Principals of Digital Signal Recording How do we represent a continuously variable signal digitally? • Sampling – Sampling rate – number of measurements per unit time – Sampling depth or quantization – number of gradations by which the measurement can be recorded How do we represent a continuously variable signal digitally? • Sampling – What would be the advantage to higher sampling rates? How do we represent a continuously variable signal digitally? • Sampling – What would be the advantage to higher sampling rates? • Nyquist limit How do we represent a continuously variable signal digitally? • Sampling – What would be the advantage to higher sampling rates? • Nyquist limit • Aliasing – What would be the disadvantage? • Data size • Compute time How do we represent a continuously variable signal digitally? • Sampling – What would be the advantage to greater sampling depth? • Finer resolution – What would be the disadvantage? • Data size • Possibly compute time How do we represent a continuously variable signal digitally? • Sampling – A note about data size and compute time: • New data size = increase in quantization x number of samples x number of electrodes! Filters used in EEG What is a filter? What is a filter? • Filters let some “stuff” through and keep other “stuff” from getting through – What do we want to let through? – What do we want to filter out? What is a filter? • The goal of filtering is to improve the signal to noise ratio – Can the filter add signal? Different Kinds of Filters • • • • Low-Pass (High-Cut-Off) High-Pass (Low-Cut-Off) Band-Pass Notch • Each of these will have a certain “slope” How do Filters Work? • Notionally: – Transform to frequency domain – Mask some parts of the spectrum – Transform back to time domain Are There Any Drawbacks? • Yes • Filters necessarily distort data – Amplitude distortion – Latency distortion • Forward/backward/zero-phase Recommendations • Should you filter? – Yes, when necessary to reveal a real signal • Problem: how do you know it’s “real” – No, always look at the unfiltered data first • What filters should you use? – Depends on your situation (e.g. what EEG band are you interested in? Do you have 60Hz line noise?) – General rule: less aggressive filters are less distorting