Multiparametric image cytometry of nevi and melanomas

Multiparametric image cytometry was applied to 10 examples each of malignant melanoma and common, Spitz, and dysplastic nevus. DNA index, area, and 21 parameters describing chromatin texture were measured for 50 nuclei in each lesion. Linear discriminant analysis was used to derive discriminant functions based on the measured parameters. The analysis demonstrated that chromatin texture provides more diagnostic information than either DNA index or nuclear area. The discriminant functions allowed 68% of nuclei to be accurately classified among the four groups, and allowed 37 of the 40 lesions to be accurately classified as nevus or melanoma.