Effect of simulated intraoral variables on the accuracy of a photogrammetric imaging technique for complete-arch implant prostheses

Abstract

Statement of problemConventional impression techniques to obtain a definitive cast for a complete-arch implant-supported prosthesis are technique-sensitive and time-consuming. Direct optical recording with a camera could offer an alternative to conventional impression making. PurposeThe purpose of this in vitro study was to test a novel intraoral image capture protocol to obtain 3-dimensional (3D) implant spatial measurement data under simulated oral conditions of vertical opening and lip retraction. Material and methodsA mannequin was assembled simulating the intraoral conditions of a patient having an edentulous mandible with 5 interforaminal implants. Simulated mouth openings with 2 interincisal openings (35 mm and 55 mm) and 3 lip retractions (55 mm, 75 mm, and 85 mm) were evaluated to record the implant positions. The 3D spatial orientations of implant replicas embedded in the reference model were measured using a coordinate measuring machine (CMM) (control). Five definitive casts were made with a splinted conventional impression technique of the reference model. The positions of the implant replicas for each of the 5 casts were measured with a Nobel Procera Scanner (conventional digital method). For the prototype, optical targets were secured to the implant replicas, and 3 sets of 12 images each were recorded for the photogrammetric process of 6 groups of retractions and openings using a digital camera and a standardized image capture protocol. Dimensional data were imported into photogrammetry software (photogrammetry method). The calculated and/or measured precision and accuracy of the implant positions in 3D space for the 6 groups were compared with 1-way ANOVA with an F-test (α=.05). ResultsThe precision (standard error [SE] of measurement) for CMM was 3.9 μm (95% confidence interval [CI] 2.7 to 7.1 μm). For the conventional impression method, the SE of measurement was 17.2 μm (95% CI 10.3 to 49.4 μm). For photogrammetry, a grand mean was calculated for groups MinR-AvgO, MinR-MaxO, AvgR-AvgO, and MaxR-AvgO obtaining a value of 26.8 μm (95% CI 18.1 to 51.4 μm). The overall linear measurement error for accurately locating the top center points (TCP) followed a similar pattern as for precision. CMM (coordinate measurement machine) measurement represents the nonclinical gold standard, with an average error TCP distance of 4.6 μm (95% CI 3.5 to 6 μm). All photogrammetry groups presented an accuracy that ranged from 63 μm (SD 17.6) to 47 μm (SD 9.2). The grand mean of accuracy was calculated as 55.2 μm (95% CI 8.8 to 130.8 μm). ConclusionsThe CMM group (control) demonstrated the highest levels of accuracy and precision. Most of the groups with the photogrammetric method were statistically similar to the conventional group except for groups AvgR-MaxO and MaxR-MaxO, which represented maximum opening with average retraction and maximum opening with maximum retraction.