BCI2000 - will be used to create framework of the BCI
MATLAB - more of an end step, used to analyze the signals (in my case CCA [read the articles for more information about this])
C# - will be used to modify the SSVEP signal; the SSVEP signal had to be created seperately by Nicholas, so to modify the size, color, shape of the SSVEP signal I will need to learn how to program CStar (I hope to start this on Wednesday)

General Schedule - I can arrive at the lab generally whenever I want (I'll probably stay from 11:00 - 6:00 pm or so)

Monday/Tuesday - Read SSVEP articles and read book (the one listed below). Create the general framework of what I am modifying and other specifics
Tuesday - Friday - Continue reading books and articles; learn how to program/use CStar; will also use BCI2000 to set everything up (this may be already done by the pHd students)
Weekend - maybe continue reading and such; more freedom

Monday - Wednesday - Not sure, will probably be testing on human subjects (the volunteers are typically people in the lab, probably myself included)
Thursday - Friday - Use MATLAB to analyze signals; ensure that everything is complete; IF you do not finish by this time, make sure you can analyze the signals at home or obtain the information through the pHd students in some way
ALSO, ask about possibly having this article published? << Need more details regarding this <<



Notes-Statistical Signal Processing For Neuroscience and Neurotechnology

Spike detection and estimation - Different types of signals require different types of analysis (in my case I am using EEG, probably a power analysis of the signals using MATLAB and BCI2000)

Transient Signal: a signal known to be nonzero only for a subset of samples L within an observation interval of length N (basically, a signal created by a stimuli that is different from calm state [no moving or thinking] over a set time interval)
HOW TO DETECT? 3 Parameters: Arrival time (K sub0), the support/time window (L), and the waveform shape (s)
<Arrival time> MOST IMPORTANT *** Carries the information
<L and s> mainly used to sort the spikes when multiple neurons are recorded simultaneously (AKA classification)

Test statistic is used to find the liklihood of a certain event occuring.
Ha: y=s+n
H0: y=n
y=observations
s=L-dimensional spike to be detected
n=additive noise term

Contaminating Noise (n), crucial role in assessing how far apart the two models are

Thermal and electrical noise due to electronic cuircitry (includes amplifiers, filters, and ADC)
	Noise can be modeled as a stationary, zero-mean, Gaussian white noise process (AKA normal 		curve/Gaussian curve); only unknown parameter is the noise variance
Biological noise
	Already have notes on this; still need to figure out how to filter this information




Brain-Computer Interfaces Based on the Steady-State Visual-Evoked Response
Note: This article was published in 2000 and was the first article to use SSVEP BCI. Although the idea of SSVEP has been around for a while, this article was the first to make a BCI out of it.

The researchers did 2 tests:
1. Operators are trained to exert voluntary control over the strength of the SSVEP (it is possible to train users to manipulate to amplitude of the signal)
2. Multiple SSVEP's are used for control (requires little to no training because the system capitalizes on naturally occurring responses)

Generally results were good; it required a lot of training however (in my case I will not need training)


Frequency Recognition Based on Canonical Correlation Analysis for SSVEP-Based BCIs
CCA is applied to analyze the frequency components of SSVEP in EEG: extract a narrowband frequency component of SSVEP in EEG
Recognition results better than FFT

Frequency-coded selection by SSVEP - an SSVEP signal will be created in the brain with a similar fundamental frequency as visually seen on the moniter, CCA extracts that information and registers it as a command
PSDA=Power Spectral Desnity based Analysis

CCA: helps improve SNR, uses channel covariance information; it is a multivariable statistcal method used when there are two sets of data which may have some underlying correlation,
Method used in article: focuses on relationship between the stimulus signals and EEG signals from multiple channels in a local area






SSVEP Experimentation Information:

References:
A Survey of Stimulation Methods Used in SSVEP-Based BCIs 2010
-"Further research on the influence of color on the SSVEP is necessary"
-"Power of the SSVEP resonse is affected by both frequency and color of the stimuli [11]"

Brain-Computer Interfaces Based on the Steady-State Visual-Evoked Response 2000

Frequency Recognition Based on Canonical Correlation Analysis for SSVEP-Based BCIs 2007

An Online Multi-Channel SSVEP-based Brain-Computer Interface Using a Canonical Correlation Analysis Method 2009
-The framework was based off of this article

<ADD OTHER ARTICLES>

Testing: Color and Size of the stimuli (maybe focus on color?)
Why? To improve classification rates and stability of the SSVEP signal in EEG

Specs of the materials used?
EEG:
Amplifier:
ADC:
Filter(s):
Programs: BCI2000 for paradigm and layout, CStar to manipulate the SSVEP signal, MATLAB for analysis and conclusion sections

SSVEP Frequencies: 
1-12 Hz = low frequency band
13-30 Hz = medium frequency band
31-60 Hz = high frequency band
Choose from 6-30 Hz (in case computer specs or other issues), pattern reversal stimuli (AKA checkerboard)
Low/medium frequencies cause more visual fatigue, can provoke epileptic seizures, and covers the alpha band which can cause a considerable amount of false positives. Should I use a high frequency band? 
<Perhaps choose 8 Hz?> <Perhaps choose 31 Hz?>
If I have time I will experiment with both frequencies

Introduction to Experiment:
No User Training Required
The brightness of the room and location will all be the same
Subjects will all be treated in a similar manner
Using Canonical Cortical Analysis (CCA) to classify signals
Users will be monitored to avoid eye blinking and muscle movement artifacts
Using 5 healthy human subjects (probably people from the lab)
ITR, classification accuracy (%), (I will need to find a way to analyze the signal and generate a conclusion from it) <ADD OTHER RESULTS> will be used as results

SSVEP Size: Checkerboard and Solid Colors
10 cm length (100 cm^2 area)
SSVEP will be placed in the center of the screen

Color Combinations that Will be Tested with Each Different Size:
Background color is gray
Black <=> White
Red <=> White
Blue <=> White
Green <=> White

Session Information
Each session lasts 30 minutes.
There will be one session for each of the stimuli type (checkerboard and solid colors) and color combinations (Checkerboard; White <=> Black, Solid Color; White <=> Black, Checkerboard; Red <=> White, Solid Color; Red <=> White, etc)

Time window and trial organization: (needs to specify how long the object will flash, when it dissapears, how long the user break is, etc)
Initial State: User will sit in a comfortable chair 1.5 meters away from the TV screen where the SSVEP will be displayed.
Once the user is ready the BCI will start. 

Loop Starts Here:
1. There will be a 14 second pause. The screen will be blank at this time.

2. The stimuli will appear for 5 seconds before dissapearing.
3. There will be a black square for 10 seconds.
4. Repeat steps 2 and 3 for all checkerboard/solid colors until the end of the experiment.

User Interface
Single SSVEP in the middle of the screen
-This SSVEP will flash or dissapear in accordance with the time window
The users job will be to look at the SSVEP