Friday April 25th, 2003
NEUROINFORMATICS 2003
Speakers
David A. Rottenberg, M.Sc. (Cantab), M.D.
Professor of Neurology and Radiology, University of Minnesota and
Director, International Neuroimaging Consortium
As described in a 1993 NIH Program Announcement (PA-93-068), the Human Brain Project began as a broadly-based federal research initiative to encourage and support basic and clinical neuroscience research and research on informatics that could be used to facilitate neuroscience research. Particular emphasis was placed on computer storage, the manipulation of neuroscience data, networked computer systems, and software tools that would allow neuroscientists to access stored information. This initiative was initially supported by five NIH Institutes (most notably the National Institute of Mental Health), the National Center for Research Resources, the National Science Foundation, the National Library of Medicine, the Office of Naval Research, and the National Aeronautics and Space Administration. The Human Brain Program as well as subsequent NIH initiatives, e.g., the Biomedical Information Science and Technology Initiative (BISTI), have served to define the emerging discipline of neuroinformatics and to provide researchers with the ability to manage, integrate, and synthesize neuroscience information across disciplinary and geographic boundaries.
Top | Back to Neuroinformatics 2003 Symposium
Stephen H. Koslow, Ph.D.
Associate Director and Director, Office on Neuroinformatics
National Institute of Mental Health
Neuroinformatics: What, Why, and How
The Information Technology (IT) revolution has changed contemporary society and has begun to alter the conduct of science, as best exemplified by Bioinformatics. Neuroscience is also being impacted by IT in order to meet the challenge of enormous datasets and the integration of data across the multiple levels of analysis at which neuroscientists now work. To meet this challenge the new field of Neuroinformatics is developing. Neuroinformatics combines research in neuroscience, informatics and computation to develop distributed databases, advanced analytical and computational tools, and novel approaches for understanding the brain.
This presentation will provide the contextual need for Neuroinformatics, examples of new capabilities initiated through Human Brain Project grants, as well as new International initiatives and societal challenges to shifts in the scientific paradigm.
Top | Back to Neuroinformatics 2003 Symposium
Edward H. Herskovits, M.D., Ph.D.
Associate Professor of Radiology, University of Pennsylvania
Principal Investigator, The Brain-Image Database Project
Microarray Analysis: Overview and Potential Applications to Neuroscience
Microarrays show great promise for the delineation of genetic determinants of neurological disease and of embryonal development. Fulfillment of this promise requires close collaboration between neuroscientists and computer scientists, attention to experimental design, and the development of image-processing and data-analysis or data-mining methods that accomodate the unique nature of microarray data.
I will describe several approaches to the analysis of microarray data and discuss how these analyses might contribute to our understanding of the human brain.
Top | Back to Neuroinformatics 2003 Symposium
Arthur W. Toga, Ph.D.
Associate Professor of Neurology and Director, Laboratory of Neuro Imaging
UCLA School of Medicine
Databases are Atlases of Brain Structure and Function in Health and Disease
The ubiquitous application of neuroimaging to the study of brain structure and function in health and disease has resulted in extraordinary collections of brain images worldwide. However, until recently, these collections were largely independent and uncoordinated efforts. That is changing with advances in technology and changes in the sociology of science policy makers and scientific interest. Is there anything special about neuroimaging that lends itself to the creation of databases, and are there certain characteristics of databases that either must be present or cannot be included? The complexity and variability of the human brain (as well as that of other species) across subjects is so great that simple cataloguing of multisubject brain imagery is inadequate to fully utilize the constructs of a database. Instead, the variability across subjects that initially poses obstacles to their combination should be considered valuable information about the similarities and differences within and between populations.
This talk makes the argument that a population brain atlas is equivalent to a database. A useful database consisting of neuroimages is an atlas. An atlas consists of brain image data, its interpretation, a coordinate system to facilitate indexing, and innumerable demographic, cognitive, behavioral and even genetic characteristics of each subject. The atlas as database is essential to effectively manipulate, analyze and interpret brain data. Central to these tasks is the construction of averages, templates and models to describe how the brain and its component parts are organized. Design of appropriate reference systems for human brain data presents considerable challenges, since these systems must capture how brain structure and function vary in large populations, across age and gender, in different disease states, across imaging modalities, and even across species. Only in an atlas with these systems in place will the full richness and complexity of a database be realized.
Top | Back to Neuroinformatics 2003 Symposium
David N. Kennedy, Ph.D.
Associate Professor of Neurology and Radiology, Harvard Medical School
MGH/MIT/HMS Athinoula A. Martinos Center for Biomedical Imaging and Center for Morphometric Analysis
Departments of Radiology and Neurology, Massachusetts General Hospital
Neuroinformatics and Human Brain Morphometry
The neuroinformatics landscape in which human brain morphometry resides has advanced dramatically over the past few years. We are actively developing neuroinformatics tools that span a number of important areas in this landscape.
Tools need validation prior to their use and dissemination. Thus, we have created a website to facilitate the testing, validation, and comparison of neuroanatomic image analysis tools, the Internet Brain Segmentation Repository (IBSR). This site contains test image data sets which permit a standardized mechanism for evaluating the sensitivity of a given analysis method to signal-to-noise ratio, contrast-to-noise ratio, shape complexity, degree of partial voluming, etc.
Once a tool has been developed and validated, it should be shared; a tool that only works for one or a small group of investigators is not destined to make a large impact on our fundamental understanding of neuroscience. Our development of an Internet Analysis Tools Registry (IATR), is specifically designed to facilitate tool sharing for developers and consumers of tools.
In addition to the tools themselves, the results of tool use (data) should also be shared. In order to facilitate multisite morphometry data integration, we have developed the Internet Brain Volume Database (IBVD), one of the numerous federated databases that reposit neuroscience data. Through such examples of federated and integrated neuroinformatics resources, progress towards the delivery of a comprehensive infrastructure to the neuroscience community is being achieved.
Top | Back to Neuroinformatics 2003 Symposium
Lars K. Hansen, Ph.D.
Professor, Technical University of Denmark
Director, THOR Center for Neuroinformatics and
Head, Copenhagen Image and Signal Processing Graduate School
Getting the Most out of Neuroimaging Experiments: Multivariate Modeling and Meta-analysis
Multivariate statistical tools can be used for exploratory analysis, hypothesis testing, and meta-analysis in neuroimage databases. In this talk I will review recent progress in multivariate model design and evaluation in neuroimaging, including the use of mutual information learning curves, model consensus, and datamining techniques.
Top | Back to Neuroinformatics 2003 Symposium
Stephen C. Strother, Ph.D.
Associate Professor of Radiology and Neurology, University of Minnesota
Testing and Managing Heterogeneous Tools for Neuroimaging Experiments
A wide range of techniques and software tools have become available with which to acquire, manage, and process functional neuroimaging datasets. This has only recently been accompanied by attempts to develop software environments, performance metrics, and benchmark datasets with which to manage, evaluate and compare these tools.
During this talk I will introduce the idea of a "meta-model" that incorporates all the experimental decisions on which the final neuroimaging data depend. These decisions include the experimental design, data acquisition and preprocessing strategies, and the final data analysis step. I will discuss some of the strategies and techniques that are available for testing and optimizing such neuroimaging meta-models, particularly the trade-offs that exist between simulation-driven testing using Receiver Operating Characteristic curves and data-driven testing based on cross-validation resampling techniques, such as NPAIRS and learning curves. In order to manage this testing process its elements should be usefully combined in an accessible, user-friendly software environment. I will end my talk by discussing our experience with the Java-based FisWidgets environment for managing and testing heterogeneous tools.
Top | Back to Neuroinformatics 2003 Symposium
Christopher R. Johnson, Ph.D.
Professor of Computer Science, Research Professor of Bioengineering, Adjunct Professor of Physics and
Director, Scientific Computing and Imaging Institute, University of Utah
Computational Bioimaging and Visualization
“The future of radiology is not getting the data — we have plenty of data. The need for the future is how to make use of all the data we have.”
The next decades will see an explosion in the use and the scope of medical imaging. In my opinion, advanced, multimodal imaging techniques, powered by new computational methods, will change the face of biology and medicine. These imaging modalities will produce information about anatomical structure that is linked to functional data, in the form of electric and magnetic fields, mechanical motion, and metabolism. This integrated approach will provide comprehensive views of the human body in progressively greater depth and detail, while such visualizations gradually become cheaper, faster, and less invasive. As a result, computer assisted imaging will become more ubiquitous, which will, in turn, produce new scientific and clinical specialties that rely on particular combinations of imaging, computer science, and medicine.
I will present recent research results in computational imaging and visualization, including new methods for interactive visualization of large-scale 3D MRI and CT volumes, and introduce new methods for diffusion tensor imaging.
Top | Back to Neuroinformatics 2003 Symposium
Lael C. Gatewood, Ph.D.
Professor of Laboratory Medicine and Pathology, University of Minnesota and
Director of NLM Training Program in Medical Informatics
Summary and Conclusions
The growth of Neuroinformatics depends in part on participation in collaboratories, defined by William Wulf in 1989 as "Centers without walls...where users can perfom research without regard to geographical location...interacting with colleagues accessing instruments, sharing data and computational resources [and] accessing information in digital libraries." No one institution can assemble all of the resources needed to support up-to-date studies in Neuroinformatics. Tools are needed for purposes of image acquisition, processing, modeling, analysis, visualization and archiving. Reference databases are needed for defining formats and protocols, setting regions of interest, testing informatics products and algorithms, and adding to scientific knowledge bases. The Human Brain Project (HBP), with its 15 funding agencies and its annual scientific meeting of more than 50 investigators, has been instrumental in coordinating and communicating results for the field of neurosciences. The HBP Director Dr. Koslow, a speaker today, has extended the research foci across the world by chairing the Neuroinformatics Guidelines Committee of the Global Science Forum for the Organization for Economic Cooperation and Development (OECD). Drs. Toga and Kennedy edit journals in the discipline, and other symposium speakers provide sharoed datasets, protocols, algorithms, and other manner of collaborative resources.
Other parts of the NIH/NSF organizations are building resource networks or collaboratories for the future. NIH has established two new institutes, to fund and share the results of Human Genomic Research (NHGRI), and to support Biomedical Imaging and Bioengineering (NIBIB). The NIH's National Center for Research Resouces (NCRR) has initiated a twenty-center strong Biomedical Informatics Research Network (BIRN), utilizing research involving neuroimaging as a testbed to take advantage of the advanced level of sophistication of information technology in this community. The NSF Directorate on Computer & Information Sciences & Engineering (CISE) commissioned a blue ribbon panel on Cyberinfrastructure that delivered its report at the beginning of this year. The report speaks to investing in a broad set of services ranging from the base technologies of computation, storage and communication; through mid-level suport of networking, operating systems and middleware; to supportive high-level services for computation, knowledge management, instrumentation, visualization and collaboration to provide community-specific knowledge environments for research and education. This is where Neuroinformatics 2003 needs to reach, and those presenting today are part of that realization.
Top | Back to Neuroinformatics 2003 Symposium