A study of three human dental populations using statistical shape modelling

Dental morphological measurements are typically either metric or non-metric. Non-metric measurements are qualitative and are used to identify the degree of expression of features, whilst metric approaches focus on measurements of geometric landmarks [1]. Geometric morphometrics combines landmark measurements with multi-variant statistical analysis techniques to compare populations. Statistical shape models (SSM’s) identify the geometric modes of variation of a database by resolving its principle components. SSM’s have been used to capture variation in anatomical populations including long bones and dentition [3]. SSM’s offer the opportunity to make full-field morphometric comparisons of discreet anatomical populations. The present study uses SSM to identify morphological characteristics of three populations of human mandibular canines.A digital (.stl) database of 27 mandibular canines was compiled from micro CT scanning (33m resolution, Nikon/Metris MMX scanner, mu-Vis Soton). The database consisted of three populations; modern human (MH, n=9), Anglo-Saxon (GC, n=13) and Roman (HQ, n=5). A binary material operator was assigned to each mesh node to distinguish between enamel and dentin/cementum. Each canine was aligned, registered to a baseline geometry and statistically modelled using principle component analysis (Fig.1).Synthetic geometries for each mode of variation were generated for standard deviations of the modal range. The CEJ was identified using material parameter thresholds. The modes of variation were characterised using visual observation and automated landmark identification/measurement algorithms. Finally, populations were grouped and their expression of modes of variation analysed for trends.Prominent characteristics captured by the modes of variation were as follows; mode 1 was associated with overall scaling (including crown bruxing) (Fig. 2), mode 2 with mesio-distal diameter, mode 3 with labiolingual thickness, mode 4 with crow/root length ratio and mode 5 with root thickness.Analysis of PC scores between population groups revealed the greatest inter-population variation in mode 1, corresponding to larger MH than GC and HQ populations respectively. Also observed was the similarity of the GC and HQ populations compared with MH for mode 3, suggesting that the MH specimens were relatively larger in the mesio-distally (Fig. 3). This study demonstrates the use of statistical shape modelling for morphometric measurement of populations, and identification of geometric characteristics in sub-populations. This technique has applications in osteoarchaeology and forensic medicine and may inform on disease development and treatment