Accuracy of digital models obtained by direct and indirect data capturing

With direct and indirect digitalisation, two access points to CAD/CAM-generated restorations are available. The aim of this study was to compare the accuracy of the single steps of both approaches by comparing construction datasets using a new methodology. Twelve test datasets were generated in vitro (1) with the Lava Chairside Oral Scanner (COS) (2) by digitizing polyether impressions (IMP) and (3) by scanning the referring gypsum cast by the Lava Scan ST laboratory scanner (ST) at a time. Using an inspection software, these datasets were superimposed by a best fit algorithm with the reference dataset (REF), gained from industrial computed tomography, and divergences were analysed. On the basis of average positive and negative deviations between test- and REF datasets, it could be shown that direct digitalisation accomplished the most accurate results (COS, 17 mu m/-13 mu m; SD +/- 19 mu m), followed by digitized polyether impression (IMP, 23 mu m/-22 mu m; SD +/- 31 mu m) and indirect digitalisation (ST, 36 mu m/-35 mu m; SD +/- 52 mu m). The mean absolute values of Euclidean distances showed the least values for COS (15 mu m; SD +/- 6 mu m), followed by IMP (23 mu m; SD +/- 9 mu m) and ST (36 mu m; SD +/- 7 mu m). The mean negative and mean absolute values of all groups were significantly different. Comparing the mean positive values of the groups, IMP and COS (p = 0.082) showed no significant difference, whereas ST and COS, and ST and IMP exhibited statistically significant differences. Within the limitations of this in vitro study, the direct digitalisation with Lava C.O.S. showed statistically significantly higher accuracy compared to the conventional procedure of impression taking and indirect digitalisation. Within the limitations of this study, the method of direct digitalisation seems to have the potential to improve the accuracy of impressions for four-unit FDPs.