Checking consecutively sampled routine sections of 206 melanocytic lesions with a maximum vertical diameter of at least 1 mm (133 benign dermal nevi, 20 Spitz's nevi, 53 primary malignant melanomas), we measured the morphometric features of at least 60 nuclei each from the superficial and the deep dermal tumor portion using a computer-assisted interactive image analysis system. Furthermore we calculated the so-called maturation parameter (MP) in each case as the ratio of the mean nuclear area in the deep portion and the superficial portion. When we compared the results with those obtained in a training set, we found that the lowest evidence for the discrimination of benign and malignant melanocytic lesions resulted from the application of the mean values of the nuclear area in the superficial layer (efficiency = 62.1%). The efficiency was higher when we used the mean values of the nuclear area in the deep layer (96.1%) and the maturation parameter (85.4%). By applying the mean nuclear area in the deep portion and the maturation parameter simultaneously, we gained the highest efficiency, specificity, and sensitivity for the distinction between benign dermal nevi and malignant melanomas (0.968, 0.955, 1) as well as for the distinction between Spitz's nevi and malignant melanomas (0.986, 0.950, 1). Our study shows that morphometry provides reliable diagnostic results in routinely sampled melanocytic skin tumors.
