Deep learning computer vision systems classified melanoma dermoscopy images with accuracy that exceeded some but not all dermatologists.

Results from this pilot study suggest that teledermoscopy is feasible and effective as a method for short-term monitoring of clinically atypical nevi. The implementation of teledermoscopy can potentially enhance patient convenience, optimize physician scheduling, and promote efficiency.

We developed an automated approach for generating quantitative image analysis metrics (imaging biomarkers) that are then analysed with a set of 13 machine learning algorithms to generate an overall risk score that is called a Q‐score. These methods were applied to a set of 120 “difficult” dermoscopy images of dysplastic nevi and melanomas that were subsequently excised/classified. This approach yielded 98% sensitivity and 36% specificity for melanoma detection, approaching sensitivity/specificity of expert lesion evaluation. Importantly, we found strong spectral dependence of many imaging biomarkers in blue or red colour channels, suggesting the need to optimize spectral evaluation of pigmented lesions.

Due to its particularly lethal nature and tendency to affect relatively young individuals, the timely diagnosis of melanoma remains of paramount importance for clinicians and their patients. Unfortunately, melanomas can mimic benign lesions that are overwhelmingly more common in the population than are melanomas, and misdiagnosis or delay in diagnosis of melanoma can occur. Misdiagnosis of melanoma serves as one of the most common causes for malpractice litigation brought against medical practitioners. In this review we describe seven clinical scenarios that represent challenges in melanoma diagnosis and discuss potential strategies for avoiding the errors that commonly give rise to those scenarios.

No abstract