Biopotential Data – Describe Placement Info
ECG – Full and Restricted Frequency Bands, Non-linear Phase, Notch Effects on Waveshape
EDA – Attachment points, Bipolar, Unipolar
EEG – Frequency Banding, Number of Channels
EGG – Recording Close to DC
EMG – Effects of HPF and LPF
EOG – Vertical, Horizontal and Circular Eye Movement
Bioimpedance – Impedance Cardiography
Transducer Data – Describe Attachment Location and Methodology
Stimulus/Response/Averaging – Describe Methods
MRI and fMRI
Methods for Retrieving Noise-impacted Data – Noise Periodic or Random
Highly-processed Results (BPM, Median Frequency, HRV, PSD, etc.)
Biopotential Data – Describe Placement Info
Micro-electrode measurements are biopotential recordings from very small electrodes. Typically, these measurements are performed when recording the electrical signal from a small group of cells to as small as a portion of a single cell’s membrane. Recording from a small group of cells can be managed with a wire electrode. The wire electrode would usually have an insulating layer, such as epoxy, over its entire length. Small portions of the epoxy are then removed to expose the conductive wire underneath. When the wire is inserted into the preparation tissue volume, just the conductive portion of the wire is exposed to the cellular-generated electrical signal.
Micro-electrodes can also be glass electrodes that are filled with saline. The glass electrodes are made on a puller which heats a glass tube as it is subjected to strain in the direction of the tube. As the tube heats, it stretches and gets thinner. A glass tip can be constructed, in this fashion, that can be considerably smaller than a single cell. The tube is filled with saline and a conducting wire or Ag/AgCl pellet (with embedded wire) is inserted into the saline solution, to connect the electrode sensing tip to a wire attached to the amplifier input.
The smallest micro-electrodes, because of their small contact area, can have series impedances approaching 1Gohm. This situation poses special requirements for the recording system. In particular, shielding the measurement area from external interfering sources is important. This shield is often referred to as a “Faraday cage”. The cage will intercept any stray displacement current in the room and this current will flow around the surface of the cage. If the cage has effectively zero impedance then no differential voltages will develop, between different points on the cage, so no induced displacement currents will occur inside the volume of the cage.
For a Faraday cage (shield) to operate well, all signal conductors crossing the boundary of the shield should be filter-referenced to the shield. The nature of this referencing depends on the characteristics of the signals being recorded inside the shielded volume. Generally, all signal conductors crossing the boundary of the shield are lowpass filtered with respect to the shield. For a series L and shunt C type of lowpass filter, this means that the C shunts to the cage. In this circumstance, at higher frequency, the shunt C will keep pass-through conductors at same potential as the shield, thus removing any differential voltages that could induce displacement currents inside the shield. If conductors are directly (galvanically) attached to the Faraday shield, then no filter referencing is needed. This is typically the situation for all power lines passing through the shield, assuming the associated ground line is directly attached to the shield. Given that very low impedance exists between supply voltage lines and ground, the supply lines become effectively attached to shield, by virtue of the associated ground’s attachment to shield. Even though the supply lines are at a different voltage then ground, that voltage difference is constant, so those lines will not source a mains induced displacement current inside the shield.
If conductors cross the shield boundary there is always potential to induce differential voltages, between the pass-through conductors and the shield, depending on the nature of the filtering referencing conductors to the shield. For lowpass filters, differential voltages could develop at lower frequencies, versus higher ones and the opposite is true for highpass filters.
Displacement currents flow through the volume of any room that has mains wiring. These currents are sourced from “hot” mains wiring and find pathways to flow to neutral or ground. Any conductive object in the room will intercept these currents, such as the Faraday shield. Accordingly, there may be an impressed mains voltage on the shield. This voltage is called “common-mode”. If the shield is “floating”, in that it is not tied galvanically to mains ground, then a readily observable voltage, referenced to mains ground, will be present on the shield. Alternatively, if the shield is tied to mains ground, then no common-mode voltage will be observable.
Measurement quality-wise, it’s typically helpful to galvanically-attach the Faraday shield to amplifier reference ground and mains ground. This procedure will accomplish two objectives.
1. All differential signals generated, inside the shielded volume, will be common-mode referenced to the shield.
2. The shield’s EMF will be effectively reduced to 0 volts, with respect to mains ground. Displacement currents will be sent directly to mains ground and will not flow through amplifier reference ground.
Two basic types of Microelectrode measurements
Voltage Measurement with Micro-electrodes
The amplifier Vin+ input is usually attached to the preparation and the amplifier Vin- input (differential reference) is placed in the saline holding the preparation or elsewhere on the preparation. Typically, amplifier inputs require a bias current source and so amplifier reference ground must also be attached to the preparation. Amplifier reference ground should also connect to the shield so the amplifier differential inputs are common-mode referenced to the shield. Connecting the amplifier reference ground to shield improves common mode rejection ratio by increasing the common mode input impedance. As long as all subsequent measurement systems are referenced to the same potential as the, shield connected, amplifier reference ground then there will be no opportunity for the shield common-mode voltage (if any) to appear in the measurement recording.
Micro-electrode measurements can be either singled-ended or differential. Single-ended signal measurements reference the input signal (Vin+) to amplifier reference ground. Differential measurements reference one input voltage (Vin+) to another (Vin-). Micro-electrode single-ended measurements are best performed with a Faraday shield, or other configuration, that prevents common-mode displacement currents from reaching the measurement preparation or electrode. Differential measurements are more accommodating to a recording within a less-than-ideal Faraday shield.
Micro-electrode voltage amplifiers have a number of unique qualities:
1. extremely high input impedance
2. extremely low input bias and offset current
3. low front-end gain
4. large-range offset control
5. wide bandwidth
6. measurement to DC
High impedance and low bias current inputs are required because the micro-electrodes connected to the amplifier have very high source impedance. Amplifier bias currents are typically sourced through the micro-electrode, so an offset voltage develops across the micro-electrode tip and point of contact with the preparation. If the bias current is too high, the offset voltage created can exceed the amplifier input voltage range.
Typical cellular-generated potentials are on the order of 100mV and this measured voltage is added to all the other voltage sources in the measurement loop. Given the presence of amplifier offset voltage, bias current, probe tip resistance and half-cell potential of the electrode, it’s generally required that the amplifier tolerate a range of +/- 4 vdc. Accordingly, a micro-electrode amplifier usually has a unity front-end gain (Av = 1). Because of the wide input voltage range required, and the need to be able to discern small voltage variations residing at a possibly high voltage offset, offset nulling control is typically necessary. This control is added after the initial amplification stage at the point of a summing junction.
Single-ended Micro-electrode Voltage Measurements
A key aspect to performing single-ended micro-electrode measurements is the successful redirection of stray displacement currents in the measurement room. Any laboratory room is circled with mains high voltage (hot) wiring in the walls and in all mains-powered lab equipment. Displacements currents flow in these laboratory environments because of the physical phenomena of capacitance, defined as (e*A/D), where “e” is permittivity of free space, “A” is the common cross-sectional area between conductive object and mains hot wiring and “D” is the common distance between conductive object and mains hot wiring. In a typical laboratory, a few pF of coupling capacitance is typically present between mains hot wiring and a conductive object.
When performing very high impedance micro-electrode measurements, the probe impedances might be on the order of 100Mohms or more. The probe impedance is in parallel with the coupling impedance of the preparation to mains ground.
Assuming infinite probe impedance, 5pF coupling capacitance between preparation to mains hot wiring, 120VAC @ 60 Hz hot wiring and 0.5Mohm resistance of probe tip preparation location to mains ground, then the voltage that could be impressed on the probe is calculated as follows:
Reactance of coupling capacitance (mains hot wiring to preparation) at 60 Hz (Xch) is: 1/(2*Pi*60*5E-12) = 530Mohms
Vp = Vs (sRC/(1+SRC))
Vp = 120vac * j(0.5*(1/530))/(1+j(0.5*(1/530)) = 0.113vac
Clearly, 0.113vac impressed on the preparation is unacceptable, considering that the preparation voltage requires observable nuance on the order of 0.001v (1 mv). To resolve this issue, two strategies are helpful:
1. Reduce the coupling capacitance of mains hot wiring to the probe point of measurement.
2. Reduce the electromotive force (EMF) driving the coupling capacitance between mains hot wiring and probe point of measurement.
The preparation should be placed in a saline bath and the bath should be connected to amplifier reference /mains ground. Preparation, probe and amplifier should be placed on a large conductive plate with at least a partial conductive shield to cover preparation and probe. Both plate and shield are galvanically tied to amplifier reference and mains ground. In this circumstance, both of the above strategies are implemented. The coupling capacitance of mains hot wiring to probe point of measurement is greatly reduced, due to interception of displacement currents by shielding. Simultaneously, the shield effectively establishes a zero interfering EMF, with respect to amplifier reference ground, to further suppress the flow of displacement currents.
Assuming capacitance (from probe to mains hot wiring) can be reduced to 0.001pF and resistance of probe tip preparation location to amplifier mains ground is maintained at 0.5 Mohms, then the calculation becomes:
Reactance of 0.001pF coupling capacitor (hot mains wiring to probe) – 1/(2*Pi*60*1E-15) = 2,600 Gohms
Vp = Vs (sRC/(1+SRC))
Vp = 120VAC * j(0.5*(1/2,600,000))/(1+j(0.5*(1/2,600,000)) = 120vac * 1.92E-7 = 23uvac
In this situation, the impinging 120vac source, from room hot mains wiring, has been reduced to 23uvac at the location of the point of probe tip preparation. The plate and partial preparation shields should be large (x10) relative to the preparation and amplifier. In this case, the displacement currents in the room are first drawn to the plate and the partial shields. If the plate is amplifier reference/mains grounded, then that impinging current is directed to the amplifier reference/mains ground.
The plate and partial shields establish one plate of a large capacitor which is coupled to the mains hot power lines via a small coupling capacitance (about 5pF). This coupling capacitance is just a function of (e*A/d), where “e” is permittivity of free space, “A” is common cross sectional area between plate and room hot wiring and “d” is the average distance between plate and room hot wiring. Therefore, if the amplifier and preparation are placed on this plate, and if the amplifier is grounded to the plate, then the amplifier and the preparation bulk simply track the plate voltage. If the plate is also galvanically connected to main ground, then the plate EMF, with respect to mains ground, will be 0 volts.
At a mains frequency 60Hz, the wavelength is 5000km. Any shield opening, around the preparation, probe and amplifier will not be big enough to allow the 60Hz electromagnetic field to pass. The EMF will be nearly entirely trapped by the plate and partial shield.
Differential Micro-electrode Voltage Measurements
Micro-electrode recordings will measure the voltage difference between Vin+ and Vin-, where Vin+ attaches to the preparation and Vin- attaches to the preparation’s differential reference. Both preparation and shield may have a large, identical, mains-induced voltage present on their conductive parts. The micro-electrode amplifier attempts to remove these identical voltages, called “common-mode”, and just amplify the difference between the Vin+ and Vin- inputs, attached to the preparation and preparation differential reference, respectively. By attaching the amplifier ground to the Faraday shield, the common mode rejection ratio of the amplifier is greatly improved.
When recording from differential micro-electrodes, it’s important to balance the source impedances directed to the input of the amplifier. The amplifier has input resistance and input capacitance, referenced to ground. This input resistance and capacitance are parallel connected to create a complex input impedance (Zin). The source impedance (Zs) creates a voltage divider against Zin, in accordance to the formula:
Voltage Divider Ratio: Zs/(Zs+Zin)
If the source impedances are unbalanced, namely when the Zs connected to Vin+ is different than the Zs connected to Vin-, then the voltage dividers will be different between the Vin+ and Vin- inputs. This voltage difference will degrade the common mode rejection ratio (CMRR) of the amplifier. In the case of unequal Voltage Divider Ratios, from Vin+ to Vin-, any common mode signal present on the preparation will manifest itself as a voltage difference between Vin+ and Vin-. The common mode signal becomes a differential signal (noise) that superimposes with the desired signal from the preparation. Best amplifier performance is obtained when Zin is many orders of magnitude greater than Zs and when Zs (connected to Vin+) and Zs (connected to Vin-) are similarly-sized.
When recording from micro-electrodes, it’s typically most helpful to perform the measurement inside of a Faraday cage. A special type of micro-electrode amplifier, called a “Headstage Amplifier”, is used to measured the signal from the preparation via a wire or glass micro-electrode. This signal is attached to the positive input of the amplifier (Vin+). This signal will be referenced to the voltage at the attachment point of the amplifier’s negative input (Vin-). The amplifier’s output will be directed through the Faraday shield, via a pass-through lowpass filter.
The ideal micro-electrode recording setup is:
1. Headstage Amplifier inside of Faraday shield
2. Amplifier ground attached to preparation – typically to preparation bath via wire or pellet electrode
3. Vin+ to electrode attached to preparation – typically via microelectrode
4. Vin- to electrode attached to preparation – typically via microelectrode
Estimate Zs of the Vin+ electrode, when attached to the preparation. The connection from Faraday shield to Vin- should incorporate a similar value of Zs, as the Zs from the preparation to Vin+.
If the Headstage Amplifier is single-ended (Vin+ input only), then the amplifier measures the difference between the Vin+ input and amplifier ground. In this situation, a very high performing Faraday shield is required because the measurement will not tolerate the presence of common-mode noise on the preparation. The amplifier input attaches to a micro-electrode on the preparation and amplifier ground attaches to the preparation’s saline bath in order to provide a bias current path for the amplifier input.
A differential Headstage Amplifier will behave like a single-ended Headstage Amplifier if Vin- is connected to the same point as amplifier ground on the preparation. Because of the possibility of voltage drop, across a ground lead and electrode connection, the Vin- input should be connected to the same point on the preparation as the amplifier ground, assuming single-ended measurements with respect to ground are required.
Frequency Response and Bandwidth
Microelectrode probe series impedance, stray capacitances and amplifier input capacitance can establish a lowpass filter that may greatly interfere with signal measurement. Probe series impedance is typically highly resistive and is largely determined by the contact area of the probe as it makes contact with the preparation. The smaller the probe contact area, the higher the probe series resistance. Stray capacitances are determined by the physical distance the probe is from surrounding shields and other conducting surfaces. The amplifier input capacitance is typically specified on the amplifier data sheet, and is usually 1 pF or less.
Assuming a microelectrode probe series resistance of 100Mohms, stray capacitances of 0.5 pF and an amplifier input capacitance of 0.1 pF, then the frequency response of the signal measurement pathway is calculated as:
Frequency Response Bandwidth = 1/(2*Pi*R*C) = 1/(2*Pi*100E6*(0.5+0.1)E-12) = 2,650 Hz
Given that cellular-generated electrical signals can have frequency components in excess of 100kHz, this restriction can be quite profound. Probe series resistances are fixed, and are not able to be reduced because probe contact areas are required to be small to measure signals in the region of interest. Accordingly, to increase frequency response, stray capacitance and amplifier input capacitance must be reduced. Reduction of these capacitance can be accomplished through a couple of means:
1. Increase the physical distance between microelectrode probe conductor and surrounding shield.
2. Compensate for stray and amplifier input capacitance using electronic methods.
ECG – Full and Restricted Frequency Bands, Non-linear Phase, Notch Effects on Waveshape
When recording ECG, it’s important to consider the impacts of electrode placement and amplifier filters on the measured biopotential. Electrode placement is important because different orientations of biopotential surface electrodes with respect to the heart’s position will result in different signal shapes. As the heart contracts and expands, during the course of pumping blood, a time-varying, voltage signal is propagated over the volume of the heart. This voltage signal conducts to the surface of the body and is detectable between any two points on the body that straddle the heart. The signal can look quite different depending on whether the signal is collected from left hand to right hand or from right hand to left leg. When the ECG signal is recorded far from the heart, electrode placement is not critical to obtain similar-looking ECG waveforms when recording over time. As electrodes are placed increasingly close to the heart, then electrode placement becomes quite specific to recreate similar-looking ECG signals. This is because the body is a volume conductor and the heart is not a point-source of electrical signal activity.
The ECG waveform has been specified to consist of frequency components in the range of 0.05Hz to 150Hz, even though the characteristic ECG waveform shape is largely composed of components from 0.05Hz to 35Hz. However, even if the ECG’s primary waveshape is preserved, bandwidth limits affect usability of the data. For a high resolution ECG recording, the signal bandwidth should be maximized. Although the Association for the Advancement of Medical Instrumentation (AAMI) standard recommends an ECG bandwidth of 0.05-150Hz, surface electrodes measurements can show ECG components up to 500Hz. 
As with any physiological measurement from a subject, the balance lies in attempting to isolate the signal of interest from residual noise sources. In the case of ECG recording, interfering noise is often generated by skeletal muscle that lies in the superposition path defined by electrode placement. This noise (recorded as the electromyogram) can be especially problematic if the subject is intentionally moving during the ECG measurement. Other noise sources include interference resulting from mains-generated displacement currents (50/60 Hz hum).
Depending on the nature of the ECG measurement requirements, filtering methodology can be a helpful ally to reduce the effects of interfering noise sources. The critical aspects of filtering are a function of the filter’s magnitude and phase characteristics. Choice of filter topologies can have a significant effect on the shape of the ECG. Given that AAMI guidelines are primarily concerned with ECG filter flatness from 0.05-150Hz, typically, second-order (or higher) Butterworth filters are used to define these limits.
In some situations, it can be helpful to craft a compromise between maximizing ECG signal bandwidth, preserving the essential ECG waveshape, reducing the impact of skeletal myo-electrical noise and removing mains interference. If this collective compromise is not well-considered, certain problems can arise. For example, if notch filters are used to remove mains interference in the ECG signal processing channel, significant distortion can be introduced into the ECG waveshape. Accordingly, notch filters are generally avoided or only used in combinations with lowpass filters to minimize undesired effects.
When constraining the ECG signal bandwidth, it becomes increasingly important to be aware of the phase linearity of the filters used to bandlimit the channel. For example, if a 35Hz, second order, lowpass, Butterworth filter is used instead of one at 150Hz, then an observable “ringing” will be evident after the QRS interval. This ringing occurs because of the phase non-linearity of the 35Hz Butterworth filter. However, if the 35Hz filter is adjusted to have more “Bessel-like” phase qualities, the problematic ringing can be eliminated. When limiting the ECG signal frequency band to less than 0.05-150Hz, reasonably phase linear highpass and lowpass filters will preserve the essential waveshape of the ECG. Phase linearity implies constant group delay. Constant group delay means that all frequencies in the filter passband are delayed by the same amount of time. Phase linear analog filters, and their digital equivalents, can be designed to process an ECG signal with minimal, filter-induced, distortion.
In addition to the above-listed simple strategies, a host of digital signal processing methods are available to optimize the previously mentioned compromise. Methods include linear and non-linear adaptive filters, synchronous adaptive noise cancellation, dynamic neural networks, coherent averaging, transform domain singular value decomposition and Legendre moments. 
EDA – Attachment points, Bipolar, Unipolar
Bipolar Measurements – Electrodes are placed on Medial Phalanx on index and middle fingers or on Thenar Eminence and Hypothenar Eminence.
Unipolar Measurements – Signal electrode is placed at Thenar Eminence or Hypothenar Eminence. Reference electrode is placed on the same arm, more proximally to the body, on the inside forearm.
Bipolar Measurements – Electrodes are placed on the Medial Site, on the side of the foot, over the abductor hallucis muscle and midway between the ankle and proximal phalange of the big toe.
Unipolar Measurements – Signal electrode is placed on the Medial Site, on the side of the foot, over the abductor hallucis muscle and midway between the ankle and proximal phalange of the big toe. Reference electrode is placed on the same foot, more proximally to the body, just above the inside ankle.
Isotonic electrolyte (example: 0.05 Molar NaCl, 0.3% chloride content) is used to avoid hypersaturating or hyposaturating eccrine glands. Use of highly hypertonic electrolyte gel results in reduced EDA levels and responses over time. Isotonic electrolyte gel use establishes a framework for non-invasive EDA measurements so these measurements will have minimal impact on the skin site chemistry. Skin sweat is a weak electrolyte and it tonically varies between chloride levels of 0.02 and 0.060 Molar. Blood plasma varies between chloride levels of 0.09 and 0.011 Molar. In practice, an EDA electrolyte with a 0.1% to a 0.5% chloride salt concentration is typically suitable, with concentrations in the 0.3% to 0.5% range as being somewhat more stable, using Ag/AgCl electrodes, due to higher Cl- ionic content.
Different qualities of electrode gels, other than those associated with ionic concentration, can introduce skin conductance level variations over time. These changes correspond with the viscosity of the gel and the amount of free water in the gel. Wetter gels seem to provide an increased sensitivity to sweat activity in partially-filled sweat ducts. Wetter gels appear to provide less repeatable measurements due to changes, introduced by the gel, that appear to depend on stratum corneum characteristics. Wetter gels may also be fully absorbed by the underlying skin during long-term recordings. Electrode gels with reduced mechanical stability (lower viscosity, wetter gels) appear to be more susceptible to artifacts induced by pressing on the electrode. Solid gels (hydrogels) appear to have improved ability, compared to wetter gels, to return skin conductance to baseline after a sweating period.
EEG – Frequency Banding
Many EEG studies employ methods to discriminate between the various frequencies generated by neuronal activity in the brain. The EEG frequency bands defined by Delta, Theta, Alpha, Beta and Gamma are commonly referenced. When discriminating between bands, problems may occur when attempting to classify signal frequencies that reside on the edges of the bands. This is because the filters used may not be perfectly steep at the band limits. In these cases, EEG signals may be classified as being in two bands at the same time. Also, if a large signal is present in one band, a portion of that signal may reflect into an adjacent band. In some classification studies, these phenomena may not be particularly important as long as awareness exists. When precise EEG signal frequency classification is required, then certain processing methods can be employed to assist this effort. Two commonly used methods are:
1. Power spectral density determined over a specific band
2. High-order FIR or IIR filtering
EEG – Number of Channels
The EEG measured via surface electrodes is called the sEEG. When using surface electrodes, on the scalp, quite a bit of data is measured. A significant problem in EEG is that interesting data, from differing brain regions, is all combined together. EEG is faced with a localization problem because all the signals in the neuronal mass of the brain are mixed together via superposition. The conductive pathways provided by the interstitial fluids in the brain permits the combining of all source neuron-generated signals. However, because the conductance between the volumes of the brain is finite, sEEG measures will show the relative activity of a brain region as a larger magnitude when that brain region is correspondingly closer to the recording surface electrode.
From a spatial sampling perspective at least 128 channels of sEEG are required to adequately characterize the electrical fields, generated by the brain, as they map onto the surface of the scalp. However, in many EEG studies, accurate characterization of total brain volume field mapping is not required. Certainly, EEG measurements can be performed using any two electrode locations on the scalp. These are simple differential measurements of the biopotentials measured between those contact points.
Signal processing methods, such as Independent Component Analysis (ICA) and Principal Component Analysis (PCA) can be employed on sEEG measurements to help isolate signals of interest. These are known as “blind source-signal (BSS) separation methods”. BSS implies that the axes of projection (sources) are determined through the application of an internal measure and without the use of any a priori knowledge of data structures. These techniques, and associated evolutions, can be be employed on one to N channels of sEEG data to isolate data components. Both ICA and PCA are feature extraction techniques used for dimensionality reduction. ICA employs non-Gaussian aspects of the wave data, using the third moments and above, to generate the component waves. PCA employs Gaussian aspects of the wave data, using the first and second moments, to generate component waves. In a multidimensional space, ICA identifies the component waves that maximize their statistical independence and PCA identifies the component waves that maximize their variance.
EGG – Recording Close to DC
EGG measurements are at very low frequency, in the range of 0.015 Hz to 1.5 Hz. The time constant associated with a highpass filter of 0.015 Hz is about 10.6 seconds. Because EGG measurements are performed via electrodes, and the electrode / skin interface junction incorporates both half-cell and skin potentials, these potentials can confound the EGG measurement if they vary in the same frequency band as the EGG.
Generally, when performing EGG, hypertonic electrolyte gel is used with Ag/AgCl electrodes. Measurements are performed after the electrode-electrolyte-skin half cell potentials and skin potentials have settled, after electrode application to skin (15-20 minutes).
Recording electrodes are placed cutaneously, above and along the antral axis of the stomach. One bipolar recording electrode is placed halfway between the umbilicus and xiphoid. The other bipolar electrode is placed a few centimeters (3-7) to the left of the first biopotential electrode. The ground electrode can be placed anywhere on the torso.
EMG – Effects of HPF and LPF
Surface EMG biopotentials (sEMG) have a frequency range of 10 Hz to 500 Hz. Signals are typically recorded using biopotential pairs of recording electrodes. Distances between recording electrodes vary from 2cm to 10cm depending on the size of muscle group under investigation. Ground electrodes can be placed anywhere on the body.
EMG signals can be immediately recorded after electrode application, due to the frequency range of EMG data. A wide variety of electrode types can be employed, ranging from polarizable to non-polarizable. Electrolyte use is similarly optional, as even capacitive-type electrodes (polarizable) are typically suitable for EMG recording.
A common class of EMG electrodes are active electrodes. These electrodes typically incorporate stainless steel skin contact pads, coupled with differential amplifiers, integrated into a small ergonomic carrier that can be attached to the skin via tape or elastic bandage.
EOG – Vertical, Horizontal and Circular Eye Movement
EOG signals have a wide range in frequency, from DC to 100 Hz. Because EOG measurements are performed via electrodes, and the electrode / skin interface junction incorporates both half-cell and skin potentials, these potentials can confound the EOG measurement if they vary in the same frequency band as the EOG.
Generally, when performing EOG, hypertonic electrolyte gel is used with Ag/AgCl electrodes. Measurements are performed after the electrode-electrolyte-skin half cell potentials and skin potentials have settled, after electrode application to skin. (15-20 minutes).
For recording eye movement in the horizontal axis, bipolar recording electrodes are placed on the left and right temples. For recording element movement in the vertical axis, recording electrodes are placed above and below one eye. The ground electrode can be placed anywhere else on the head.
Bioimpedance – Impedance Cardiography
Bioimpedance measurements are typically performed using a Kelvin probe (four terminal) configuration. In this setup, current injection electrodes and voltage measurement electrodes are independently attached to the subject tissue volume. This methodology is used to remove errors that can occur due to the impact of the electrode to subject attachment junctions. For example, each electrode /skin interface junction is modeled as a voltage source coupled with a resistor and capacitor network. This complex impedance junction can confound the bioimpedance measurement of the subject tissue volume unless accounted for properly. By isolating the current injection electrodes from the voltage monitoring electrodes the following circumstances arise:
1. The measurement stimulation current is injected through one electrode / skin junction, through tissue volume, and then removed through another electrode /skin junction. All current is directed through the tissue volume and is unaffected by the electrode / skin junction impedances.
2. The measurement voltage, resulting from the injected current through tissue volume, is sensed by electrodes attached to the tissue volume. Because no current flows through the voltage sensing electrodes, there is no voltage drop across the two voltage-sensing electrode / skin junctions. The measured voltage will just be the voltage developed across the tissue volume, resulting from the applied stimulation current.
Typically, an alternating current is used to sense bioimpedance for a couple of reasons:
1. The combined series impedance of the electrode / skin electrode junctions and tissue impedance can be quite high at DC. This high residual impedance can create impractical circumstances for measuring small impedance changes around the residual value. This is because only a small excitation current can be practically sourced through a high impedance. This high residual impedance, however, drops rapidly with increasing frequency of the stimulation current.
2. Half-cell and skin potentials are present in all electrode / skin interface junctions and these potentials can vary for a variety of reasons. If a DC stimulation current is used, then the resulting DC voltage developed across the tissue volume will be superimposed with the half-cell and skin potentials at the voltage measurement electrode / skin junctions, thus possibly confounding the measurement. In practice, even if the electrodes are identical, small differences in electrode /electrolyte half-cell potentials can develop between a measurement pair of voltage sensing electrodes. Furthermore, skin potentials can vary between recording sites. If alternating stimulation current is used, the potentials at the voltage measurement electrode / skin junctions can be ignored via the use of highpass filtering.
As with any electrical stimulation system, attention to voltage and current compliance is required. Typically, current stimulation is used for bioimpedance measurements. The stimulation current is directed through the tissue volume via electrode / skin junctions. The voltage compliance required by the current stimulator will be the combined series impedance of the two electrode / skin junctions and the tissue volume. For bioimpedance measurements impedance on the body, the tissue volume impedance is substantially smaller than the combined series impedance of the electrode / skin junctions. Accordingly, voltage compliance levels, for the constant current source, are largely determined by the electrode-electrolyte-skin contact areas.
Considerations for ICG
Historically, strip electrodes are used for ICG recording to establish circumferential equipotential lines at the points of current injection and voltage measurement. An equipotential line created in the contact area between the strip electrode and the skin surface because the strip electrode has a significantly lower impedance, at ICG measurement frequencies, then the volume conductor of tissue underneath.
ICG derived measurements, such as Kubicek’s and Sramek’s expressions for Stroke Volume (SV), depend on the variable “L”, which is the distance between the voltage measurement electrodes defining the measurement volume on the torso. Kubicek’s expression indicates that SV is proportional to L^2 and Sramek’s equation indicates that SV is proportional to L^3.
Given this mathematical relationship, SV is highly sensitive to the physical location of the equipotential lines defining the voltage measurement locations. When equipotential lines can be forced along a specific circumferential path, by the use of low impedance circumferential strip electrodes, then location estimation error is reduced.
If a spot electrode system is used to emulate the location of equipotential lines, then there is increasing reliance on the impedance homogenous nature of the torso tissue volume. Assuming the torso can be modeled by a three dimensional mesh of smaller volume impedances, and if those impedances are equivalent at the test frequency, then two spot electrodes can be expected to emulate a circumferential strip electrode that is the shortest circumferential path between the spot electrodes.
In the case of spot electrode measurements for ICG, each skin/electrode contact area will have a characteristic impedance. This impedance is dominant in the voltage measurement (Vt) loop because the electrode/skin contact area is smaller then the cross sectional area of any other tissue plane orthogonal to the flow of current through the torso.
When two, circumferentially-defining, voltage measurement electrodes are linked (via a “Y” junction) and then directed to one ICG voltage measurement input, they create a summing junction. The summing junction output voltage (Vto) is defined as follows:
Vto = [Vtr * (Rt + Rsel) / (2*Rt + Rsel + Rser)]
+ [Vtl * (Rt + Rser) / (2*Rt + Rsel + Rser)]
Vtr = Voltage from right torso electrode
Vtl = Voltage from left torso electrode
Rt = Impedance of torso
Rser = Impedance magnitude from right torso skin/electrode
Rsel = Impedance magnitude from left torso skin/electrode
Assuming: Rsel = Rser then:
Vto = (Vtr/2) + (Vtl/2)
If there is an imbalance between Rsel and Rser then Vto becomes more relatively dependent on Vtr or Vtl. However, this issue doesn’t matter if Vtr = Vtl.
The use of voltage measurement leads (prior to the summing “Y” junction) with embedded series resistance helps to alleviate concerns associated with this situation because of the pre-defined nature of any included series lead
resistance. This known series resistance adds to the skin/electrode junction impedances to create a larger resistance. As the added (known) lead resistance becomes larger relative to Rse, then variations in Rse will have less impact on the Vto signal.
When performing measurements with carbon composition leads, the effective resistance is approximately 156 ohms per meter. For two 30 cm voltage electrode monitoring leads, implemented prior to the Y junction, the distributed resistance (Rel) in each lead will be: 156 * (30/100) = 46.8 ohms.
Assuming Rser and Rsel are both less than Rel/X, then potential impacts on Vto, resulting from variations between Rsel and Rser, are reduced by roughly a factor of X.
Using identical resistance leads (between subject connections and the “Y” junction) helps for a consistent ICG measurement. It also helps to use identical electrode/skin prep methods on both sides of torso. However, this total effort will typically only be relevant for ICG waveform magnitude as opposed to timing relationships.
Furthermore, this issue is only relevant for the ICG voltage measurement electrodes because the voltage measurement is relatively insensitive to the current injection electrode locations. This insensitivity occurs because the current electrodes are driven from a current source, that will simply force current through the torso, subject to the compliance voltage of the current source.
Another non-invasive method to measure cardiac output (CO) is to employ the arterial blood pressure (ABP) signal. There are at least ten different types of estimators, that can be employed, to calculate CO from ABP. The simplest estimator is based on the mean arterial pressure (MAP) minus the mean venous pressure (MVP). Assuming MVP is much less than MAP, then MAP is considered to be proportional to CO. The proportionality constant is the total peripheral resistance (TPR).
CO = TPR * MAP
This equation is modeled after Ohm’s law, where Current = Resistance * Voltage, where all values are time averaged.
A slightly more complicated estimator is based on the Windkessel model, where TPC is total peripheral capacitance, PPBP is the peak to peak arterial blood pressure (Systolic – Diastolic Blood Pressure) and F is the Frequency of the blood pressure signal.
CO = TPC * PPBP * F
Plethysmography – Describe Methods
Plethysmography is a procedure for measuring changes in volume of a body part, or whole body, that results from fluctuations in the amount of liquid (typically blood or water) or gas (typically air) it contains.
The general plethysmography measurement concept is associated with the consequence of volume displacement. As the measured tissue volume changes, the circumference and/or length of the volume also changes. Many plethysmogram transducers are designed to measure volume changes via circumferential variations in tissue volume over a specific cylindrical length. Air plethysmography employs an air cuff that is wrapped around the body part of interest, typically a limb. Volume changes in the limb results in pressure changes in the partially filled air cuff. Strain gauge plethysmography typically employs an elastic tube filled with a liquid metal conductor, such as mercury or indium-gallium. The tube is wrapped around the body part and changing circumference, in the tissue volume, results in resistance changes over the length of the tube. Other types of strain gauge plethysmogram transducers utilize conductive elements that vary in resistance, capacitance or inductance as they are stretched.
Impedance plethysmography is a non-invasive method used to detect blood flow and relative volume in parts of the body. This measurement employs a high frequency excitation current that is directed through a specific tissue volume. Voltage measurement electrodes are used to record the voltage generated across a specific length of tissue volume and are placed between the current injection electrodes. Generally, voltage measurement electrodes employ conductive strip electrodes to establish a specific circumferential equipotential lines that define the tissue volume investigated. The tissue volume impedance is determined by dividing the voltage measurement by the current injected. The phase of the tissue volume impedance is simply the difference between the voltage phase and the current phase.
Blood Volume Pulse (BVP) – Typically measured using a pulse pressure pad or photo-plethysmogram transducer.
Limb Plethysmography (LPG) – Typically measured using an air or impedance plethysmogram transducer.
Penile Plethysmography (PnPG) – Typically measured using a strain gauge (liquid conductor) plethysmogram transducer.
Vaginal Plethysmography (VPG) – Typically measured using a vaginal photo-plethysmogram transducer.
Total Body Plethysmography (TBP)
In TBP, subjects sit inside a airtight chamber designed to measure flow and pressure. The principle of TBP is based on the phenomenon that, for a fixed amount of gas in a sealed chamber, relative changes in the chamber volume are a function of relative changes in chamber pressure.
The defining mathematical expression is:
P1 * V1 = P2 * V2
TBP relies on detecting changes in chamber pressure simultaneously with changes of pressure at the mouth or airflow under specific respiratory conditions. TBP is used to determine airflow resistance and fixed lung volumes.
Transducer Data – Describe Attachment Location and Methodology: See Appendix C
Stimulus/Response/Averaging – Describe Methods
Many different types of stimulation are possible and are simply subject to the range of human senses. Tactile, electrical, olfactory, taste and temperature stimulation methods can be constructed.
Any physiological measurement is concerned with the subject of stimulus and response because all physiological processes are subject to change with applied stimuli. The processes of life encapsulate the idea that environmental change precipitates sensory activity that results in stimulation. Because sensory activity is mediated by the flow of ions in the nervous system, highly-controlled electrical stimulation capability can provide a unique tool for the life science researcher to discern sensory function and associated physiological responses in living subjects. Computer-based systems can be used to define very exacting electrical stimuli and those signals can be directed to a subject via electrodes. The level of stimuli can be extremely wide and the shape of the stimuli can be finely controlled. Signals can be introduced directly into nerves or muscle fibers to evoke a specific response.
There are two types of electrical stimulators, voltage and current. A voltage stimulator behaves like a voltage source, in that it outputs a variable voltage with low output impedance. A current stimulator behaves like a current source, in that it outputs a variable current with high output impedance.
The primary physical expression that characterizes the behavior of electrical stimulators is:
Vs = Is * Zs
Is – stimulation current from stimulator
Vs – stimulation voltage from stimulator
Zs – impedance of stimulation loop
Zs is the collective impedance of the complete stimulation loop. From a simplified view, Zs consists of the series combination of two electrode / skin junction impedances and the stimulated tissue volume. Zs has resistive and capacitive components, so the capacitive components charge in accordance to the expression:
Ics = Cs * (dVcs/dt)
Ics – stimulation current through Cs
Cs – capacitive component associated with Zs
dVcs/dT – change in stimulus voltage, across Cs, as a function of time
Accordingly, electrical stimulation is associated with the charging of the stimulation loop capacitance. If fast electrical stimulation is required, then the stimulator should have high current level drive capability. For a given Cs, the speed of charging (dVcs/dt) will be higher (faster) as Ics is increased.
There are two major sub-impedances in the stimulation loop that define the portion of the stimulator output voltage (Vs) that is directly across the tissue volume (Vt):
Ze – electrode to site impedance (total of both sites)
Zt – stimulated tissue volume impedance
Vs = Is * Zs
Vs = Is * (Ze + Zt)
Vt = Vs * (Zt / (Ze + Zt))
Because of this relationship, the value of Vt might not be reliably specified, given a certain Vs.
For any given tissue volume, electrical stimulation can be specified in terms of voltage across the tissue volume (Vt) or current through the tissue volume (It). In practice, because two terminal stimulation is simple to apply, current stimulation is often used because the current through the tissue volume (It) is the same as the sourced stimulation current (Is). However, if the stimulation current (It) is specified then Vt may be indeterminate. Sometimes, both (It) and (Vt) are required to be known, such as with bioimpedance measurements. In this case, voltage monitoring electrodes can be placed on the tissue volume so both (It) and (Vt) can be identified.
All stimulators incorporate a fundamental characteristic known as “compliance”. Compliance comes in two flavors, namely voltage compliance and current compliance. Voltage stimulators emulate a voltage source, up to a certain current compliance. Current stimulators emulate a current source, up to a certain voltage compliance. The stimulator compliance limit is important information. Here are examples:
A voltage stimulator is being used to establish a field stimulation in a tissue bath. The voltage pulse is 100 volts, with a width of 1 ms. The bath establishes a 1000 ohm impedance between stimulation electrodes. Therefore, 100 volts/1000 ohms is 100 ma. In this situation, the voltage stimulator must be able to provide a current compliance up to 100 ma for at least 1 ms. If the stimulator’s current compliance is limited to 50 ma, then the output voltage will be limited to 50 ma * 1000 ohms or 50 volts, even if the stimulator is being instructed to output 100 volts.
Voltage stimulators commonly operate with the help of high voltage DC-DC converters and storage capacitors to provide the needed compliance current. If the storage capacitors can’t store enough charge or are not sufficiently supplied, then voltage stimulators may exhibit a quality known as voltage droop. Voltage droop is the quality when the stimulator output voltage can’t be maintained over a specific time interval, due to lack of sufficient current compliance. If the above stimulator can only be current compliant to 100 ma for 0.5 ms, then the output voltage pulse will be 100 volts for 0.5 ms and then will decay in direct proportion to the current compliance limit.
A current stimulator is being used for neuromuscular stimulation between two small electrodes. The current pulse is 10 ma, with a width of 1 ms and there is a 10,000 ohm impedance between stimulation electrodes. Therefore, 10 ma * 10,000 ohms is 100 volts. In this situation, the current stimulator must be able to provide a voltage compliance up to 100 volts for at least 1 ms. If the stimulator’s voltage compliance is limited to 50 volts, then the output voltage will be limited to 50 volts / 10,000 ohms or 5 ma, even if the stimulator is being instructed to output 10 ma.
Current stimulators commonly operate with the help of high voltage DC-DC converters and storage capacitors to provide the needed compliance voltage. If the storage capacitors can’t store enough charge or are not sufficiently supplied, then current stimulators may exhibit a quality known as current droop. Current droop is the quality when the stimulator output current can’t be maintained over a specific time interval, due to lack of sufficient voltage compliance. If the above stimulator can only be voltage compliant to 100 volts for 0.5 ms, then the current pulse will be 10 ma for 0.5 ms and then will decay in direct proportion to the voltage compliance limit.
In the case of transcranial Direct Current Stimulation (tDCS), a constant current of up to 2 ma (Is) may be applied through the cranium. Assuming a combined series electrode / skin junction resistance of 20 Kohms, a cranium resistance of 100 ohms and other loop series resistances considered negligible, then the total stimulation loop impedance (Zs) is 20.1 Kohms. Accordingly, in this situation, the voltage compliance (Vs) of the tDCS current stimulator must be at least:
Is * Zs = Vs
0.002 amp * 20,100 ohms = 40.2 volts
Given the variable nature of the combined series resistances in the stimulation loop, it’s generally helpful to double the expected compliance voltage requirement of the stimulator. Accordingly, in this case, a stimulator compliance voltage of 80 volts or higher is recommended. Because the series resistance in the loop is largely a function of the electrode / skin interface junctions, the choice of electrode type and contact area has a great effect on the voltage compliance requirements of the current stimulator. Other factors contributing to the voltage compliance requirement are the equilibrium potentials and over-potentials associated with the specified electrode / electrolyte / skin junctions. Furthermore, if high resistance electrode leads are used, the potential across the leads may become material.
MRI and fMRI
Performing biopotential recordings in the MRI can be quite challenging. Typical biopotential signals are very small and the generated MRI RF energy and associated magnetic field shifting can corrupt the measurement. However, typically to a significant degree, the characteristic artifact introduced by MRI scanning process can be removed from biopotential recordings. MRI scanning-induced artifact results from a combination of RF energy pulsing and magnetic field gradient switching. Scanner-generated RF energy induces high frequency current flow, via coupling, in the tissue volume being measured and the associated measurement conductors. Magnetic field switching induces transient current flow in the tissue volume and in the measurement conductors. Accordingly, scanner-induced signal artifact qualities are a function of scanning sequence type and timing. Other factors that affect the nature of artifact are:
1. Spacing between measurement conductors
2. Type of measurement conductors
3. Type of electrodes, electrolyte and skin preparation methods
4. Patch panel filtering characteristics
5. Measurement conductor shielding
6. Volume of tissue being measured and location with respect to bore
Methods for Retrieving Noise-impacted Data – Noise Periodic or Random
Procedures for extracting meaningful data from physiological measurements typically encompasses methods to remove artifact from the desired data. Common methods to isolated desired data include:
1. Linear Filtering
2. Template Matching
3. Modeling – such as with Fourier Linear Combiners
4. Non-linear Filtering
5. Heuristic Algorithms
6. Orthogonal Transformation – via Fourier Transformation, Wavelets
Highly-processed Results (BPM, Median Frequency, HRV, PSD, etc.)
These algorithmically-based methods are used to extract higher-ordered information from a physiological signal data set. Examples of this type of processing include:
1. Heart Rate Variability
2. Beats Per Minute
3. Respiration Rate
4. Interbeat Interval
5. Mean and Median EMG Frequency
6. Power Spectral Density
7. ECG Wave Analysis (PQRST complex timing)
8. Stroke Volume, as derived from ICG
9. Cardiac Output, as derived from ICG
10. EEG Frequency Band Power (delta, theta, alpha, beta, gamma)
11. Volume of CO2 Produced (VCO2) or O2 Consumed (VO2)
12. Velocity or Acceleration of Joint or Angle, as calculated from position data
13. Integrated EMG
14. EDA Analysis
15. Actigraphy Analysis
16. Sleep Staging
17. Nerve Conduction Velocity
18. Pulse Transit Time, estimate Blood Pressure
19. Tidal Volume
20. Forced Expiratory Volumes
21. Body Fat (BF)
22. Total Body Water (TBW)
23. Fat Free Mass (FFM)
24. Body Cell Mass (BCM)
25. Extracellular Water (ECW)
26. Respiratory Exchange Ratio
27. Quantitative EEG
28. Resting Metabolic Rate
29. Energy Expenditure
30. Systolic, Diastolic and Mean Blood Pressure
31. Eye-blink Artifact Removal from EEG Data
32. Baroreflex Sensitivity
33. Pressure/Volume Loop Analysis
34. EEG Seizure Detection
35. Cardiac Output Estimation from Blood Pressure
36. Respiratory Sinus Arrhythmia
37. Cognitive State Classification
38. Eyeblink Removal
Ionic content in sweat
Capacitance in tissue
Excellent Study on Electrode Gel Types for EDA
Drift Characteristics Electrodes and Gels
Ideal EGG electrodes
Cardiac Output Estimation from Arterial Blood Pressure Waveform
Interference Cancellation Methods:
Single Channel EEG
Spatial sampling of EEG
EEG and ICA / PCA