Transcript presentation_nada_si..

by Omar Nada & Sina Firouzi
Introduction
 What is it
 A communication channel between brain and electronic device
 Computer to brain/Brain to computer

Why we need it
 Medical purposes

Repairing eyesight, hearing, movement of body parts
 Entertainment and multimedia communications

Toys, video games, activity in virtual reality environments, controlling devices with thought,
synthetic telepathy
 Military

Mood control , commanding and telepresence

How does it work
 Algorithms are used to translate brain activity into control signals
 Brain can handle signals generated by electronic devices
Overview

Source: wingsforlife.com
How does it work
 Brain’s electrical activity produced by firing of electrically charged
neurons is observed by sensors
 Invasive sensors

Electrodes are implanted directly into gray matter
 High quality of signals, risk of scar-tissue
 Partially invasive
 Electrodes are implanted inside skull but not into gray matter
 Lower quality, less risk of scar-tissue
 Non-invasive
 Signals are observed from outside the skull
 Low quality as skull dampens signal, no surgery, no scar-tissue, safest method
 Electroencephalography (EEG) by observing the wave of ions released by neurons
 Magnetoencephalography (MEG) by observing magnetic fields produced in brain
 Functional magnetic resonance imaging (FMRI)
 This information is translated using algorithms and used by electronic
devices and vice versa
Using Invasive Sensors
BCI Projects
 Assist Arm Robot
 Carleton University
 BCI + Assist
 Berlin Brain-Computer Interface
 Health Care
1) Assist ARM Robot
 Early phase One degree of freedom Assist Arm
 Uses nerves and force sensor as input
 Assist in a desired motion ( for recovery)
MEG
(using electrodes)
Biceps & triceps
impedance
control schema
Motion
(force sensor)
Up & Down directions
Projected
Motion
Assisted Movement
force
sensor
electrodes
Initial movement
Assisted movement
Challenges
 Same group muscles can control different joints
 Body fat, muscle mass, muscle fatigue affect
measurements
 Different people give different values ( like PWM)
 Lack of volunteers!!!! (especially for invasive methods)
 Guessing the user Intensions!
Work Arounds / Solutions
 Session Calibration
 Using min and max values of voltages
 Muscle Group Calibration
 Run the above technique for all the group muscles used
for readings
 THEN: Work relatively
 Use the session and group muscle boundaries to predict
user intention
2) Berlin BCI
 The Berlin Brain-Computer Interface: EEG-based
communication without subject training
Benjamin Blankertz, Guido Dornhege, Matthias Krauledat, Klaus-Robert Müller, Volker Kunzmann, Florian
Losch, Gabriel Curio
 Non-invasive
 Key features
 Use of well-established motor competences as control
paradigms
 High-dimensional features from 128-channel EEG
 Advanced machine learning techniques
2) Berlin BCI
 Establishing a BCI system based on motor imagery that
works without subject training
 ‘Let the machine learn’
 System automatically adapts to the specific brain signals of each user
by using advanced techniques of machine learning and signal
processing
 It is possible to transfer the results obtained with regard to
movement intentions in healthy subjects to phantom
movements in patients with traumatic amputations.
 High information transfer rates can be obtained from
single-trial classification of fast-paced motor commands
3) Health Care
 Health care example
 Repairing damaged hearing

Sounds are received by an external device and signals are sent
to brain
 Repairing damaged eyesight

A camera sends signals to brain
 Helping people with spine injuries and paralyzed limbs
by electrically stimulating muscles
 Moving paralyzed body parts with help of robotic parts
Brain
Computer
Moving
part
3) Health Care
 Replacing damaged or lost body parts

Mechanical hands, fingers.
 Helping people with severe paralysis to communicate
with outside world using a computer.
 Restore speech

Patient concentrates on a letter and computer receives and
pronounces it
Feasible Future
 What is in research
 Are people able to willingly fire specific neurons in realtime?
 Images seen by human eyes have been recorded in black
and white. Recording color images is in research
 Recording dreams and thoughts
 What is coming out soon
 Affordable non invasive sensors
 Calibration using heart rates ( more accurate results)
Omar’s view of the future
 BCI


Better control algorithm to decode the brain activities (cheap
non invasive) coming to reality
Check out TED video emotiv by Tan Le
 Application
 ‘HandsFree’ Driving
 Thinking Pattern Authentication
Sina’s view of the future
 Virtual reality
 Being able to interact with others in a virtual 3D environment
without using muscles or mouse
 Using electronic devices without touch or any muscle
movement
 Well functioning moving body parts
 Mood control
 Sending signals to your brain can improve your mood
Conclusion
 BCI = Brain + Computer + Communication Channel
 BCI Applications
 Carleton Assist ARM
 Berlin BCI
 Health applications
 How we view the future from the BCI lens