A critical assessment of the potential for Structure-from-Motion photogrammetry to produce high fidelity 3D dental models.

Abstract

The recent proliferation of methods of 3D model generation has enabled the development of new approaches to the analysis of dental form, function and wear. This article assesses whether Structure-from-motion (SfM) photogrammetry is capable of producing virtual 3D models of teeth of adequate quality for assessing fine scale surface details, such as dental macrowear patterns. Reference models were generated using a high resolution structured light scanner to assess the accuracy of the photogrammetric models generated. Dental gypsum models of the molar teeth of human individuals from St. Michael's Litten, Chichester, Post-medieval assemblage (n = 17) were used for 3D model generation. Photogrammetry was performed using Agisoft Metashape and reference 3D models were generated using a GOM ATOS 80 scanner. Focus stacking was explored as a method of enhancing 3D model detail. Differences between the photogrammetric and reference models were assessed using CloudCompare and the quality of the surface detail was examined quantitatively using Occlusal Fingerprint Analysis. Photogrammetric model generation was highly replicable and the tooth models produced closely approximated the overall geometry of those derived from the structured light scanner. Dental wear facet area measurements on the photogrammetric models differed significantly, however, from those derived from the structured light scanning reference models. Photogrammetry can create virtual dental models from which crude quantitative size and shape data can be obtained. Finer scale surface details are not accurately reproduced on SfM models using the methods outlined in the current article due to high levels of surface noise.