It is estimated that over 12 million of people will die in 2030 due to skin cancer. This fact (and the low number of experts in skin cancer) become relevant for computer aided diagnosis (CAD) and its use for evaluating malignant skin lesions. The main goal of these systems is to detect candidate malignant lesions at early stages. A CAD system is integrated by four blocks: image processing, skin lesion segmentation, region description, and region classification. Image processing is used to isolate and prepare the digital region of interest. Region segmentation makes use of colour and border detection algorithms to separate the skin region from the background. After that, the third block computes several shape descriptors. Finally, the classification step is used to output the desired diagnosis. The general approach used by a CAD system consists in describing the skin lesion by means of a set of textural and geometrical shape features known as the ABCD rule (asymmetry, border, colour and diametre). In this paper we propose to measure the A and B features, which are considered to be the two main characteristics to accurately diagnose skin cancer, through the Normalized E Factor (NEF). Excellent results of classifications are presented.