Aspectos destacables del an´ alisis

Properties of the electrodes and the electrode-skin interface have an impact in any biopotential recordings, and can be an accuracy limiting factor when performing EIT. The interface between the electrode and the skin resists the flow of current, which is termed the contact impedance. Large contact impedances can give rise to large potential drops at the electrode-skin interface which can mask those caused by the underlying tissues, therefore making impedance measure- ments difficult[95]. Reactive components of the contact impedance can also generate phase shifts leading to misleading measurements[96]. In principle, the four-electrode measurement system should cancel out the effects of contact impedance. However this is only the case if the impedances are exactly matched at all four electrode sites. In reality, variations in skin impedance and stray capacitance between channels cause significant errors in recorded impedance values[38] [97]. The mismatch of contact impedances in the current drive circuit develop a common mode voltage on the body, while those found in the voltage measurement

circuit effectively reduce the common-mode rejection ratio (CMRR) of the differential ampli- fiers, and thus reduce the elimination of common-mode voltages[98]. Mismatches in contact impedance as small as 20 % have been shown to render EIT images “almost meaningless” [73],[99]. Results from the UCL MFEIT study show that contact impedances variations of this order can be expected on scalp recordings in the ward[100]. Therefore reducing the contact impedance mismatch is of primary concern in EIT of acute stroke.

Traditionally, gel is placed between the electrode and the skin which has a number of benefits (described below). Therefore the contact impedance represents the electrode-gel-skin interface which can be simplistically modelled as a resistor-capacitor network shown in figure 1.13 [95]. Contact impedances are dominated by the impedances of the outer layers of the skin, and are largest at low frequencies. It is at these frequencies (≈20 Hz) where the

contrast between healthy and unhealthy brain tissues is greatest, see section 1.3.1, so contact impedance causes particular difficulties in MFEIT of acute stroke.

Figure 1.13: Simple equivalent circuit model of electrode-gel-skin interface from[95]

Skin impedance and abrasion

The skin impedance is dominated by the impedance in the epidermis -R_pandC_pin parallel, which can be further divided into two sub layers: the Stratum Corneum, the outer keratinised layer and deeper tissues including the stratum gerninativum. The Stratum Corneum is the outer skin layer consisting of dead cells, it acts as a fluid barrier and this has electrical isolation characteristics [101]. The equivalent circuit is now given in figure 1.14, with by far the greatest contribution fromR₁, the resistance of the Stratum Corneum[102]. Removal of layers of the Stratum Corneum therefore results in a significant decrease inR₁and thusRp (and hence overall contact impedance). A study using adhesive tape to progressively remove layers of the epidermis of approximately the same thickness demonstrates dramatic reduction

inRpas shown in figure 1.15[102]. The outer layers of the Stratum Corneum are the most resistive, so the most dramatic decreases are found after the first few strippings.

Figure 1.14: Equivalent electrical circuit of the epidermis from[102]

Figure 1.15:Frequency characteristics of skin impedance from from[102]represented as values for the equivalent circuit in figure 1.14. Number of tape strippings given by numbers in parenthesis, Roman

numerals denote corresponding data in R_pand C_p

Due to these effects it is standard clinical practice to perform abrasion of the electrode contact site before taking measurements. This abrasion is typically performed by hand by a clinician using a cotton bud with abrasive paste, or by rubbing with sandpaper as shown in figure 1.16.

It is important to consider the inter and intra human variations of skin properties as they dictate the potential decrease is overall contact impedance. The thickness of the Stratum Corneum has been found to vary from as little as 10µm to well over 100 µm in a single patient depending upon the area of the body measured[95]. Skin hydration greatly affects the value ofR_p and considerable daily variations have been shown, as well as long term changes over seasons[103]. These issues coupled with the expected inter patient variations may explain why there is little agreement in the literature on the exact values of the impedance of human skin[104]. Therefore it is difficult to create any rules as to the amount of abrasion

stratum corneum stratum germinativum dermis 100 ₁₀2 ₁₀4 0 1 2 3x 10 4 Impedan ce | Z| Ω Frequency (Hz) 100 ₁₀2 ₁₀4 0 1 2 3x 10 4 Impedan ce | Z| Ω Frequency (Hz) Cotton Bud Abrasive Paste Electrode

Figure 1.16: Effect of abrasion of electrode site on the resultant contact impedance. Impedance values

taken from[102]for zero and 9 strippings

that is required to reach a low contact impedance as the impedance before abrasion, and the amount of tissue which can be safely removed vary to such a large extent, and depend upon the frequency of interest. However, as the impedance of the underlying tissues (R₂ andC₂ in figure 1.14) is relatively small, variations in thickness or characteristics will have less of an effect on the overall contact impedance. Hence, if the Stratum Corneum is largely removed, it would be expected that the variation in contact impedance across electrode sites and between subjects would decrease. This assumption allows for the use of general guidelines of acceptable contact impedances in clinical settings. For example, a value of 1 kΩat 10 kHz per electrode is deemed acceptable for use in the UCL stroke trials[7]. The removal of the Stratum Corneum is not controlled or repeatable with manual abrasion. Instead the contact impedance is checked after electrode preparation, and the process is repeated for electrodes with contact impedances above the threshold. However, as has been described previously, impedances mismatches of as little as 20% are unacceptable for imaging and performing abrasion by hand does not allow for accurate control of the impedance.

Gel impedance

Electrode gels decrease the contact impedance by ensuring optimal electrical contact between the electrode and the skin. The gel rapidly fills up the pores and wrinkles in the skin, providing the maximum possible contact area. To ensure electrical conductivity electrode gels also contain ionic salts which are biocompatible, usually NaCL and KCL. A high concentration of electrolyte is used to decrease the charge transfer resistanceR_{c t} and the skin impedance, therefore reducing the overall contact impedance[95]. However, biological tissues cannot tolerate long-term exposure to salt concentrations which differ from physiological levels. Therefore, high electrolyte concentration gels are not suitable for long term monitoring.

Electrode impedance

Theoretically an electrode can be made of any conductive material, and often practical consid- erations such as re-usability and corrosion resistance take priority over electrical characteristics. However in EIT, the electrical characteristics, particularly the contact impedance, are vital. The electrical properties generally considered desirable for an electrode-electrolyte interface are given in[95]as:

• low, stable offset potentials

• low, matched interface impedances • low polarisation

An electrode which meets all of these criteria is the silver-silver chloride (Ag-AgCl) elec- trode. The offset potential of this electrode is determined by the activity of the Cl ion and hence is small and stable when the electrode is in contact with an electrolyte containing Cl− _{ions. This is true for both fluids in human tissue and electrolyte gels[105]. Ag-AgCl} electrodes also have low interface impedances due to the roughness of the electrode surface caused by the deposited layer which effectively increases the contact surface area. A simplistic model of this decrease in impedance has been proposed, in which the interface impedance of a rough surface is proportional to the square root of that of the smooth surface[97]. Elec- trode polarisation describes the additional voltage or “over potential” caused by the current flowing through the charge transfer resistance which gives rise to an additional voltage[106]. Ideally the electrode would be entirely non-polarisable, with no voltage being dropped across the interface. However, all electrodes are polarisable to a certain extent as charge transfer resistances (R_{c t}) are unavoidable. Ag-AgCl have low values ofR_{c t} and thus are relatively non-polarisable. All noble metals have low polarisation (as well as low corrosion) properties, and as such electrodes are often made from these materials such as gold[107], platinum [46], [108]and iridium [109]. However, as they are considered to be inert, they do not react chemically with tissue, and thus do not gain the benefits offered by chloriding silver to produce Ag-AgCl. It is possible however to “platinise” a platinum electrode which deposits a fine layer of platinum on the surface, greatly increasing its roughness and decreasing the interface impedance[95].