Presentation on theme: "Presented By Qingwei Zhang Mu Li To understand the definition and classification of brain-computer interaction To explorer various non-invasive brain-"— Presentation transcript:
To understand the definition and classification of brain-computer interaction To explorer various non-invasive brain- computer interfaces
Introduction to Brain-Computer Interaction Non-invasive BCI EEG BCI by Qingwei Zhang fMRI BCI by Mu Li Summary & Conclusion
A communication system ; Enables the brain to send messages to the external world ; Without using traditional pathways as nerve or muscle.
Different types of BCI Animal BCI research Human BCI research Invasive BCIs Non-invasive BCIs EEGMRI
Has the least signal clarity Most safest Various technologies EEG: Electroencephalography fMRI: Functional magnetic resonance imaging
EEG: Why best choice of BCI Research ? Based on functional imaging Offer safer operation Excellent temporal resolution and usability
The usability depends on three aspects Running workload Effective bit rate portability
Problems Subject to artifacts from ocular or muscle movements Overlapping of electrical activity from different brain areas Sensitive to apparatus noise Low-pass filtering of the signal by the skull and skin EEG: loss in spatial resolution
Solution The signal need to be transformed to a new space that more suitable for classification. We need to reconstruct and deblur the signal from different sources.
Methods Data Feature Extraction Features Selection Classification Scalp Electrodes Data i.Data: Signals recorded with a bandpass filter. ii.Feature Extraction: Based on change in frequency power due to movement. iii.Features Selection: Applied a Model-independent ranking method. iv.Classification: Data divided into a training set and a test set. v.Scalp Electrodes Data: Features extracted based on frequency band powers.
What is fMRI BCI How it works Applications Based on reference paper 
What is fMRI BCI? fMRI - Functional magnetic resonance imaging a noninvasive BCI technique that measures the task-induced blood oxygen level-dependent (BOLD) changes correlating with neuronal activity in the brain. 
Signal Acquisition To acquire images slice by slice and at the same time reconstruct the whole brain image Influencing factors Static magnetic field strength Spatial resolution Temporal resolution Echo time Magnetic field inhomogeneities
Signal Preprocessing To correct noises/artifacts Head motion correction Retrospective methods Prospective methods Physiological noise correction Noise caused by breath Noise caused by pulse
Signal Analysis To find out how a particular perceptual or cognitive state is encoded (classification) Univariate analysis Exam brain activity from thousands of locations and analyze them separately Multivariate analysis Exam spatial pattern of brain activity (multiple locations)
Signal Feedback To compute and present feedbacks Feedback identification Prior to experiment, need to identify region of interest. (ROI) Feedback computation To compute suitable feedback Feedback presentation Modalities – verbal, visual, auditory, olfactory, tactile…
fMRI BCI Applications Clinical rehabilitation and treatment To train subjects to modulate a brain region, e.g. perception of chronic pain Neuroscientific research To study neuroplasticity, emotional processing, pain and language processing Decoding brain state e.g. “brain reading”
Both are non-invasive BCIs. Captures two different brain signals: brain wave by EEG and BOLD by fMRI. EEG is much more portable, while fMRI requires big MRI scanning machine, several workstations for the intensive computing and a local network. fMRI has much better spatial resolution and better bit rate. Intuitively, EEG BCI is much cheaper than fMRI BCI.
Non-invasive BCI may be the most attractive BCI research direction in the future due to its safeness. There are various technologies supporting non-invasive BCI. Researchers keep improving them and look for new ones. EEG and fMRI are two different non-invasive BCI techniques, each of them has its own pros and cons. They could be used together to provide better BCI experience.
1) R. Sitaram, N. Weiskopf, A. Caria, R. Veit, M. Erb, N. Birbaumer, "fMRI Brain- Computer Interfaces," In Signal Processing Magazine, vol. 25, no. 1, pp. 95- 106, 2008 2) A. Eklund, M. Andersson, H. Ohlsson, A. Ynnerman, H. Knutsson, "A Brain Computer Interface for Communication Using Real-Time fMRI," In Pattern Recognition (ICPR), 2010 20th International Conference on, pp. 3665-3669 3) Wikipedia, Brain-computer interface, http://en.wikipedia.org/wiki/Brain%E2%80%93computer_interface. Accessed on Oct. 31, 2010 4) Wikipedia, Functional magnetic resonance imaging, http://en.wikipedia.org/wiki/Functional_magnetic_resonance_imaging. Accessed on Oct. 31, 2010 5) Quentin Noirhomme, etc. Single-Trial EEG source Reconstruction for Brain- Computer Interface, IEEE Transactions on Biomedical Engineering, 55(5), pp: 1592 - 1601, 2008. 6) Wolters, C.H.,etc. Geometry-Adapted Hexahedral Meshes Improve Accuracy of Finite-Element-Method-Based EEG Source Analysis, IEEE Transactions on Biomedical Engineering, 54(8), pp: 1446 - 1453, 2007.