Aim: Dental research involves variations between actual and reference datasets of master models to determine the metric accuracy through transfer accuracy tests. Various methods of measurement are used to analyze the results, which are often subjected to direct comparisons. Hence, the aim of the present study was to analyze the influence and effect on results of different methods of digital data analysis, being coordinate-based analysis (CBA) and best-fit superimposition analysis.
Keywords: dimensional measurement accuracy, accuracy, trueness, precision, intraoral scanner, digital dentistry, implant impression, best-fit analysis
Materials and methods: A model with four implants and a reference cuboid was digitized through computed tomography (CT), which served as the master model. Ten implant impressions were made using a Trios (3Shape) intraoral scanner, and three different scan bodies (nt-trading, Kulzer, and Medentika) were used. The deviations between the master model and the digital impressions were analyzed using CBA and best-fit superimposition analysis. Statistical analysis was performed using SPSS 25.
Results: The deviations in the CBA and best-fit superimposition analysis ranged from 0.088 ± 0.012 mm (mean ± SE; Medentika, 14) to 0.199 ± 0.021 mm (Kulzer, 26), and from 0.042 ± 0.010 mm (Medentika, 16) to 0.074 ± 0.006 mm (Kulzer, 16), respectively. Significant differences were observed between the implant positions in the CBA and the digital measurements at each implant position, whereas the best-fit analysis showed no significant difference between the scan bodies and implant positions.
Conclusion: CBA displays an advantage over best-fit superimposition analysis in the detection of possible influencing factors for primarily scientific purposes. However, a global analysis and visualization of angles and torsions is difficult, for which a best-fit evaluation is needed. However, a best-fit analysis better represents the clinical try-in. It is associated with the risk that possible disturbing factors and resulting errors might be leveled out and their identification camouflaged.