|
|
| |
New! Look through the list and talk to Raj or Pradeep about a topic
that interests you.
A group can comprise of 1-3 persons enrolled in the class. Each group
will pick a particular "mini-research" topic pertinent to this course
and investigate the topic through literature search and/or analysis of
available BCI data. Each group will present their results to class on
May 31 and will submit a short printed report (10-15 pages including
figures and references) before the presentations.
Presentation details: You may email Raj your slides by 11am on the day of
the presentation and use his laptop for your presentation or bring your
own laptop if your presentation requires special software or uses demos.
In the latter case, you may want to print out transparencies as a
back-up in case the projector does not recognize your laptop.
ECOG Data
Electrocorticography (ECoG) is a technique for recording
signals directly on the surface of the brain, and is widely used
for monitoring brain activity in the course of medical treatment
for epilepsy. We have a large number of datasets where the
subject in question performed a variety of motor actions (or
imagined the coresponding action), and the challenge is
to build classifiers that can distinguish between different
motor actions. The following projects will look at various
techniques for improving the classification performance of ECoG
Data. We will provide you with a small number of datasets, and
your finished code can optionally be run on all the other
(unseen) datasets.
Spatial Filters for ECoG Data:
Classification techniques for EEG data
use spatial filters for integrating information from multiple
sensors and creating a lower-dimensional feature space that is
more tractable for classification. The project would examine one
of the following filters:
a) CSP—a supervised technique for selecting linear projections. b) PCA/ICA—unsupervised techniques that are useful in dimensionality reduction, c) CSSSP - a supervised technique that simultaneously optimizes the spatial and frequency filters used on the signals. d) an algorithm of your choice.
We will keep the choice of classifier fixed, and also supply you with some basic preprocessing code. References [3-8].
Spectral Filters for ECoG Data: We know that
motor activity or imagery is manifested in the spectrum of the
EEG data as changes in the power of
particular frequency bands—e.g., the so-called mu and beta
bands in EEG. Similar behavior is
observed in ECoG data as well. Here we will examine algorithmic
techniques for “fitting” a frequency band filter on
ECoG data in order to maximise the discriminability of the
resulting data. We will keep the choice of classifier fixed, and
also supply you with some basic preprocessing code. References
[3-6, 9].
EEG Data
BCI Competition: ( link to webpage)
pick your favorite dataset, and either implement one or two of
the submitted solutions, or compare the winning solution with
your favourite algorithm. Talk to us for choices
Local Field Potentials and EEG: The idea is to estimate local
field potentials directly from EEG,
and thus get a better understanding of the underlying neural
processes recorded on the surface of the scalp as EEG. These estimates could presumably also
be useful for BCI purposes. This is a
harder project—see refs [1,2].
Possible projects using our EEG recording system:
- Classification of motor imagery data: we will explore the
classical EEG-based BCI paradigm, that of classifying data
recorded during imagined movements of hands, feet, tongue.
See refs [10,11]
- Classification of P300 evoked
response: this is also a common BCI paradigm where the subject's
reaction to a visual stimulus is gauged and used for
communication. See refs [12,13]
- Image processing using cortical computing [14]. Here, we
use human brains as a computing substrate that can sort out
pictures of faces from pictures of inanimate objects, solely by
virtue of the reaction to presented visual stimuli.
- Neural correlates of motor learning: Psychophysical research
literature aims to build qualitative models of human
sensorimotor learning. Evidence of learning can also be observed
at the neural level. Can we also do this at the EEG level?
See refs [14-17].
Reading Projects
The idea in these projects would be to read a set of closely
related papers investigating a topic of interest to BCI. You will make a presentation to the
class explaining the new issue and summarising the papers, and,
based on the topic, present a written report consisting of a
synthesis, comparison, or possible future work and
improvements.
- EEG artifacts and
modeling/rejection. A significant problem in the analysis of EEG
data is its extreme sensitivity to artifacts arising from muscle
activity, eyeblinks and movement. What techniques do people use
to deal with these artifacts? Can they be removed from the EEG?
This can also be converted into data analysis project with our
EEG recording system. See refs
[18-21].
- Learning, longitudinal study, rehab issues, etc-- What does
it take to build an EEG-based BCI? Who can use one? How do
things change over time? See refs [22-25].
- Other sources of information: phase synchrony, presence of
an error potential-- in the vanilla BCI paradigms, you can
squeeze extra information out of the EEG data, and also by
observing user reactions to errors made by the BCI system [26-30].
- Neural correlates of motor learning: Psychophysical research
literature aims to build qualitative models of human
sensorimotor learning. Evidence of learning can also be observed
at the neural level. Can we also do this at the EEG level?
See refs [15-17].
- The BCI Taxonomy
project—summarize the results. The idea here is to
build a generic description, structure and terminology that
covers the variety of BCI systems being designed.
- Other paradigms for BCI: This
could be either a different interaction scheme, e.g.,
steady-state somatosensory evoked potentials, or a different
recording scheme--e.g., infrared, FMRI, MEG, etc.
References
- Non-Invasive Estimation of Local Field Potentials for Neuroprosthesis Control, R. Grave de Peralta Menendez, S. Gonzalez Andino, L. Perez, P.W. Ferrez, and J. del R. Millán, in “Cognitive Processing, Special Issue on Motor Planning in Humans and Neuroprosthesis Control” link
- Electrical neuroimaging based on biophysical constraints. Grave de peralta menendez R, Murray MM, et al. Neuroimage 21 (2004)
- A brain-computer interface using electrocorticographic signals in humans. Eric C Leuthardt, Gerwin Schalk, Jonathan R Wolpaw, Jeffrey G Ojemann and Daniel W Moran. J. Neural Eng. 1(2), 2004.
- Optimizing spatio-temporal filters for improving Brain-Computer Interfacing. Guido Dornhege, Benjamin Blankertz, Matthias Krauledat, Florian Losch, Gabriel Curio, Klaus-Robert Mueller, NIPS 05.
- Pregenzer, M.; Pfurtscheller, G.: Frequency Component Selection for an EEG-Based Brain Computer Interface (BCI). . – in: IEEE Trans Rehab Engg. (1999) S. 413 – 419
- Ramoser, H.; Müller-Gerking, J.; Pfurtscheller, G.: Optimal spatial filtering of single trial EEG during imagined hand movement. – in: IEEE transactions on rehabilitation engineering 8 (2000) , S. 441-446
- ICA For Dummies
- Jung T-P, Makeig S, Lee T-W, Mc Keown? MJ, Brown G, Bell AJ, and Sejnowski TJ, “Independent Component Analysis of Biomedical Signals,” (.pdf, 863k) Second International Workshop on Independent Component Analysis and Signal Separation, Helsinki, pp. 633-44, 2000. link
- Pfurtscheller, G.; Lopes da Silva, F.: Event-related EEG/MEG synchronization and desynchronization: Basic principles. – in: Journal of clinical neurophysiology 110 (1999)
- Guido Dornhege, Benjamin Blankertz, Gabriel Curio, and Klaus-Robert Müller. Boosting bit rates in non-invasive EEG single-trial classifications by feature combination and multi-class paradigms. IEEE Trans. Biomed. Eng., 51(6):993-1002, 2004.
- Schlögl, A.; Lee, F. Y.; Bischof, H.; Pfurtscheller, G.: Characterization of Four-Class Motor Imagery EEG Data for the BCI-Competition 2005. – in: Journal of neural engineering 2 (2005)
- Robust Classification of EEG Signal for Brain–Computer Interface. Thulasidas, M.; Guan, C.; Wu, J. TNSRE 14 (1) 2006.
- ERPs evoked by different matrix sizes: implications for a brain computer interface (BCI) system
Allison, B.Z.; Pineda, J.A. TNSRE 2003
- P. Sajda, A. Gerson and L.C. Parra (2003) High-throughput Image Search via Single-trial Event Detection in a Rapid Serial Visual Presentation Task, IEEE Conference on Neural Engineering Capri Island, Italy, March 20-22, 2003
- M.G. Philiastides and P. Sajda (2005) Temporal Characterization of the Neural Correlates of Perceptual Decision Making in the Human Brain, Cerebral Cortex.
- Gandolfo, F., Li, C.-S., Benda, B.J., Padoa-Schioppa, C. & Bizzi, E. (2000) Cortical correlates of learning in monkeys adapting to a new dynamical environment. Proc. Natl Acad. Sci. USA, 97, 2259-2263
- Emerging Patterns of Neuronal Responses in Supplementary and Primary Motor Areas during Sensorimotor Adaptation. R. Paz, C. Natan, T. Boraud, H. Bergman, and E. Vaadia (2005). J. Neurosci. 25: 10941-10951
- Delorme, A., Sejnowski, T., Makeig, S. (2005) Improved rejection of artifacts from EEG data using high-order statistics and independent component analysis. Submitted to neuroimage link
- Schlögl, A.; Anderer, P.; Roberts, S. J.; Pregenzer, M.; Pfurtscheller, G.: Artefact detection in sleep EEG by the use of Kalman filtering. – in: Medical & Biological Engineering & Computing, European Medical & Biological Engineering Conference EMBEC ‘99. (1999) S. 1648-1649
- Roberts, S. J.; Everson, J.; Rezek, I.; Anderer, P.; Schlögl, A.: Tracking ICA for EEG Eye Movement Artifact Removal. – in: Medical & Biological Engineering & Computing, European Medical & Biological Engineering Conference EMBEC ‘99. (1999)
- Jung T-P, Makeig S, Westerfield M, Townsend J, Courchesne E, and Sejnowski TJ. ” Removal of eye activity artifacts from visual event-related potentials in normal and clinical subjects (.pdf, 1.3Mb),” Clinical Neurophysiology 111(10), 1745-58, 2000.
- How many people are able to operate an EEG-based brain-computer interface (BCI)? Guger, C.; Edlinger, G.; Harkam, W.; Niedermayer, I.; Pfurtscheller, G. TNSRE 2003
- Learning to control brain rhythms: making a brain-computer interface possible
Pineda, J.A.; Silverman, D.S.; Vankov, A.; Hestenes, J. TNSRE 2003.
- EEG changes accompanying learned regulation of 12-Hz EEG activity. Delorme, A.; Makeig, S. TNSRE 2003.
- Evolution of the Mental States Operating a Brain-Computer Interface, J. Mouri no, S. Chiappa, R. Jané, and J. del R. Millán, in “Proceedings of the International Federation for Medical and Biological Engineering”, 2002
- Phase Synchrony Rate for the Recognition of Motor Imagery in Brain-Computer Interface. Le Song, Evian Gordon, Elly Gysel. NIPS 2005
- Brunner, C.; Graimann, B.; Huggins, J. E.; Levine, S. P.; Pfurtscheller, G.:
Phase relationships between different subdural electrode recordings in man. – in: Neuroscience letters 375 (2005) 2 , S. 69-74
- Response error correction-a demonstration of improved human-machine performance using real-time EEG monitoring. Parra, L.C.; Spence, C.D.; Gerson, A.D.; Sajda, P. TNSRE 2003
- Schalk, G.; Wolpaw, J.; MCFarland, J.; Pfurtscheller, G.: EEG-based communication: presence of an error potential. – in: Clinical neurophysiology (2000) , S. 2138-2144
- You are wrong! Automatic detection of errors from Brain Waves. P. W. Ferrez, J. del R. Millan. IJCAI 2005
|