Keynote Speakers Abstract:The central nervous system has an inbuilt capacity to modify pain intensity adaptively in relation to the context. As such, pain perception for the most part, does not directly reflect activity in peripheral nociceptive signals. Instead pain is a composite of top-down expectations/prior learning and bottom-up nociceptive processes. Studies on placebo response indicate that initiating a new treatment induces positive expectations and shifts mental states to be conducive for stronger analgesic responses. However, inability to build positive expectations and appropriate mental states (i.e., the right ‘mind set’) before starting treatment can negatively affect treatment outcomes. Previously, Dr. Hashmi and others have reported that expectation-effects on pain can be predicted by observing brain network properties at baseline. New data from her lab offers insights on the nature and extent to which top-down signals in the brain can modulate pain. This talk will discuss how mapping activity in top-down brain circuits in people with chronic pain and healthy subjects is expected to offer new directions for improving chronic pain diagnosis and treatment outcomes. Dr. Hashmi’s research shows that configurations of connectivity in resting fMRI are predictive of intrinsic aspects of treatment response. This finding was observed in experimental models of placebo analgesia and in clinical trials in a few different studies and in different groups of participants. Recent findings from her lab show that the identified connectivity patterns also predict participant engagement in cognitive training programs. Her lab has developed new cognitive models for evaluating top-down bias in pain perception. Overall, this research offers new evidence for a role of prior states in treatment response. Hence, this knowledge combined with frameworks such as network analysis and machine learning can be expected to improve diagnosis and treatment selection for brain disorders. Dr. Hashmi is an Assistant Professor and a Canada Research Chair Tier II at Dalhousie University, Canada. She received her PhD from the University of Toronto and has trained as a postdoctoral fellow at Northwestern University and Harvard Medical School. Dr. Hashmi’s research focuses on applying computational methods to understand the role of brain activity and connectivity in pain perception and cognition. She aims to generate ideas and tools from brain imaging that can impact clinical management of brain-related disorders.  Title: Mapping individual variability and development of the brain connectivity architecture Abstract: The variability and development of the architecture of the human brain are closely interlinked. The relationship between anatomy and function observed across individuals is complex, and at the same time revealing regarding its underlying mechanisms. In this talk I will show machine learning approaches to capture and decouple functional and anatomical variability. Based on the observations gained with these methods we will discuss the link between individuality of function and anatomy, early brain development, and evolution.  Juan (Helen) Zhou Title: Brain functional connectome topology and dynamics: applications in psychiatric disorders and behavioral prediction. Abstract: Neuropsychiatric disorders target large-scale neural networks. In the past decade, network-sensitive neuroimaging methods have made it possible to test the notion of network-based degeneration in living humans. Previous work has demonstrated that the spatial patterning of each disease relates closely to a distinct functional intrinsic connectivity network mapped in the healthy brain with task-free or “resting-state” functional magnetic resonance imaging (fMRI). This talk will first describe how brain network functional connectome using graph theoretical analyses shed light on the differential vulnerable neural substrates in dementia subtypes and network-breakdown mechanisms in neurodegenerative disorders. Our recent work on persons at-risk for psychosis found that baseline functional connectome topological disruptions mainly in the salience network were associated with future transition to psychosis. Furthermore, brain dynamic functional connectivity methods and how dynamic connectivity states relate to vigilance fluctuations over time will be presented. Further developed and integrated with structural brain measures, brain network connectome signatures may help us predict behavioral changes, reveal disease mechanism, track disease progression, and monitor treatment response. Dr. Juan (Helen) Zhou is an Assistant professor at the Center for Cognitive Neuroscience, Neuroscience and Behavioral Disorders Program, Duke-National University of Singapore Medical School, Singapore. Her lab studies the human neural bases of cognitive functions and the associated vulnerability patterns in aging and neuropsychiatric disorders, particularly neurodegenerative diseases. Multimodal neuroimaging methods and psychophysical techniques are employed, including MRI/fMRI/DTI/EEG. Prior to joining Duke-NUS, Helen was an associate research scientist at the Phyllis Green and Randolph Cōwen Institute for Pediatric Neuroscience in the Child Study Center, New York University. She did a post-doctoral fellowship at the Memory and Aging Centre, Department of Neurology, University of California, San Francisco, from 2008 to 2010. Helen received her Bachelor degree in Computer Engineering with first class honour in 2003 under the scholarship from Ministry of Education, Singapore and her Ph.D. in Neuroimaging in 2007 from School of Computer Engineering, Nanyang Technological University, Singapore. Helen has published in several international peer-reviewed high impact journals such as Neuron, Brain, PNAS, Neurology, Molecular Psychiatry, Biological Psychiatry, Journal of Neuroscience, NeuroImage, Human Brain Mapping and so on. She is a program committee member of the Organization of Human Brain Mapping, a member of the Society for Neuroscience, American Academy of Neurology, and Alzheimer’s Association. Helen also serves as editors and reviewer for various journals and grants. Dr. Zhou has been the recipient of research support from National Medical Research Council and Biomedical Research Council, Singapore as well as the Royal Society, UK.  Title: Deep image reconstruction from the human brain. Abstract: The internal visual world is thought to be encoded in hierarchical representations in the brain. However, previous attempts to visualize perceptual contents based on machine-learning analysis of fMRI patterns have been limited to reconstructions with low level image bases or to the matching to exemplars. While categorical decoding of imagery contents has been demonstrated, the reconstruction of internally generated images has been challenging. I introduce our recent study showing that that visual cortical activity can be decoded (translated) into the hierarchical features of a pre-trained deep neural network (DNN) for the same input image, providing a way to make use of the information from hierarchical visual features (Horikawa & Kamitani, Nature Comm,. 2017). Next I present a novel image reconstruction method, in which the pixel values of an image are optimized to make its DNN features similar to those decoded from human brain activity at multiple layers (Shen, Horikawa, Majima, Kamitani, bioRxiv 2017). We found that our method was able to reliably produce reconstructions that resembled the viewed natural images. While our model was solely trained with natural images, it successfully generalized to artificial shapes, indicating that our model was not simply matching to exemplars. The same analysis applied to mental imagery demonstrated rudimentary reconstructions of the subjective content. Our method can effectively combine hierarchical neural representations to reconstruct perceptual and subjective images, providing a new window into the internal contents of the brain.
 Title: Generative modelling of resting-state cortical parcellations and dynamics. Abstract: I will discuss recent generative models from my lab for estimating cortical parcellations and exploring dynamical brain organization using resting-state fMRI. First, we develop a gradient-weighted Markov Random Field model to integrate two popular parcellation approaches: local gradient and global similarity. The local gradient approach detects abrupt transitions in functional connectivity patterns, while the global similarity approach clusters similar functional connectivity patterns regardless of spatial proximity. Our fusion model yields a new population-level parcellation with hundreds of highly homogeneous parcels, which approximate classically defined cortical areas from systems neuroscience. Second, we propose a multi-session hierarchical Bayesian model (MS-HBM) for estimating individual-specific cortical networks. The multiple layers of the model explicitly differentiate intra-subject from inter-subject network variability. By ignoring intra-subject variability, previous individual-specific network mappings might confuse intra-subject variability for inter-subject differences. Furthermore, many studies have shown that individual-specific network topography varies substantially across participants. We show that individual-specific network topography and size can predict cognition, personality and emotion. Finally, to bridge the gap between micro-scale and macro-scale brain organization, we extend an influential biophysical generative model to allow different brain regions to possess distinct micro-scale properties. Inverting the model reveals a macro-scale hierarchy with sensory-motor cortex and the default network at opposite ends. In contrast to the default network, sensory-motor regions exhibit strong recurrent connections and strong subcortical inputs. The strength of recurrent connections is associated with neuronal density, suggesting a cellular basis for a large-scale hierarchy in the dynamic resting brain. Dr.Thomas Yeo is an Assistant Professor at the National University of Singapore (NUS) Department of Electrical and Computer Engineering, Clinical Imaging Research Center, and Singapore Institute for Neurotechnology. He is also an affiliated faculty at Duke-NUS Medical School and Harvard Medical School. Thomas received his B.S. and M.S. from Stanford University and Ph.D. from the Massachusetts Institute of Technology. Prior to NUS, Thomas was a research fellow at Harvard University and Duke-NUS Graduate Medical School. With the deluge of data in multiple scientific disciplines, future scientific breakthroughs will be made by new algorithms exploring these massive datasets. Thomas’ lab develops machine learning algorithms to make scientific discoveries from large-scale brain imaging data, comprising thousands of subjects and billions of measurements per subject. Thomas is an editor at NeuroImage and is a recipient of the A*STAR National Science Scholarship (2004), the MICCAI Young Scientist Award (2007), the MICCAI Young Investigator Publication Impact Award (2011), and the National Research Foundation Fellowship (2017). |