Teledermatology viable for skin cancer diagnosis

Teledermatology may play a role in skin cancer diagnosis,  an accumulating body of literature indicates, says Carrie Kovarik, M.D., associate professor of dermatology, University of Pennsylvania, Philadelphia. 

“Teledermatology can be a useful tool for evaluating individual pigmented lesions, particularly with the adjunctive use of dermoscopy,” she says. “Nevertheless, as noted in practice guidelines for teledermatology from the American Telemedicine Association, pigmented lesions may present a diagnostic challenge and should require a higher index of suspicion when being interpreted,” Dr. Kovarik adds.

“Image quality is important, and the person taking the photographs must be properly trained, she says. “Furthermore, the remote consultation should occur in a setting where the patient will have access to any necessary follow-up care.”

Evidence base

Speaking at the annual meeting of the American Academy of Dermatology, Dr. Kovarik discussed selected published studies investigating teledermatology for diagnosis of skin cancer.

The most recent article was a systematic review of 21 studies conducted to evaluate the accuracy of teledermatology, its potential benefits, and factors influencing its adoption.¹ The findings of the systematic review provide support for the use of technology for remote evaluation of pigmented lesions, as well as its diagnostic performance. In addition, they show that teledermatology through the use of triage for skin cancer can reduce wait times and lead to earlier assessment and treatment. She notes, however, that the research authors pointed out the remaining need for high-quality, patient-oriented evidence from robust implementation studies.

“More outcomes data are needed, and it will be important to consider that outcomes can depend on clinician competence,” she says.

Image issues

Research by investigators from Spain demonstrates the value of adding dermoscopic images to clinical images for internet-based skin cancer screening of patients with pigmented skin lesions.² In this study, which randomized 454 patients either to evaluation of clinical images alone or in conjunction with dermoscopic images, the addition of dermoscopy significantly improved all measures of diagnostic performance, including sensitivity, specificity, false-negative and false-positive rates, positive and negative predictive values, and accuracy index.

“Other studies also support the value of attaching a dermatoscope to the camera when taking images of pigmented lesions to be evaluated by teledermatology,” Dr. Kovarik says.

She also discussed a study comparing treatment decisions made by dermatologists after in-person examination vs. in a teledermatology visit that included macroscopic and dermoscopic images.³ The study found almost perfect agreement in management recommendations made for high-risk melanoma patients after the face-to-face vs. remote assessments. The face-to-face dermatologist recommended excision of 23 lesions, of which nine were diagnosed as melanoma. All nine melanomas were identified by teledermatology.

“An important aspect of this study is that a person with some dermatology experience identified the lesions that were to be photographed. In some other studies where lesion selection was left to the patient, the results were different,” Dr. Kovarik says.

Findings of another study supported the feasibility of patient-initiated teledermoscopy for short-term monitoring of clinically atypical nevi.⁴ The patients in this study, however, received training in photography and had specific lesions to follow. Dermoscopic images were obtained at baseline and after three to four months, both at home by the patients and in the office by a dermatologist. All patients except one were able to acquire evaluable images for teledermatologist review. There was high concordance in the decisions made by the teledermatologist reviewing the patient-acquired images and the office-based dermatologist.

“This study supports the potential for teledermatology to enhance the efficiency of monitoring specific skin lesions. What we worry about with patient-driven teledermatology, however, is that it can lead to data overload if patients have free rein for submitting images of moles to their dermatologist,” Dr. Kovarik says.

Patients in the study were also highly receptive to the teledermatology short-term monitoring, but another investigation highlighted that not all patients have confidence in the approach or are able to get good images.⁵

Artificial Intelligence

Esteva and colleagues at Stanford University developed a neural network for classifying skin lesions based on image evaluation.⁶  The system was trained using a dataset of almost 130,000 clinical images representing more than 2,000 different diseases. Its performance for discriminating keratinocyte carcinomas from benign seborrheic keratosis and malignant melanoma from benign nevi was tested using biopsy-proven clinical images and found to match or even exceed the diagnostic accuracy of 21 board-certified dermatologists.

“This system still needs to be validated in a clinical scenario, but the authors suggested it could be scalable and deployable technology on mobile devices. It could be particularly valuable to primary care providers practicing in rural settings where patients need to travel far distances to access dermatology care. Guided by the neural network diagnosis, providers in this setting could triage patients who should be referred for further evaluation by a dermatologist,” Dr. Kovarik says.

References

1. Finnane A, Dallest K, Janda M, Soyer HP. Teledermatology for the diagnosis and management of skin cancer: A systematic review. JAMA Dermatol. 20171;153(3):319-327.

2. Ferrándiz L, Ojeda-Vila T, Corrales A, et al. Internet-based skin cancer screening using clinical images alone or in conjunction with dermoscopic images: A randomized teledermoscopy trial. J Am Acad Dermatol. 2017;76(4):676-682.

3. Arzberger E, Curiel-Lewandrowski C, Blum A, et al. Teledermoscopy in high-risk melanoma patients: A comparative study of face-to-face and teledermatology visits. Acta Derm Venereol. 2016;96(6):779-783.

4. Wu X, Oliveria SA, Yagerman S, et al. Feasibility and efficacy of patient-initiated mobile teledermoscopy for short-term monitoring of clinically atypical nevi. JAMA Dermatol. 2015;151(5):489-496.

5. Horsham C, Loescher LJ, Whiteman DC, Soyer HP, Janda M. Consumer acceptance of patient-performed mobile teledermoscopy for the early detection of melanoma. Br J Dermatol. 2016;175(6):1301-1310.

6. Esteva A, Kuprel B, Novoa RA, et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 2017;542(7639):115-118.