A hand-held, computerized imaging device identified 98% of melanomas in a set of suspicious pigmented skin lesions, according to results published online in the Archives of Dermatology.

The study findings were also published online on Skin & Allergy News Digital Network on May 6, 2010 (AAD: Digital Dermoscopy Device Detects Melanoma).

Results of the automated device – MelaFind from MELA Sciences Inc. – were compared with dermatologists who examined a subset of lesions from the same group, for a 78% identification rate. On lesions that had been previously biopsied to rule out melanoma, MelaFind's average specificity was 10%, significantly better than the 3.7% rate of dermatologists.

The device's 98% sensitivity rating for malignant melanoma was significantly better than that of dermatologists, who showed a wide range of variability about which lesions would have been recommended for biopsy and which relegated to observation, Dr. Gary Monheit and his colleagues reported (Arch. Dermatol. 2010 [doi:10.1001/archdermatol.2010.302]).

The study compared MelaFind's performance on a group of 1,632 lesions with a reading study of 50 lesions randomly extracted from the same set. All of the lesions had been biopsied by the patients’ attending dermatologists.

Lesion data, including the clinical overview and close-up and dermatoscopy images, were provided to 39 dermatologists, who were blinded to the biopsy results. The dermatologists evaluated which lesions they believed should be biopsied as probable melanoma and which should be followed clinically, wrote Dr. Monheit, a dermatologist in private practice in Birmingham, Ala., and his coauthors. The lesion samples came from a group of 1,257 patients whose mean age was 46 years. Most (98%) were white.

The MelaFind study's primary end point was the device’s sensitivity and specificity for lesions classified before biopsy as "melanoma cannot be ruled out" or as "not melanoma." After reviewing each lesion, MelaFind generated one of two conclusions: positive (biopsy) or negative (follow clinically).

Pigmented nevi made up the bulk of the lesion group (77%; 1,258), with 61% being low-grade dysplasia. Another 15% of the lesions were nonmelanocytic. Melanomas accounted for 8% of the group (127), with 57 in situ and 70 invasive; however, the invasive lesions were mostly thin, with a median thickness of 0.36 mm. Only two were relatively thick (1.0 mm and 1.2 mm). "Thus," the investigators noted, "almost all melanomas in this trial were early lesions that are difficult to differentiate from benign simulants."

Other components of the group included keratoses (119), lentigos (76), pigmented basal and squamous cell carcinomas (33), and other lesions (14).

In the reader study, the average biopsy sensitivity of the 39 dermatologists was 78%. However, the authors noted, the inter-reader variability was high. "Only 5 of the 25 melanomas would have been biopsied by all readers, and different readers missed different melanomas." Since all of the lesions had previously been biopsied, the paper did not compare clinician and MelaFind biopsy ratios. However, the authors noted, historical biopsy ratios among dermatologists are about 8:1 for the general population and up to 47:1 for high-risk patients.

The authors concluded that the device could be a valuable tool in accurately assessing pigmented lesions – an area that is largely dependent on clinicians’ individual observations and conclusions.

In a statement from MELA Sciences, coauthor Dr. Kenneth Gross stressed the need for an objective tool in the biopsy decision-making process.

"The pilot reader study found that dermatologists do miss early melanomas," said Dr. Gross, who practices surgical dermatology in San Diego. "Only 20% of the melanomas in the pilot reader study would have been biopsied by all readers and different readers missed different melanomas. Thus, even though all lesions in the clinical trial were biopsied by the examining dermatologists, many of the melanomas would not have been biopsied by other dermatologists, underscoring the need for an objective tool to aid in the decision to biopsy. We believe that the results of the pivotal trial and the pilot reader study underscore the clinical utility of MelaFind as an objective tool to aid clinicians in the detection of early melanoma."

The Food and Drug Administration will consider MelaFind on Nov. 18, during the General and Plastic Surgery Devices Panel of the Medical Devices Advisory Committee meeting. The committee will make recommendations and vote on data related to the device's premarket approval.

According to the company Web site, MelaFind is a hand-held device that acquires and displays multispectral digital images from blue to near-infrared, for pigmented skin lesions. The computerized device uses automatic image analysis and statistical pattern recognition to help identify lesions that should be considered for biopsy to rule out melanoma. The device consists of an illuminator that produces light in 10 wavelengths, a lens system that creates the images, a light sensor, and an image processor that identifies discrete characteristics of each picture.

MELA Sciences sponsored the study. Dr. Monheit and 7 of the other 15 investigators are on the MELA Sciences Advisory Committee. Dr. Gross did not disclose any financial relationship with the company.

MelaFind Device Surpassed Dermatologists in Identifying Melanoma