Objective: The aim of this study is to compare the accuracy of conventional and digital impression techniques used at the implant and abutment levels.

Materials-Methods: Eight implants were inserted at different angles and distances into the polyurethane edentulous mandible model. Angled and straight multi-unit abutments are placed on the implants to correct the angles at an optimum level. Scan bodies were fixed on the abutments and a 3-dimensional (3D) reference model was obtained by scanning the polyurethane model with an optical scanner. Then, ten abutment level digital impressions were made with 3Shape Trios 3 and STL data were obtained. A verification jig was produced to record the angle and the hexagonal positions of the multi-unit abutments, then the abutments were removed, and open tray impression copings were fixed to the implants. First, without splinting and then by splinting the copings, twenty open tray conventional impressions were made (n=10). Master casts were obtained for each conventional impression and then multi-unit abutments were fixed by placing them on the casts with the help of the verification jig. Then, scan bodies were placed on multi-unit abutments for each master cast and 3D models were obtained with a digital laboratory scanner. STL files of these 3 different impression groups were transferred to a reverse engineering program to measure distance and angle deviations from the 3D reference model. Data were analyzed in IBM SPSS V23. The suitability of the data for normal distribution was examined with the Shapiro Wilk test; one- way analysis of variance test was used to compare normally distributed data according to three groups, and multiple comparisons were examined with the Tukey test. The significance level was taken as p<0.05.

Results: No statistically significant difference was observed between the digital (243.12 ±207.58 μm) and splinted (288.74 ±222.39 μm), non-splinted (311.62 ±183.84 μm) conventional groups in terms of distance deviations (p=0.82). In terms of angular deviation, the digital group (0.37° ±0.18°), was statistically significantly superior to splinted (0.59° ±0.12°) and non-splinted (0.58° ±0.14°) conventional impression groups (p=0.02).

Conclusion: In full-arch implant treatments with severe angulations, high deviations can be observed in both digital and conventional impressions, which can compromise the passive fit of the restoration to be produced. However, thanks to intraoral scanners, better impressions can be made, especially by minimizing angular deviations.