Digital skin lesion analysis may help dermatologists classify skin lesions

A multispectral digital skin lesion analysis device (MSDSLA, MelaFind; MELA Sciences) may be useful in helping clinicians to classify clinically ambiguous lesions, detect early melanomas in young adults, and identify outlier lesions that should be biopsied, according to a collection of posters presented at MauiDerm 2014.

Melanoma can be particularly difficult to detect in young adults, according to one abstract submitted by study author Marcella Kollmann, M.D., of Munich, Germany. Early melanoma lesions can mimic benign lesions and may present a challenge, researchers from a second abstract report. The MSDSLA device collects multispectral data in 10 wavelengths of light to provide information about a lesion's morphology up to 2.5 mm into the skin. The device then characterizes information through use of a score where zero and higher is considered "high disorganization," and below zero is considered "low disorganization."

The first study investigated the use of the device in young adults. Dr. Kollmann reviewed data for 183 lesions in 136  young adults ages 16 to 25 who had been enrolled in a pivotal study, which has been published in Archives of Dermatology.

Also included was real-world data from two clinical cases: a 17-year-old female and a 24-year-old male, as well as six invasive melanomas in young adults who had participated in the pivotal trial and were randomly selected to be reviewed by 211 American and German dermatologists as part of a reader study.

The 17-year-old patient was clinically unsuspicious for more than three years. Use of the MSDSLA reported a 3.5 score and pathology results indicated melanoma in situ. The 24-year-old patient had a lesion that appeared within four months of excision and grew. He was clinically diagnosed with a Spitz/Reed Nevus. Use of the MSDSLA reported a score of 3.3 and pathology results indicated melanoma in situ.

Next: Analysis results 

More on melanoma detection

Airport screeners hold potential for cancer detection

'Chaos' algorithm simplifies biopsy decision

New test differentiates melanoma from benign lesions

Of the 183 lesions examined in the pivotal study, 13 were high-grade melanoma lesions. MSDSLA correctly considered 11 of these lesions as possessing "high disorganization."

Average false negative rates for the 211 American and German dermatologists for the six melanomas reviewed in the reader study were: 48.6 percent (0.17 mm median Breslow thickness), 57.3 percent (0.60 mm median Breslow thickness), 1.9 percent (0.32 mm median Breslow thickness), 76.2 percent (0.15 mm median Breslow thickness), 46.0 percent (0.70 mm median Breslow thickness), and 78.1 percent (0.28 mm median Breslow thickness).

"MSDSLA can be used to help dermatologists decide which lesions may be suspicious for melanoma by providing information about the structure of a lesion from under the skin," Dr. Kollmann reported in the abstract.

In a second abstract, researchers looked retrospectively at MSDSLA-computed classifier scores pigmented lesions relative to clinical characteristics and histopathology. The researchers noted a range of scores between -5.2 to 9.0. Average classifier scores for melanomas, high grade lesions, nonmelanoma skin cancers, and benign lesions were 3.5, 2.7, 2.6 and 1.7 respectively, further validating use of the device.

A third abstract investigated whether the MSDSLA scores can be useful in identifying benign nevi that should be further evaluated. Noting that a pivotal study indicated that, in patients with multiple nevi, scores for melanoma and high grade lesions tend to be higher than benign lesions, researchers included 10 patients with at least 200 nevi and analyzed MSDSLA score distributions for each patient and for all patients.

The researchers found significantly different score distributions among some patients. Individual thresholds ranged from 2.8 to 4.5 with 76 percent of lesions falling above the zero threshold used for the general population and 2.1 percent falling above the individual thresholds calculated for each patient. The authors concluded that the analysis of MSDSLA scores can be used to identify typical lesions on a patient and outliers, which should then be considered for biopsy.

More on melanoma detection

Airport screeners hold potential for cancer detection

'Chaos' algorithm simplifies biopsy decision

New test differentiates melanoma from benign lesions