Clinical diagnosis of pigmented lesions using digital epiluminescence microscopy. Grading protocol and atlas.

BACKGROUND AND DESIGN

Epiluminescence microscopy (ELM) is a clinical technique that permits in vivo visual inspection of pigmented anatomic structures of the epidermis, dermoepidermal junction, and papillary dermis. A protocol is proposed for systematic visual inspection of pigmented lesions. Seventy pigmented lesions were imaged with a digital ELM camera system. Images were visually inspected for eight "global" ELM features, 23 "local" ELM features, and 18 network features. An atlas of the most clinically significant ELM features is presented with pilot estimates of their sensitivity and specificity for detecting melanoma.

RESULTS

Preliminary data suggest that ELM features that may be most specific for melanoma include multicomponent pattern, nodular pattern, pseudopods, radial streaming, blue-gray areas, whitish veil (milky way), and sharp network margins. Epiluminescence microscopic features that may be most sensitive for melanoma include pigment dots, peripheral erythema, peripheral dark network patches, marked mean network irregularity, network line thickness variability, radial streaming, blue-gray areas, and whitish veil (milky way). Epiluminescence microscopic features that may be most sensitive for severe melanocyte atypia include pigment dots, peripheral erythema, hypopigmented network patches, peripheral dark network patches, marked mean network irregularity, and focal absence of network. In addition, features that may have a very high specificity for benign lesions include saccular pattern (suggests hemangioma), globular pattern (suggests a compound or dermal nevus), and multiple comedolike openings (suggests seborrheic keratosis).

CONCLUSIONS

Features most sensitive for severe atypia and melanoma could form the basis for a screening test for considering biopsy. Features most specific for melanoma then could be applied to further triage management of pigmented lesions that meet initial screening criteria. In addition, features with very high specificity for benign lesions may help develop ELM criteria to avoid unnecessary surgery.