These modern alternatives aim to isolate measures of microscopic tissue structure (e.g., fiber density) from potential confounds produced by local fiber geometry (e.g., crossing fibers), allowing for more reliable, biologically meaningful measures of anatomical connectivity. I will demonstrate how these novel approaches can be used to test hypotheses about anatomical connections between regions and their relationships with other variables of interest.

Slides from talk

Diffusion MRI: Introduction and Modern Methods (.pdf)

Diffusion MRI: Introduction and Modern Methods (.pptx)

Cultural Neuroscience: Linking Context to Genes and the Brain

Shinobu Kitayama, Social Psychology Area Chair; Robert B. Zajonc Collegiate Professor of Psychology; Director of the Culture and Cognition Program

Date: Oct 27, 2017

Time: 2:10 PM

Location: 4464 East Hall

Presented as part of the CCN Forum series

Prior work in cultural psychology and cultural neuroscience shows that culture (defined as a loosely organized set of practices and meanings) has systematic influences on psychological processes that are reflected in judgment, memory, and decision-making, as well as in functional brain activations that are linked to these domains (1, 2). However, it is not as yet clear whether the cultural influences might extend to structural properties of the brain such as the volume of different cortical and subcortical regions. Nor is it clear whether the cultural difference in the regionally specific brain volume could be attributed to cultural experience. To explore these questions, we have investigated whether the cultural dimension of independent versus interdependent self-construal would be related to brain volume. In support of the hypothesis that interdependence is linked to the down-regulation of a strategic pursuit of self-interest, we have found that the volume of the bilateral orbitofrontal cortex inversely predicts interdependent self-construal (3). Further, we have underscored the critical role of experience in the OFC volume difference by utilizing a genetic means. Specifically, while the OFC volume tends to be less in interdependent vs. independent cultures, this cultural difference is apparent only among those carrying genetic polymorphisms that are known to support environmental influences. Implications for future research on culture, genes, and the brain will be discussed.

An introduction to mean and variance modeling for studies with repeated measures

Kerby Shedden, Director of UM Consulting for Statistics, Computing and Analytics Research (CSCAR)

Date: Oct 24, 2017

Time: 4:00 PM

Location: 4464 East Hall

Jointly presented by UM NII and the Psychology Methods Hour

This talk will describe the marginal mean/covariance models as a convenient way to think about several popular statistical approaches for working with dependent data. It will also touch on the influence of variance structure on statistical power, and the possibility that misspecification of the variance structure as a one reason for over-statement of evidence supporting research findings.

The slides from the talk

Neurogenetics approaches to understanding the developement of antisocial behavior

Luke Hyde, Assistant Professor, Psychology, Research Faculty Affiliate, Center for Human Growth and Development

Date: Oct 24, 2017

Time: 12:00 PM

Location: Room 1000, 10th floor, N Ingalls Building

Dr. Hyde’s recent program of research has been merging imaging genetics techniques that aim to understand genetic and molecular contributions to neural reactivity with longitudinal developmental studies of at risk children in order to inform our understanding of the development of antisocial behavior, psychopathy, and psychopathology across the lifespan.

His research focuses on mechanisms linking early risk to adolescent antisocial behavior. In particular, the role of cognitions, empathy (and callous/unemotional traits), genes (using candidate genes), and neural processes (using fMRI) as they are affected by and interact with harsh environments to increase risk for psychopathology.

First U-M Workshop on Computational Neuroscience

Organized and presented by MICDE and U-M's Neuroscience Graduate Program

Date: Monday, Oct 16, 2017 (Fall study day)

Time: 9:00 AM to 1:00 PM

Location: 2435 North Quad, 105 South State St

Description: The Michigan Institute for Computational Discovery & Engineering (MICDE) and the U-M Neuroscience Graduate Program have organized the 1st U-M workshop on Computational Neuroscience. The goal is to bring together the large U-M community of neuroscientists that use computational tools in their research, and to start building new bridges across departments and disciplines.

For the Agenda and speakers, please see the 1st U-M Workshop on Computational Neuroscience web site.

Fall 2017 Functional MRI Symposium

Organized and presented by the Functional MRI Laboratory at the University of Michigan

Date: Sep 29, 2017

Time: 9:30 AM – 4:00 PM

Location: 4448 East Hall (the final talk will be in 4464 East Hall)


Naftali Raz, Wayne State University,
Structural Characteristics of the Brain in the Context of Adult Development and Aging

Alex Iordain, University of Michigan,
Brain Graphs: Network Analysis of Function MRI Data

Doug Noll, University of Michigan,
New Approaches to fMRI Image Acquisition

Chandra Spripada, Daniel Kessler, and Mike Angstadt, University of Michigan,
Meta-Bases–Searching for the Fundamental Units of Brain Activation in Massive Repositories of Task Contrasts

Bennet Fauber, University of Michigan,
Advanced MRI processing capabilities at UM

Scott Peltier and Krisanne Litinas, University of Michigan,
New Initiatives in the Functional MRI Laboratory and Hands-On Data Quality Inspection

Navigating resting-state streams: Making sure you're up the creek with the right paddle

Date: Aug 1, 2017

Time: 4:00 PM

Location: 4464 East Hall

Resting state data are an important part of many projects, and recommendations for how to analyze it are complex and frequently updated. What's the “best” method for dealing with potential motion confounds? Should you or shouldn't you control for the global signal–and if so, why and how? How should you analyze connectivity: using ICA, seed-based analyses, graph theory–or something else?

Hailey Dotterer from Luke Hyde's lab will tell us about a new project involving resting state data and ask people's opinions about current recommendations for preprocessing and other aspects of the analysis. This will be an opportunity to revisit resting state analysis and issues with preprocessing data for it. We will also examine a paper by Vergara and Mayer that demonstrates how to use pipelines to quickly and systematically evaluate how the decisions you make about these questions may be affecting your results. Come tell us how you are doing your analyses, and learn from how others are doing theirs.

The global signal in fMRI: Nuisance or information, Thomas T Liu, Alican Nalci, Maryam Falahpour, Neuroimage, 150 (2017) 213–229
Studying brain organization via spontaneous fMRI signal, Jonathan D Power, Bradley L Schlaggar, Steven E Petersen, Neuron Volume 1, Issue 4, 19 Nov 2014, 681–696
The effect of preprocessing pipelines in subject classification and detection of abnormal resting state functional network connectivity using group ICA, Victor M Vergara, Andrew R Mayer, Eswar Damaraju, et al., Brain Connectivity Volume 1, Number 1, 2011, 365–376

Click here for the slides from the talk


Report from the Exploring the Human Connectome workshop

Date: Jul 25, 2017

Time: 4:00 PM

Location: 4464 East Hall

Saige Rutherford recently attended a Human Connectome workshop prior to the OHBM conference. She will do a demonstration of the HCP's Connectome Workbench software and some of the impressive visualizations that it provides, particularly for DTI data. In addition, Saige will review some of what was said with regard to dense versus parcellated and surface versus volume analyses and some recommendations the HCP folks had.

Additional resources for HCP

If you want to go through all of the workshop materials on your own, you can use

Exploring the Human Connectome. Be aware this takes up a substantial amount of space as it includes the virtual machine and data used to complete some of the exercises. (You will need at least 1TB of free disk space for downloading/extracting the data. The Practical Data are very large (~350 GB) and are downloaded as a compressed archive.)

BALSA (HCP's website) that discusses the multimodal parcellation in detail. This is the scene file that can be downloaded and opened in Connectome Workbench and shows the parcellation with labels.

Study: A Multi-modal Parcellation of Human Cerebral Cortex

A web version of a poster given at OHBM that has good (brief) info about why you should try to use the parcellation in surface space, not volume.

Impact of Traditional Neuroimaging Methods on the Spatial Localization of Cortical Areas. See the link to E-Poster on that page for the .pdf version.

Some links for the HCP's parcellation in volume space

Here are two links to download the HCP parcellation in volume space, to use as an atlas.

HCP-MMP1.0 projected on MNI2009a GM (volumetric) in NIfTI format

MMP 1.0 MNI projections

Here is a link to recreating the parcellation in subject-specific volume space

HCP-MMP1.0 volumetric (NIfTI) masks in native structural space

An article on projecting the HCP-MMP1.0 parcellation onto fsaverage, along with files and notes for doing it yourself.

HCP-MMP1.0 projected on fsaverage

Where is your white matter going (and how sure can you be)?: An overview of probabilistic tractography and how to scale up from processing from one subject to processing 171 while you go enjoy your weekend

Date: Jul 18, 2017

Time: 4:00 PM

Location: 4464 East Hall

ReadingA Network of Amygdala Connections Predict Indvidual Differences in Trait Anxiety, Steven Greening G and Derek GV Mitchell, Human Brain Mapping, 36 (2015) 4819–4830

Leigh Goetschius, Bennet Fauber will talk about a method for exploring anatomical connections in the brain (which don't always map onto functional connectivity!) and their integrity, illustrating this in an actual dataset from the TAD Lab. This is still a work in progress, so we'll also describe some of the issues we struggled with and how we resolved them (or are still figuring them out) – and suggestions are welcome!

We'll also use this as a way to illustrate a method for automating your analyses to make it easier to deal with big datasets, and how this can facilitate moving your analyses to the cluster if your data processing needs outgrow the computers in your lab.

The slides from the talk (PowerPoint)

We will post the scripts, or links to the scripts soon.

The policies for the LSA public Flux allocations

The policies for the Engineering public Flux allocations

Preprocessing and Quality Assurance

Date: May 23, 2017

Time: 4:00 PM

Location: 4464 East Hall

Scott Peltier and Krisanne Litinas from the fMRI Lab will continue the previous preprocessing discussion, showing how to check the output of each preprocessing step for both functional and anatomical streams, and give examples of different kinds of artifacts to watch out for. They will also discuss recent work done with Alex Iordan of Psychology and Mike Angstadt of MethodsCore, examining iterative reconstruction and fieldmap correction methods.

Video recording of the talk
The slides from the talk
The slides from the previous talk on artifacts
Sample fMRI Lab flex file This is the file provided by the fMRI Lab with many helpful QA graphs and images for each scan.

Inferring, Summarizing and Mining Large-Scale Graph Data

Date: Friday, Apr 21, 2017

Time: 4:00–5:00 PM

Location: Rackham Amphitheater

Abstract: Networks naturally capture a host of real-world interactions, from social interactions and email communication to brain activity. However, graphs are not always directly observed, especially in scientific domains, such as neuroscience, where monitored brain activity is often captured as time series. How can we efficiently infer networks from time series data (e.g., model the functional organization of brain activity as a network) and speed up the network construction process to scale up to millions of nodes and thousands of graphs? Further, what can be learned about the structure of graph? How can we summarize its most important properties by taking into account the properties of other graphs in that domain (e.g., neuroscience)? In this talk I will present our recent work on scalable algorithms for inferring, summarizing and mining large collections of graph data. I will also discuss applications in various domains, including connectomics and social science.

This talk is part of the MIDAS Seminar series

Danai Koutra

Multivoxel pattern analysis (MVPA) of fMRI data: Tutorial and examples

Date: Mar 7, 2017

Time: 4:00 PM

Location: 4464 East Hall

Thad Polk from the Psychology department will give a presentation introducting Multivoxel pattern analysis (MVPA) of fMRI data. Basic concepts will be presented along with a demonstration. There will be a code example written for Matlab shown.

The slides from this talk are posted here. Multi-variate/voxel pattern analysis (MVPA) (PDF)

The supplemental slides that were mentioned are posted here. Caveats for MVPA (PDF)

The example script, which has both k-nearest neighbor and SVM classification examples it. Example script

The Matlab .mat file that it gets its data from. Note, you might need to Ctrl-Click on this link to get the menu and select Save link as for this one. Life is different in different browsers. Matlab data file

Brain Hack, 2017

Date: March 2–5, 2017

Time: Varies by day

Location: North Campus

Brainhack was a project-based work party, along with a couple of good talks. We looked at converting the fMRI center data organization (MIDS) to a minimally viable Brain Imaging Data Structure (BIDS) format, and we looked at getting the fmriqc Nipype pipeline to work on the BIDS format. Both projects are incomplete, but when they are, we will post them. If anyone would like to help with those, please let us know at michigan-nii@umich.edu.

These are the slides from the talk given by Cindy Lustig, Pushing Beyond the PPI: Regional Homogeneity and Dynamic Functional Connectivity Analysis.

We will post slides from the talk given by Ivo Dinov on the LONI Pipeline later.

Introducing the Advanced Computational Neuroscience Network (ACNN)

Date: Feb 8, 2017

Time: 2:00 PM

Location: 4464 East Hall

Ivo Dinov and Rich Gonzalez will give an introduction to the Advanced Computational Neuroscience Network (ACNN). ACNN (NeuroscienceNetork.org) is a collaboration of six Midwestern universities that aims to develop standards, build tools and techniques, and promote open collaborative neuroscience. Specifically, ACNN investigations are focused on integration of 'Big Neuroscience Data', Neuroimaging, and high-throughput analytics projects.

Rich and Ivo will speak, in general, to existing techniques and technologies, to some of challenges -- theoretical, methodological, technical -- that will be available over the next several years, and to some of the opportunities for integrating imaging and non-imaging, structured and unstructured, and complex data with different representation bases.

Computing Skills Series: Neuroimaging workflows: What works, what flows?

Date: Jan 31, 2017

Time: 4:00 PM

Location: 4464 East Hall

At the Jan 31 meeting of the Computing Skills group, we will use the Askren, et al., paper as a jumping off place to discuss What is a workflow (or pipeline)? Why make them? Processing many subjects at once? How not to process needlessly. We hope others will contribute their experience. Later in the semester, we hope to start looking at the practical issues in constructing pipelines using various tools: the shell, Python, Matlab, etc.

Reading Mary K Askren, et al. Using Make for Reproducible and Parallel Neuroimaging Workflow and Quality-Assurance http://journal.frontiersin.org/article/10.3389/fninf.2016.00002/full

Decision Consortium

Mark Reimers, Statistical methods for big data can help elucidate neural processes during decisions


Big data is transforming functional neuroscience, as microarrays and sequencing transformed functional genomics. In particular high-throughput technologies are changing the relationship between theory and experiment. I will argue that time is the crucial element needed in models of cognitive processes that will make them comparable with emerging high-throughput data.

I will discuss several examples of how high-throughput data is giving us insight into cognitive processes. The first example is described in a paper from the Newsome lab (Mante et al.), which claims to falsify several leading models of perceptual decision-making. I will present their methodology and we will discuss the robustness and implications of their conclusions. Then I will present some results from my analyses of high-throughput recordings obtained during decision processes, using data provided by other labs, and we will discuss what other insights are emerging.

Date: 20 Oct, 2016
Time: 3:00 – 4:30 PM
Location: 4464 East Hall


Context-dependent computation by recurrent dynamics in the prefrontal cortex”, Valerio Mante, David Susillo, Krishna V Shenoy, and William T Newsome, Nature, 503, 78-84, doi:10.1038/nature12742

Neuroscience: What to do and how” (Commentary on Mante, et al.), Jeffrey C Erlich and Carlos D Brody Nature, 503, 45-47 doi:10.1038/503045a

Decision Consortium and Journal Club

Daniel Kessler, Which Findings from the Functional Neuromaging Literature Can We Trust? Putting ‘Cluster Failure’ in Context

Date: 6 Oct, 2016
Time: 3:00 – 4:30 PM
Location: 4464 East Hall


Slides from the talk

Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates. Anders Eklund, Thomas E. Nichols, and Hans Knutsson. PNAS 113:22, 7900-7905. DOI: 10.1073/pnas.1602413113

Which Findings from the Functional Neuromaging Literature Can We Trust?. Daniel Kessler, Michael Angstadt, Chandra Sripada. arXiv:1608.01274v1 [stat.AP]

Multiple testing corrections, nonparametric methods, and random field theory. Nichols, TE. DOI: 10.1016/j.neuroimage.2012.04.014 (UM affiliates can access the full text vi UM Library).

fMRI Symposium

The Fall 2016 Functional MRI Symposium will be held in the Colloquium Room, located in East Hall, 4th floor, room 4448. Lunch and beverages will be provided. The day will be devoted to talks that represent a range of research from methodology to bioengineering to biostatistics to neuroscience.

Date: 30 Sep, 2016
Location: 4448 East Hall
Agenda: The agenda will be updated regularly
Event web site