Outline of procedures for data mining of morphology.

(a) Seven main steps from 3D data acquisition and processing, feature extraction, data mining procedures (classifiers and feature selection) to possible applications of the method. (b–d) illustrations of feature extraction methods. (b) Section areas and section convolutions. (i) Tooth (upper molar of Rhinolophus blasii) is divided into 10 equal sections perpendicular to the z-axis. Upper and lower bounds (relative to the zApex) for every second section are given along the z-axis. (ii) Occlusal view with every second section highlighted in red, with areas and convolutions for each section. (c) Orientation patch count (OPC). (i) Surface of the tooth (upper molar of Felis silvestris) is grouped into surface vertices according to their orientation in the xy-plane (ii). (iii) Vertices are further grouped according to their 4-cell connectivity followed by (iv) exclusion of small patches. (v) The resulting OPC value is the final number of patches. (d) The effect of surface folding and elongation on surface relief. Relief is calculated by dividing the 3D surface area by its 2D projected area. A flat, unspecialized surface (i) has a relief of 1. If the surface is folded (ii), such as in Otomys irroratus, or elongated (iii), such as in Felis silvestris, its relief increases.