Electrical Impedance Spectroscopy: Augmented Intelligence Technology for the Early Detection of Melanona

As dermatologists, we pride ourselves on our ability to visually identify suspicious lesions. Data confirm that our assessments of lesions are generally accurate—especially relative to those of our physician peers in other specialties. Research going back nearly two decades shows that dermatologists have about an 88 percent accuracy in identifying melanomas based on visual assessment alone.¹ In one biopsy-confirmed assessment study, dermatologists were shown to have a diagnostic accuracy of 24.75 percent, compared to 3.52 percent and 12.75 percent, respectively, for general practitioners and family physicians.²

A 2012 study attempted to assess dermatologists’ assessment skill through a number needed to treat (NNT) analysis. Researchers reviewed pathology reports from 2,021 biopsies performed by dermatologists at a single institution. The NNT for non-melanoma skin cancer was 1.6, while the NNT for melanoma was 15. The analysis showed that patient age, anatomical location, sex, and physician all significantly impacted on NNT values.³ Additional studies have suggested that patients with melanomas detected by dermatologists have better prognoses than patients with self-detected melanomas.⁴ This is likely a reflection of the previously shown beneficial impact on prognosis of early detection and treatment of melanoma lesions.

There has been intense interest in developing an effective, non-invasive, in-office technique to support assessment of lesions and allow dermatologists to make a more informed decision regarding which lesions to biopsy and which to watch. Dermoscopy has emerged as a reliable, low-tech tool to support visual assessment, but it ultimately relies on the same visual cues as un-aided assessment. Similarly, other more high-tech options have been developed that harness light to “visualize” lesions to depths below what can be seen with the unaided eye. AI-driven electrical impedance spectroscopy (Nevisense, Scibase) is emerging as an alternative method to assess lesions using a non-optical approach providing valuable complementary information to the visual assessment. The innovative augmented intelligence (AuI) approach focuses on artificial intelligence’s (AI) assistive role. Nevisense is designed to enhance human intelligence when examining atypical pigmented skin lesions that are difficult to assess.

Understanding EIS

Impedance is the physical phenomenon that describes opposition to a current. It is, strictly speaking, different from the electrical resistance, although the concepts are similar. Understanding the distinction is not essential to understanding EIS technology; simply imagine something that may impede the flow of a current from its source to its terminal.

Dating back more than two decades, researchers at the Karolinska Institute in Sweden noted that low voltage electrical current passes through well-organized, healthy skin tissue at a consistent and predictable rate. They also showed that the same current passes through unhealthy tissue at a different rate. They hypothesized that this impedance was a function of disarrangement of the cells in the diseased tissue and theorized that measurement of impedance could be used to detect malignancy in skin cells.

Proof of concept studies began in the late 1990s and involved 1,200 patients. These studies established that there were, in fact, predictable changes in impedance across the range from healthy skin to atypical lesions to melanoma. With time, researchers shifted focus to develop algorithms to classify melanoma. The algorithm was trained from a study involving 1,300 lesions from 19 sites in Europe. This culminated in the development of the Nevisense system, which has been validated in an international, multi-site study.

In the multicenter, prospective, blinded clinical study conducted at five American and 17 European investigational sites, Nevisense was found to be an accurate and safe device to support clinicians in the detection of cutaneous melanoma. This was the largest study ever conducted on melanoma to date. A total of 1,951 patients with 2,416 lesions were enrolled into the study; 1,943 lesions were eligible and evaluable for the primary efficacy end point, including 265 melanomas (112 in situ and 153 invasive melanomas), 48 basal cell carcinomas (BCCs), and seven squamous cell carcinomas (SCCs). Lesions were examined with the EIS-based Nevisense system, photographed, removed by excisional biopsy, and evaluated histopathologically.⁵

Nevisense had a 96.6 percent (256 of 265 melanomas) sensitivity and 34.4 percent specificity. The positive and negative predictive values of Nevisense were 21.1 percent and 98.2 percent, respectively. The observed sensitivity for nonmelanoma skin cancer was 100 percent.⁵

One study compared visual evaluation by a dermatologist to assessment with Nevisense. There was an overlap in which 85 percent of all lesions where identified by both the dermatologist (using unaided ABCD visual assessment) and by EIS; just one percent of lesions were not identified by either assessment.⁶ Interestingly, two percent of lesions were identified by dermatologist evaluation and not by EIS, whereas 12 percent were detected by EIS alone. Clearly, use of the Nevisense system augments clinical diagnosis by a dermatologist. Nevisense combines human intelligence with artificial intelligence to improve detection of melanoma. The utility of the systems relies on the skill of the dermatologist to select those lesions that warrant EIS assessment. While such assessments are done quickly and can be completed on multiple lesions in a single office visit, to apply EIS to every lesion on a patient’s skin would be time-consuming and generally ineffective.

Clinical Experience

The Nevisense system received FDA approval in June 2017 and is commercially available in the US. I am fortunate to have the system in practice since December of 2017, where I find it has a number of beneficial uses. Each time I use the system, it generates a report and risk assessment score on a scale from 0 to 10, with 10 being highly suspicious for melanoma. The device does not recommend clinical action; rather the precise, objective information provided by the system is additional information I discuss with the patient and we make the decision of whether the lesion is biopsied or closely monitored. It’s important to note that I do not use the device for every lesion. Rather, I use it for lesions that look abnormal on clinical and dermoscopic exam, but not necessarily abnormal enough to immediately warrant a biopsy or removal. These are often referred to as “lesions in the grey.” If the lesion is on the higher end of the scale, I am much more likely to remove it. If it’s on the lower end of the scale, I usually take high resolution dermoscopic images and the following the lesion closely, monitoring for clinical and dermoscopic change.

The EIS process is an efficient, point-of-care, exam that is painless and comfortable for patients. The Nevisense System was designed to fit into all practice settings and we were able to efficiently integrate it into our workflow.

1. Brochez L, Verhaeghe E, Bleyen L, Naeyaert JM. Diagnostic ability of general practitioners and dermatologists in discriminating pigmented skin lesions. J Am Acad Dermatol. 2001 Jun;44(6):979-86.

2. Martinka MJ, Crawford RI, Humphrey S. Clinical Recognition of Melanoma in Dermatologists and Nondermatologists. J Cutan Med Surg. 2016 Nov;20(6):532-535.

3. Wilson RL, Yentzer BA, Isom SP, Feldman SR, Fleischer AB Jr. How good are US dermatologists at discriminating skin cancers? A number-needed-to-treat analysis. J Dermatolog Treat. 2012 Feb;23(1):65-9.

4. Avilés-Izquierdo JA, Molina-López I, Rodríguez-Lomba E, Marquez-Rodas I, Suarez-Fernandez R, Lazaro-Ochaita P. Who detects melanoma? Impact of detection patterns on characteristics and prognosis of patients with melanoma. J Am Acad Dermatol. 2016 Nov;75(5):967-974.

5. Malvehy J, Hauschild A, Curiel-Lewandrowski C, Mohr P, Hofmann-Wellenhof R, Motley R, Berking C, Grossman D, Paoli J, Loquai C, Olah J, Reinhold U, Wenger H, Dirschka T, Davis S, Henderson C, Rabinovitz H, Welzel J, Schadendorf D, Birgersson U. Clinical performance of the Nevisense system in cutaneous melanoma detection: an international, multicentre, prospective and blinded clinical trial on efficacy and safety. Br J Dermatol. 2014 Nov;171(5):1099-107.

6. Svodoba RM, Franco AI, Rigel DS. Electrical Impedance Spectroscopy Versus Clinical Inspection Approaches: Melanoma Efficacy Detection Comparison. SKIN-The Journal of Cutaneous Medicine 2018. (2)3