[Clinical recognition of early forms of malignant melanoma]

The clinical diagnosis of malignant melanoma (MM) is based on the subjective evaluation of objective measurable parameters (criteria). The accuracy of melanoma diagnosis by dermatologists is only 75%. Particularly difficult is the diagnosis of precursors or early stages of MM. Therefore, we have studied on the one hand the intra- and interindividual reproducibility of the clinical diagnosis of pigmented lesions, and on the other hand the clinico-histopathological correlation. In addition, we have conducted a preliminary investigation designed to evaluate whether image analysis (objective and reproducible) could be used as an auxiliary instrument to differentiate between benign and malignant melanocytic lesions. In the clinical study, the intraindividual reproducibility of the combination of criteria was 69%. The interindividual reproducibility of single criteria even exhibited a range of up to 36%. Histologically "atypical/dysplastic" melanocytic lesions were considered to require excision as frequently as histologically regular melanocytic lesions. Using image analysis (single threshold segmentation, standard deviation of intensity distribution, ratio of area to circumference, Fourier analysis), we could show that it may be possible to differentiate between benign and malignant melanocytic lesions. Therefore, image analysis may be very helpful in determining the dignity of melanocytic lesions.