ICASSP'2000 Tutorial
ICASSP'2000 Tutorial
Signal processing strategies in Neuroinformatics


Lars Kai Hansen
Department of Mathematical Modeling
Building 321
Technical University of Denmark
DK-2800 Lyngby, DENMARK
email: lkhansen@imm.dtu.dk
http://eivind.imm.dtu.dk

Abstract

This is the homepage for an ICASSP'2000 tutorial on Neuroinformatics with specific emphasis on signal processing aspects. Neuroinformatics is a new interdisciplinary research field which combines ``learning from the brain'' with ``learning about the brain''. This tutorial will focus on signal processing strategies for neuroimaging, and includes discussions of data preprocessing, data modeling by clustering and neural networks, and information visualization by VRML (virtual reality modeling language). A comprehensive Matlab toolbox for fMRI signal processing including example neuroimaging datasets is offered free of charge for scientific use.

Background


Neuroinformatics is a research field rooted in classical disciplines like biology, physics, computer science, and engineering. Neuroinformatics combines learning from the brain with learning about the brain. By studying information processing in the brain neuroinformatics invents new engineering paradigms (e.g. artificial neural networks) with the grand but distant objective to understand the dynamics of the conscious mind. Neuroinformatics is a key component of the US research program, the Human Brain Project which is supported by major American governmental funding agencies. The vast databases created by current brain scanning techniques calls for new innovative signal processing strategies and for for new visual and interactive means of communication and form a significant engineering challenge for the next century. A point strongly emphasized in Human Brain Project is the need for using the sc www for communication and dissemination of results as well as information infrastructure buildup. Neuroinformatics uses of multimedia techniques are now beginning to appear.

One of the major efforts in neuroinformatics concerns neuroimaging -- mapping brain function by PET and fMRI (positron emission tomography and functional magnetic resonance imaging). Neuroimaging is opening a new window to the working human brain and brain mapping will have immense influence on the way we think about the brain both in research and in the clinic.
Adaptive systems and neural networks in particular construct abstract models by learning. From an application point of view such knowledge gains credibility by being expressed in intuitively accessible forms, say by computer graphics, and interactively in the form of hypertext. With the advent of visual hypertexts such as clickable maps or VRML (Virtual Reality Modeling Language) graphics and hypertext are combined. The DTU-group has pioneered the use of VRML in neuroscience contexts. The group's work can be accessed at http://eivind.imm.dtu.dk/thor (group homepage) http://hendrix.imm.dtu.dk (THOR center repository).
On the American site http://neuroguide.com the work is characterized as ``A glimpse of the future. Virtual Reality Modeling Language (VRML) as applied to the human brain, among other technical advances''.




Tutorial outline

The tutorial will review the basic neuroimaging data acquission schemes. State the engineering challenges associated with data analysis: normalization, intersubject co-registration by warping, modeling using principal and independent component analysis (PCA, ICA) an linear and non-linear representations, information visualization and communication. The presentation will illustrated by experimental data involving PET and fMRI studies of motor activity, visual stimulation as well as cognitive tasks.
  • Neuroinformatics: definitions, background and major projects around the world
  • Neuroimaging: 3+1 D, the spatio-temporal structure of PET and fMRI data sets.
  • Signal processing challenges and current strategies for preprocessing and representation: PCA, ICA.
  • The problem of extremely ill-posed learning (learning high-dimensional models from small databases).
  • Learning non-linear relations between blood flow and function in the brain.
  • Information visualization: VRML
  • Perspectives and open research issues

    Our group is offering a comprehensive Matlab toolbox for fMRI signal processing ``lyngby''. The toolbox is accompanied by example fMRI data sets, offers an ideal introduction to the field, and is offered free of charge for the scientific community. Passcode for lyngby can be obtained by emailing lyngby@hendrix.imm.dtu.dk, the documentation (postscript and html) can be accessed via our software page

    The presentation and the hand-outs will be based on the following publications:

  • Benny Lautrup, Lars Kai Hansen, Ian Law, Niels Morch, Claus Svarer, and Stephen C. Strother: "Massive Weight Sharing: A cure for Extremely Ill-posed Problems" Workshop on Supercomputing in Brain Research: From Tomography to Neural Networks, HLRZ, November 21-23, pp 137-144, (1994).
  • S.C. Strother, J.J. Sidtis, J.R. Anderson, L.K. Hansen, K. Schaper, and D.A. Rottenberg: "15-O Water PET: More ``Noise'' than signal?" In Quantification of Brain Function Using PET, Eds R. Myers, V. Cunningham, D. Bailey, and T. Jones. Academic Press Chapter 73, p 378-383, (1996).
  • S.C. Strother, N. Lange, J.R. Anderson, K.A. Schaper, K. Rehm, L.K. Hansen, and D.A. Rottenberg: "Activation pattern reproducibility: Measuring the effect of group size and data analysis models". Human Brain Mapping vol. 5, 321-316 (1997).
  • N. Morch, L.K. Hansen, S.C. Strother, C. Svarer, D.A. Rottenberg, B. Lautrup, R. Savoy, O.B. Paulson: "Nonlinear versus Linear Models in Functional Neuroimaging: Learning Curves and Generalization Crossover". Proceedings of the 15th International Conference on Information Processing in Medical Imaging, 1997, Eds. J. Duncan and G. Gindi. Lecture Notes in Computer Science vol. 1230, Springer Verlag 259-270, (1997).
  • F.AA. Nielsen and L.K. Hansen. "Interactive Information Visualization in Neuroimaging". NPIVM'97 - Workshop on New Paradigms in Information Visualization and Manipulation. David S. Ebert, Charles K. Nicholas (eds.) ACM ISBN 1-58113-051-1. (1997)
  • L.K. Hansen, N. Morch, and F.AA. Nielsen: "Neural Net in Functional Neuroimaging". Tutorial Paper for NORSIG'98. In Proceedings of the IEEE Nordic Signal Processing Symposium, Vigso, Denmark 1998. Eds. P. Dalsgaard and S.H. Jensen, pp. 1-8, (1998).
  • C. Goutte, P. Toft, E. Rostrup, F.AA. Nielsen and L.K. Hansen: "On clustering fMRI time series". NeuroImage vol. 9:298-310, (1999).
  • L.K. Hansen, J. Larsen, F.AA. Nielsen, S.C. Strother, E. Rostrup, R. Savoy, N. Lange, J.J. Sidtis, C. Svarer, O.B. Paulson: "Generalizable Patterns in Neuroimaging: How Many Principal Components?" NeuroImage vol. 9, 534-544 (1999).
  • U. Kjems, S.C. Strother, J. Anderson, I. Law, L.K. Hansen. "Enhancing the Multivariate Signal of 15-0 water PET Studies With a New Non-Linear Neuroanatomical Registration Algorithm". IEEE Transactions in Medical Imaging vol. 18 306-319, (1999)
  • N. Lange, S.C. Strother, J.R. Anderson, F.AA. Nielsen, A.P. Holmes, T. Kolenda, R. Savoy, L.K. Hansen. "Plurality and Resemblance in fMRI Data Analysis" NeuroImage, vol. 10 (3):282-303, (1999)



    Links to neurosites

  • NIMH Neuroinformatics home Human Brain Project
  • International Neuroimaging Consortium Rottenberg groups homepage.
  • 6'th Int. Conf on Functional Mapping of the Human Brain (San Antonio, June 2000).
  • Keith Worsley's homepage (statistical analysis).
  • Statistical Parametric Mapping (SPM) home
  • Roger Woods' AIR co-registration software
  • Ulrik Kjems' 3D non-linear Warp toolbox (UNIX)

    Return to homepage.