Careers  |  Sign In  |  Register

Computerized image assessment system helps detect melanoma

Reuters Health • The Doctor's Channel Daily Newscast

NEW YORK (Reuters Health) – The evaluation of pigmented lesions for early detection of melanoma is aided by the use of MelaFind, a non-invasive automated vision-computer system, according to a report in the Archives of Dermatology published online October 18.

“Early melanoma may be difficult to differentiate from many benign simulants,” write the investigators, led by Dr. Dina Gutkowicz-Krusin with MELA Sciences Inc., in Irvington, New York. They explain that MelaFind acquires digital images of pigmented lesions in 10 spectral bands, and uses automatic image analysis and statistical pattern recognition to help identify possible melanomas.

The team evaluated the performance of MelaFind in terms of sensitivity and specificity for recommending a biopsy by classifying pigmented lesions as “melanoma cannot be ruled out” or “not melanoma,” with dermatopathology as the reference standard

For the study, 1383 patients with 1831 pigmented lesions were registered and 1632 lesions could be evaluated. A total of 127 were confirmed as melanoma.

The sensitivity of MelaFind for biopsy recommendation was 98.4%, while the specificity was just 9.9%. However, that was still better than the 3.7% specificity of clinicians in the study.

In the trial, 10.8 biopsies were performed to identify 1 melanoma. As the authors point out, “If all atypical lesions were to be biopsied to rule out melanoma, the biopsy ratio (number of false-positive biopsy findings per true-positive biopsy finding) of about 200:1 would be very high.”

They conclude, “MelaFind is a safe and effective tool to help identify pigmented lesions to be considered for biopsy to rule out melanoma.”

The study was supported by MELA Sciences, Inc., according to a disclosure statement.


The Performance of MelaFind

Arch Dermatol 2010.