Electrical Impedance Dermography: Background, Current State, and Emerging Clinical Opportunities

Electrical impedance dermography (EID), based on electrical impedance spectroscopy, is a specific technique for the evaluation of skin disorders that relies upon the application and measurement of painless, alternating electrical current. EID assesses pathological changes to the normal composition and architecture of the skin that influence the flow of electrical current, including changes associated with inflammation, keratinocyte and melanocyte carcinogenesis, and scarring. Assessing the electrical properties of the skin across a range of frequencies and in multiple directions of current flow can provide diagnostic information to aid in the identification of pathologic skin conditions. EID holds the promise of serving as a diagnostic biomarker and potential to be used in skin cancer detection and staging. EID may also be useful as a biomarker in monitoring effectiveness of treatment in individual patients and in therapeutic research. This review highlights ongoing efforts to improve mechanistic understanding of skin electrical changes, study of EID in a variety of clinical contexts, and further refine the technology to find greater clinical use in dermatology and dermatologic research.


Introduction
While skin cancer screening has been improved by the addition of noninvasive techniques, methods such as dermoscopy and confocal microscopy require subjective interpretation and signifcant training for efective use [1][2][3].In contrast, electrical impedance techniques, based on applying an alternating electrical current to a tissue and measuring the resultant voltages, ofer an objective and quantitative alternative to evaluate suspicious lesions [4].Te term electrical impedance dermography (EID) refers specifcally to the application of electrical impedance techniques to evaluate skin disorders [5].EID provides a quick and painless method to assess electrical changes in the normal architecture of the skin across a range of frequencies of electrical current.Previous studies have shown its potential as an adjunct diagnostic tool when applied to pre-cancerous and cancerous skin lesions [5][6][7][8][9][10][11].In addition, EID may be useful for assessing skin barrier function and can provide valuable information on treatment efcacy in conditions such as atopic dermatitis [12][13][14].Tere has been limited review of EID in dermatology, and previous work focused specifcally on its use in skin cancer detection [15].Tis article aims to discuss the basic electrical properties of the skin, review the progression of device development for measuring electrical impedance, and highlight a range of clinical and research applications of EID.

Electrical Impedance Dermography
2.1.Basic Electrical Impedance Principles.Electrical impedance methods are based on the application of alternating electrical current and the measurement of generated electrical voltage (Figure 1(a)) [16].As the electrical current fows through tissues, the amplitude of the voltage measured will be altered due to the extracellular and intracellular ionic media.Due to cells' ability to create a potential diference by holding electrical charges on both sides of the membrane, the timing of the voltage waveform will experience a time delay.Tis time delay will result in a phase shift in the measured voltage signal with respect to the applied electrical current.Te two voltage waveforms detected by the sensing electrodes are fed into a diferential amplifer (Figure 1(b)).Te diferential amplifer has two inputs, designated as noninverting and inverting inputs.Te noninverting input multiplies the high-voltage signal by a positive gain, whereas the inverting input inverts the low-voltage signal with respect to its polarity (by multiplying the signal by a negative gain).Te signals detected by the high-and low-voltage sensing electrodes are thus frst amplifed by the same amount and then summated electronically by the diferential amplifer.As a result, any voltage signals that are common to both electrodes are efectively subtracted while those that are diferent are amplifed and subsequently used to calculate the impedance.At various frequencies of electrical current, the voltage-current signals' peak amplitude ratio and their timelag relationship are used to determine the apparent electrical impedance of the tissue using Ohm's law [4], i.e., impedance equals voltage divided by current.Te measured impedance is determined by two quantities, its resistance R and reactance X components or, equivalently, its magnitude M and phase P, via the trigonometric relationships M � ������ � R 2 + X 2 √ and P � tan −1 X/R, or R � M cos P and X � M sin P.

Representation of EID Data.
When EID reactance and resistance data are plotted against each other (Figure 2(a)), the impedance Z is represented as a dot where the x and y coordinates are R and X, respectively.Te length of the line that goes from the coordinate origin to the impedance is the impedance magnitude M, and the angle P is the inclination of this line.At various frequencies of electrical current, X represented against R describes an arc of a circumference (Figure 2(b)).Equivalently, R and X can be represented explicitly against the frequency (Figure 2(c)).Sweeping across frequencies, the resistance R will exhibit monotonically decreasing dependence with increasing frequency, whereas the reactance X will show a bell-like curve.

Relationship between EID Data and the Underlying Skin
Electrical Properties.From EID, it is also possible to solve for R and X data and infer the underlying electrical properties of the skin, namely, the conductivity and the relative permittivity properties, including their directional dependence, a concept known as electrical anisotropy [5].When plotted against frequency, conductivity has a monotonically increasing dependence on frequency, while relative permittivity is the opposite (Figure 2(d)).
Skin conductivity is a measure of how skin resists or conducts alternating electrical current, whereas skin relative permittivity is a measure of its capacity to store electrical charge [17].Measured skin R and X values refect a complex combination of skin underlying electrical conductivity and relative permittivity properties.Structural changes that occur as a result of disease pathology alter the ionic and cellular components of the skin, afecting its conductivity and relative permittivity properties and impacting the EID measurements.

Instrumentation and Measurement Principles of EID.
EID measurements are made over a small area of interest, and the outcomes are objective, quantitative values that provide an electrical characterization of the skin.If skin conductivity and relative permittivity values are reported, then these outcomes are standardized and can be compared between studies.Although it is also possible to obtain an electrical impedance image of the skin, this technology is not readily available and its application has been minimally explored [18].
A minimum of two electrodes are needed to perform EID measurements (Figure 3(a)).Te frst electrode (current source) applies electrical current to the skin and the second electrode (current sink) closes the electrical circuit, allowing the current to fnd its way back to the device.At the same time, the two electrodes are used to measure the voltage signal, and their diference is amplifed by the diferential amplifer.In a two-electrode confguration, the measured impedance is the sum of the skin impedance plus the polarization impedance resulting from the contact between the two electrodes with the skin.
Measurement complexity may be increased by the addition of a third electrode (Figure 3(b)) [19].In a threeelectrode confguration, the measured impedance is the sum of the skin impedance plus the polarization impedance resulting from the contact between the current source electrode and the skin.By using a third independent voltage electrode, the voltage measured is not afected by experimental electrode artifacts that may occur at the sink current electrode (e.g., temperature drifts or poor electrode contact with the skin).
Importantly, two-and three-electrode setups are limited in that measurement of skin impedance also contains the skin-electrode polarization impedances resulting from the contact between the current electrodes and skin.Skin-electrode polarization impedance is a poorly controllable experimental factor where alterations in skin-electrode contact area, skin humidity, or temperature can give large impedance variations between measurements and corrupt readings, particularly at low frequencies where the skin-electrode polarization impedance is larger [20].Skin-electrode impedance polarization artifacts can be efectively mitigated by adding a fourth electrode (Figure 3(c)), in which independent pairs of electrodes are used to apply current and measure voltage which results in measuring only the electrical impedance of the skin.[21].

2
Dermatology Research and Practice 2.5.Electrodes.Conventional, noninvasive wet electrodes use gels and often provide the best impedance recordings.However, skin irritation with wet electrodes can be significant, and their large size limits their spatial resolution [22].Noninvasive dry electrodes typically cause less skin irritation and only require the skin to be moistened with saline to improve the electrical contact between the electrode and the skin [23].Dry electrodes also have an advantage over wet electrodes in that they can be constructed to measure small skin lesions [23].Minimally invasive electrode arrays penetrate the stratum corneum which reduces the electrode-skin contact impedance and improves the accuracy of measurements compared to noninvasive electrode arrays.However, the design of these electrodes is more complex and expensive to manufacture [24].

Historical Perspective
Historically, electrical impedance techniques have been widely used for nondestructive characterization applications in engineering and biosciences, but more recently they have been studied for potential clinical applications.Te use of EID to diferentiate normal and abnormal skin conditions has gained traction over the past 30 years (Table 1).Early impedance devices were used in research investigating the skin barrier and skin irritation in the cosmetic industry [25,26].Subsequent research evaluated electrical impedance diferences in basal cell carcinoma (BCC) [27,29].Further development of impedance techniques which penetrate the stratum corneum allowed for more accurate identifcation of melanoma [10,30].EID also shows promise in objectively evaluating skin barrier function and has been used in studies to assess the efcacy of treatments for conditions such as atopic dermatitis [12,14,38].

Limitations of Existing Noninvasive Imaging Technologies.
While accurate diagnosis of skin cancer requires biopsy and histologic confrmation, there are several noninvasive modalities available to facilitate visual examination of deeper structural components of the skin.Dermoscopy is primarily used for pigmented lesions to assist in the identifcation of melanoma, but it can also be used to identify features associated with BCC [39] and squamous cell carcinoma (SCC) [40].However, even in expert hands, dermoscopy has a sensitivity and specifcity for BCC of approximately 80% with limited ability to distinguish between BCC subtypes [1].Similarly, even with dermoscopy, SCC in situ can exhibit overlapping features with invasive SCC and infamed seborrheic keratosis (SK) [41].
Confocal microscopy allows for direct visualization of histologic structures beneath the skin surface and can facilitate the identifcation of both BCC [42] and SCC [3].However, this technology has signifcant limitations, including visualization only to the depth of the superfcial dermis, high cost, and extensive training required for image interpretation [2].Tere is interest in applying machine learning for the interpretation of clinical and dermoscopic or confocal images, but existing systems trained on twodimensional photographs have not performed well in real-world clinical settings [43].Terefore, there is a need for more objective, easy-to-use devices to assist in the assessment of concerning lesions.

Electrical Properties and Management of Skin Cancer.
Based on the biophysical electrical properties of skin, the histologic diferences between nonmelanoma skin cancer (NMSC), melanoma, actinic keratosis (AK), and SK should impact the fow of electrical current through the lesions and consequently afect skin impedance values measured by EID in four-electrode setups.For example, alterations in composition and tissue structure associated with NMSC and SK (e.g., epidermal hyperplasia, tumor cells in epidermis and dermis, stroma in BCC, keratin pearls in SCC, and UVinduced elastosis and infltration of infammatory cells) are predicted to change the ionic content of the skin and will afect its electrical conductivity.However, there is limited research characterizing the distinct electrical properties of the diferent subtypes of NMSC or melanoma.

Nonmelanoma Skin
Cancer.NMSC, specifcally BCC and SCC, is the most prevalent type of cancer with an estimated fve million cases diagnosed in the United States annually [44].BCC arises from the basal layer of the epidermis while SCC arises from the suprabasal squamous layers.When diagnosed at an early stage, NMSC can usually be treated by destruction or topical therapy, but larger or deeply invasive tumors require surgical excision that may be associated with substantial morbidity [45,46].
Tere are multiple subtypes of BCC that can be difcult to distinguish clinically, and biopsy is required for optimal histologic analysis and therapeutic decision-making [46].Histologically, the superfcial form of BCC is confned to the epidermis and the nodular form consists of round collections of tumor cells occupying the upper part of the dermis.Te more invasive BCC subtypes, micronodular and infltrative (or morpheaform), consist of smaller aggregates of tumor cells or angulated or stranded tumor cells, respectively, infltrating the deeper dermis.Te invasive forms of BCC cannot reliably be distinguished clinically from nodular or superfcial subtypes of BCC.
Similar to BCC, superfcial and invasive forms of SCC require a biopsy to diferentiate.SCC in situ is diferentiated from invasive SCC by depth of invasion histologically.SCC in situ can often be treated nonsurgically (although SCC  Dermatology Research and Practice in situ that involves hair follicles may be associated with higher rates of recurrence) while invasive SCC usually requires surgical excision [45].Having a real-time ability to diferentiate superfcial versus invasive subtypes of BCC and SCC would be invaluable for informing treatment decisions.

Melanoma.
Melanoma is the deadliest form of skin cancer, and melanoma-specifc survival improves signifcantly when detected at earlier stages [47].Melanoma develops from uncontrolled melanocyte proliferation in the epidermis, and cells may infltrate into the upper layers of the epidermis and dermis.Melanoma does not impact the stratum corneum to the same extent as NMSC.However, minimally invasive EID techniques have been successful in identifying melanoma with high sensitivity [7,10,11].Te four main subtypes of melanoma are superfcial spreading, nodular, acral lentiginous, and lentigo maligna.Each has distinct histologic features that would be expected to impact electrical impedance; however, there are no published studies on EID in diferent melanoma subtypes to date.

Benign Skin Cancer Mimics.
Another potential clinical application of EID is diferentiating skin cancers from benign mimics.SK is a benign neoplasm characterized by epidermal hyperplasia [48].Tese lesions occasionally become infamed and might be tender or exhibit peripheral erythema.Tese histologic changes can make it difcult to distinguish an infamed SK from SCC in situ.Multiple studies of EID have reported high false-positive rates for SK due to the impedance measurements being similar to NMSC and SK being incorrectly identifed as malignant [6,7].

Efcacy of EID Devices in Skin Cancer Identifcation.
Multiple studies have found signifcant, reproducible electrical impedance diferences between skin cancer and normal skin or nevi using both commercially and internally developed EID devices.Te most used device, Nevisense (SciBase AB, Stockholm, Sweden), is a system initially designed to measure pigmented lesions to distinguish melanoma from nevi, but it has been efectively applied to detect NMSC [8,9,36].Nevisense was preceded by SCMI, SciBase I, II, and III.Te diferent EID devices discussed are outlined in Table 2. Te ability to diferentiate skin cancer from normal skin or nevi was established by multiple preclinical and early clinical studies [27,29].A study of 35 BCC utilizing SciBase I examined diferences between impedance measurements of nodular (n � 15) and superfcial (n � 20) subtypes of BCC [31].However, they found no signifcant diferences in the impedance measurements between the two BCC subtypes [31].A study of noninvasive versus micro-invasive EID techniques using SciBase II found that noninvasive probes were the most efective in separating BCC (n � 28) from nevi with 86% specifcity at 96% sensitivity, whereas microinvasive probes were more efective when distinguishing melanoma (n � 13) from nevi with 80% specifcity at 92% Dermatology Research and Practice  Dermatology Research and Practice sensitivity [30].A study of 511 lesions using SciBase II found that skin cancer could reliably be diferentiated from nevi with a reported 75% specifcity at 100% sensitivity for melanoma (n � 16) and 87% specifcity at 100% sensitivity for NMSC (n � 94) [10].In a subsequent multicenter study using SciBase II, an automated classifcation algorithm to distinguish between melanoma and benign lesions showed an observed sensitivity of 95% (59/62) and overall specifcity of 49% (72/148) [11].Two large, multicenter studies provided evidence for the Nevisense system as an adjunct tool for skin cancer screening [6,7].Te primary aim of the frst study was to develop a classifcation algorithm for diferentiating melanoma from benign lesions, and two diferent algorithms were tested [6].SciBase III was used to measure a total of 1,300 lesions.Te observed sensitivities for the frst and second algorithms for melanoma were 98.1% (101/103) and 99.4% (161/162), respectively [6].Te study also identifed NMSC with 100% (25/25; 21 BCC, 4 SCC) sensitivity for the frst algorithm and 98% (49/50; 39 BCC, 11 SCC) sensitivity for the second algorithm [6].Te overall specifcities for the two algorithms were 23.6% and 24.5%, respectively [6].Te second study used Nevisense to measure 1943 pigmented lesions and reported an observed sensitivity for melanoma of 96.6% (256/265) and an overall specifcity of 34.4% [7].Te study also identifed 48 BCC and 7 SCC with a reported 100% observed sensitivity for NMSC overall [7].
Te fndings of these larger trials were replicated in subsequent studies focused on the use of Nevisense to identify NMSC.A study of 200 lesions reported a sensitivity of 94% for NMSC, missing three BCCs, and a specifcity of 43% [8].Similarly, a study of over 1700 suspicious lesions found an observed sensitivity of 100% for both BCC (82/82) and SCC (13/13) [9].EID measurements were found to strongly correlate with clinical and dermoscopic grading of AK with an observed sensitivity of 98% (49/50) [36].
In addition to the SciBase devices, several other EID devices have been used to diferentiate skin cancer from normal skin.A single probe with a matrix of electrodes (TransScan) showed that impedance changes were associated with melanoma tumor growth in mice and histopathologic fndings in human melanoma lesions with 67% specifcity and 92% sensitivity [28].
In addition, a pilot study of 17 BCCs using a fourelectrode EID device (URSKIN) found signifcant, reproducible electrical impedance diferences between BCC and normal skin [5].More recently, a pilot study using URSKIN on 35 skin lesions was able to diferentiate SCC in situ from infamed SK and normal skin (p < 0.001) [37].Unlike the Nevisense, URSKIN takes only 30 seconds for measurements, and data are transmitted directly from the device to a smartphone for data visualization and analysis.

Clinical Biopsy Decision-Making.
Te integration of EID into clinical practice could be benefcial for improving the clinical decision-making process for biopsy and treatment of suspicious lesions.Two survey-based studies evaluated the impact of EID measurements on biopsy decisions [49,50].In the frst study, 267 dermatologists were shown images of 43 pigmented lesions (16 were melanomas) and asked whether biopsy was indicated based on clinical appearance alone.Tey were then provided with corresponding EID measurements and asked if the information changed their biopsy decisions.After incorporating the impedance scores into their clinical decision-making process, the number that needed biopsy improved from 6.3 to 5.3 (p < 0.001), sensitivity from 84% to 98% (p < 0.001), and specifcity from 34% to 44% (p < 0.001) [49].Te second study of 164 dermatology trainees found that providing the impedance scores with the clinical images (45 pigmented lesions; 17 were melanomas) altered the biopsy decision in 24.3% of cases, and the addition of the score resulted in 402 fewer missed melanomas and an overall decrease of 376 biopsies of benign lesions (p < 0.001) [50].Te sensitivity for ruling out melanoma increased from 80.7% to 95.2% (p < 0.001) and specifcity from 50.4% to 58.6% (p < 0.001) [50].However, a signifcant limitation of both studies was the use of clinical images rather than in vivo examination and the omission of clinical information that would be collected in a typical visit.
Another study evaluated the impact of adding baseline EID measurements for suspicious melanocytic lesions to inform who would beneft most from short-term digital dermoscopy imaging follow-up [33].Of the 160 lesions, 6 were determined to be melanoma, and of those monitored by imaging, one was later determined to be melanoma.Te observed sensitivity and specifcity were both 83%.Using the proposed cutof scores, the need for dermoscopic imaging follow-up was decreased by 47% [33].Tese studies suggest that when used for clinically suspicious lesions, EID has the potential to increase the efectiveness of skin cancer screening by improving detection of malignancy and decreasing unnecessary biopsies and intensive follow-up.4.6.Epithelial Tissue Barrier.EID has been shown to be comparable to more established techniques such as measuring total epidermal water loss and ofers an alternative method to assess skin barrier function.An EID study using Nevisense in atopic versus nonatopic skin found signifcant changes in impedance scores between atopic and healthy skin [12].Te study determined that impedance-based techniques were useful in detecting skin reactivity and evaluating lipid content in the stratum corneum in atopic skin [12].Similarly, another study assessing lesional and unafected skin in patients with atopic dermatitis found that increased EID scores correlated with healing and decreased severity and itch scores, demonstrating that EID measurements were reliable for evaluating skin barrier dysfunction [14].A study evaluating the change in biophysical properties of atopic skin following the application of an emollient showed normalization of impedance indices following treatment [13].Te ability to objectively measure atopic dermatitis severity and monitor treatment response would be particularly useful in the development of new therapeutics.
EID has been used to assess skin hydration, mechanical damage, and response to UV irradiation.Studies evaluating skin hydration dynamics [35] and changes in impedance 8 Dermatology Research and Practice following mechanical damage to the skin barrier [32] provide a rationale for using EID in transdermal drug delivery research.In addition, EID measurements have been found to correlate with macroscopic tissue damage in UV-treated pig skin membranes.[34] Tese studies suggest that EID could be used to determine the sun protection factor value for sunscreens and assess their efcacy.

Conclusions and Perspectives
Given the promising fndings of studies using electrical impedance-based techniques, EID has demonstrated potential as an adjunct skin cancer screening tool, a reliable method to assess treatment response in atopic dermatitis, a clinical research device to study transdermal drug delivery, and a measurement tool to assist in sunscreen formulation.Larger datasets, further development of measurement techniques, and new machine learning analytics are warranted to enable accurate diferentiation of skin cancer mimics from malignant lesions and discrimination between NMSC subtypes.

Figure 1 :
Figure 1: Simplifed illustration representing the principles of an EID measurement.(a) Electrical current is applied between the source and sink electrodes, whereas the resultant generated voltages are picked by the sensing electrodes.(b) Schematic of a diferential amplifer used to amplify the voltage diference sensed by the voltage electrodes.Te common voltage present at the electrodes is cancelled by the amplifer.

Figure 2 :
Figure 2: Schematic illustrating the representation of EID data.(a) Diagram showing the relationship between EID data, resistance R, and reactance X, and their equivalent counterparts, magnitude M and phase P. (b) Representation of X versus R at various frequencies of electrical current.(c) Equivalent representation of R and X against the frequency.(d) Representation of the conductivity and relative permittivity properties of skin against the frequency.

Figure 3 :
Figure 3: Basic EID recording circuits, consisting of a current source and diferential amplifer.(a) Two-electrode setup.(b) Tree-electrode setup.(c) Four-electrode setup.Note only the impedances in black color contribute to the overall EID data measured.

Table 1 :
Summary of studies using EID.

Table 2 :
Advantages and limitations of EID devices.