Results of the 2016 International Skin Imaging Collaboration International Symposium on Biomedical Imaging challenge: Comparison of the accuracy of computer algorithms to dermatologists for the diagnosis of melanoma from dermoscopic images.

BACKGROUND: Computer vision may aid in melanoma detection. OBJECTIVE: We sought to compare melanoma diagnostic accuracy of computer algorithms to dermatologists using dermoscopic images. METHODS: We conducted a cross-sectional study using 100 randomly selected dermoscopic images (50 melanomas, 44 nevi, and 6 lentigines) from an international computer vision melanoma challenge dataset (n = 379), along with individual algorithm results from 25 teams. We used 5 methods (nonlearned and machine learning) to combine individual automated predictions into "fusion" algorithms. In a companion study, 8 dermatologists classified the lesions in the 100 images as either benign or malignant. RESULTS: The average sensitivity and specificity of dermatologists in classification was 82% and 59%. At 82% sensitivity, dermatologist specificity was similar to the top challenge algorithm (59% vs. 62%, P = .68) but lower than the best-performing fusion algorithm (59% vs. 76%, P = .02). Receiver operating characteristic area of the top fusion algorithm was greater than the mean receiver operating characteristic area of dermatologists (0.86 vs. 0.71, P = .001). LIMITATIONS: The dataset lacked the full spectrum of skin lesions encountered in clinical practice, particularly banal lesions. Readers and algorithms were not provided clinical data (eg, age or lesion history/symptoms). Results obtained using our study design cannot be extrapolated to clinical practice. CONCLUSION: Deep learning computer vision systems classified melanoma dermoscopy images with accuracy that exceeded some but not all dermatologists.

Digital imaging biomarkers feed machine learning for melanoma screening.

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.

In vivo intraoral reflectance confocal microscopy of an amalgam tattoo.

The majority of oral pigmentations are benign lesions such as nevi, melanotic macules, melanoacanthomas or amalgam tattoos. Conversely, mucosal melanomas are rare but often lethal; therefore, excluding oral melanomas in this setting is crucial. Reflectance confocal microscopy is a non-invasive, in vivo imaging system with cellular resolution that has been used to distinguish benign from malignant pigmented lesions in the skin, and more recently in the mucosa. However, lesions located posteriorly in the oral cavity are difficult to assess visually and difficult to biopsy due to their location. Herein we present a patient with previous multiple melanomas presenting with an oral amalgam tattoo in the buccal mucosa, which was imaged using an intraoral telescopic probe attached to a commercially available handheld RCM. In this case report we describe this novel probe, the first RCM description of an amalgam tattoo and we discuss its differences with the findings described in oral melanomas.

Variation in the depth of excision of melanoma: A survey of US physicians.

OBJECTIVE: To assess current practices and recommendations of US physicians regarding depth of excision for melanomas of varying histologic thicknesses. DESIGN: A 2-page, 13-question survey of depth of excision practices for the treatment of melanoma was developed and distributed. SETTING: Both private and academic settings. PARTICIPANTS: A total of 1184 US physicians (1000 dermatologists and 184 melanoma specialists) were sent the survey. The 184 melanoma specialists included dermatologists, oncologists, and surgeons working in pigmented lesion clinics. MAIN OUTCOME MEASURES: Depth of excision practices reported for melanomas of varying histologic thicknesses and comparison of treating physician groups. Results were tabulated, and descriptive frequencies were used to describe demographics and survey responses. RESULTS: The final study analysis included 498 completed surveys. The overall response rate was 45% (498 of 1097 [1184 total respondents - 87 ineligibles]). The response rate for the specialists was 63% (115 of 183 [184 total respondents - 1 ineligible]), and for nonspecialist dermatologists it was 43% (383 of 892 [1000 total respondents - 108 ineligibles]). Specialists were more likely to practice in an urban setting than were nonspecialist dermatologists (78% vs 46%) (P < .001). Fifty-eight percent of nonspecialist dermatologists reported more than 400 patient visits per month compared with only 16% of specialists (P < .001). While specialists reported fewer patient visits per month, 51% reported diagnosing over 20 invasive melanomas in the previous year compared with 11% of nonspecialist dermatologists. There was no significant difference in excision depth reported among the specialties for melanoma in situ (P = .15). For invasive melanoma, significant differences were observed among treating groups, with the greatest incongruence reported for thin invasive melanoma (<0.50 mm, P = .02; 0.50-0.75 mm, P < .001; and 0.76-1.00 mm, P < .001). Specialist nondermatologists consistently reported excising more deeply than specialist dermatologists and nonspecialist dermatologists. More specialist nondermatologists report excising to the fascia for thin invasive melanoma than do both specialist and nonspecialist dermatologists. For thicker melanomas (>1.00 mm), differences in excision depths among treating physician groups decreased: most physicians in each group reported excising to the fascia. CONCLUSIONS: There is considerable variation among physician groups with regard to depth of excision practices for the treatment of melanoma. Given the current lack of clinical data available, studies assessing depth of excision and patient outcomes are needed to better define our surgical management of melanoma.