Recording EEG signals

Course: CLNP5002 | Diagnostic Electroencephalography.
Speaker: Andrew Bleasel, Westmead Hospital and Children’s Hospital at Westmead.
Origin: AOCCN - EEG course, 2007, hosted by The University of Sydney.
Core Topic: Recording EEG Signals.

Primary Functions:
- Differential discrimination.
- Amplification of EEG signals.
Differential Amplifiers:
- These devices measure the difference in voltage (potential) between two points on the scalp.
- They do not measure absolute values.
- It is impossible to measure the potential between ground and the scalp directly using this method.
- Bioelectric signals require differential or balanced amplifiers to be accurately captured.
- Purpose: To remove electrical noise from very small signals.
- Metaphorical Example: Measuring the vertical distance between two aircraft provides information about their relative distance but tells you nothing about their actual altitude relative to the ground.

Definition: Because differential amplifiers only record the differences between inputs, any potential that affects the voltage equally at both inputs (Input 1 and Input 2) will be "rejected" rather than "seen."
Mechanism: Potentials or activity that are in-phase and have the same amplitude at both inputs will result in a flat line as the output.
Applications of CMR:
- It is useful for recording localized brain activity (EEG signals at the scalp).
- It eliminates certain artifacts, such as $50\,Hz$ AC interference and sweat potentials.
Limitations and Risks of CMR:
- It can be misleading when activity is widespread; by rejecting the common signal, it can make widespread activity appear falsely localized.
- It is often observed in bipolar montages.
- CMR may fail if electrodes have unequal impedance or if the patient is not properly grounded.

Sensitivity:
- Defined as the ratio of input voltage to the signal deflection produced.
- Expressed in units of $\mu V/mm$ .
- Calculation Example: Comparing $10\,\mu V/mm$ vs. $5\,\mu V/mm$ .
- Note: A higher numerical sensitivity value (e.g., $10\,\mu V/mm$ ) will produce a smaller output signal on the screen compared to a lower numerical value.
Gain:
- Defined as the ratio of signal voltage at the input to the signal voltage at the output.
- Calculation Example: If the input is $10\,\mu V$ and the output is $10\,V$ , the gain is $1,000,000$ .
- Note: A higher gain produces a bigger output signal.

Electrodes (Inputs): The physical points on the scalp used to pick up electrical activity (e.g., $Fp2, F8, T4, T6, O2, Fp1, F7, T3, T5, O1$ ).
Channel (Output): The resulting signal line displayed after the difference between two electrode inputs is processed (e.g., $Fp2-F8$ is one channel).
Montage: A specific arrangement or "chain" of electrode pairs (channels) being viewed simultaneously (e.g., an AP Hemisphere montage).
Bipolar Montage Example:
- Channel 1: $Fp2-F8$
- Channel 2: $F8-T4$
- Channel 3: $T4-T6$
- Channel 4: $T6-02$

Definition: A device used to attenuate unwanted frequencies.
Functionality Constraints:
- Filters are not intended to make the EEG look "tidy" or "neat."
- They should not be used to compensate for or conceal poor recording techniques.
- Using filters always results in some degree of data loss.
Significant Frequency Range: The most important frequencies for EEG are between $1-70\,Hz$ .
Filter Types:
- Low Frequency Filter (LFF) / High Pass: Allows higher frequencies to pass while attenuating lower frequencies. Used to attenuate slow artifacts like slow rolling eye movements.
- High Frequency Filter (HFF) / Low Pass: Allows lower frequencies to pass while attenuating higher frequencies. Used to attenuate muscle activity (EMG).
- Notch Filter / Band Reject: Specifically used to attenuate AC interference ( $50\,Hz$ ).
Filter Characteristics:
- The frequency setting of a filter is the point where the signal is attenuated by approximately $30\%$ , not blocked completely.
- Frequencies beyond the setting are attenuated to a progressively greater degree.
- Roll-off: Refers to the specific shape of the frequency-attenuation curve.
Problems with Filters:
- Can distort waveforms.
- May cause a phase shift.
- Risk of concealing actual pathological activity.

Definition: Describes the effect of low frequency filters upon a square wave pulse.
Measurement: The time taken for the amplitude to be attenuated by $63\%$ , or to fall to $37\%$ (approximately "one third") of the original amplitude.
Formula: $TC = \frac{1}{2 \pi f}$
Specific Conversions:
- $TC = 1.0\,second$ corresponds to $LFF = 0.16\,Hz$
- $TC = 0.3\,s$ corresponds to $LFF = 0.5\,Hz$
- $TC = 0.03\,s$ corresponds to $LFF = 5.3\,Hz$

Definition: The transformation of a continuous analog EEG signal into a series of discontinuous data points.
Accuracy Factors:
- Sampling Rate: Measured in $Hz$ or samples per second.
- Number of Amplitude Levels: Determined by the Bit number.
- Input Voltage Range: The range over which the ADC can operate.
Nyquist Frequency:
- The sampling rate must be at least twice the fastest frequency present in the waveform to avoid errors.
- It is considered best practice to "over-sample."
Aliasing: A phenomenon that occurs when the sampling rate is too low, causing high-frequency signals to appear as lower-frequency distortions.

Standard Display Examples:
- Sensitivity: $7.50\,\mu V/mm$ or $10\,\mu V/mm$ .
- HF (High Frequency Filter): $70\,Hz$ or $35\,Hz$ .
- LF (Low Frequency Filter): $1.6\,Hz$ or $0.3\,Hz$ .
- CAL (Calibration Pulse): $50\,\mu V$ .
- Time Scale: $13.2\,mm/s$ or $26.5\,mm/s$ .
- Time Interval: $1.0\,s$ .

EEG is a complex and delicate biological signal.
Fundamental mastery requires understanding:
- Generators of the EEG signal.
- Limitations of the recording process.
- EEG amplifiers and the principles of common mode rejection.
- The specific roles and impacts of filters.
Digital EEG provides significant flexibility for post-processing compared to traditional analog methods.