Reference model fabrication
The reference model was fabricated with type IV dental stone (NEW FUJIROCK IMP, GC, Tokyo, Japan) and implant replicas (Abutment Replica Multi-unit Brånemark System RP, Nobel Biocare, Kloten, Switzerland), which emulated a completely edentulous maxilla with implants (NobelSpeedy Groovy, Nobel Biocare, Kloten, Switzerland) and abutments (Multi-unit Abutment Brånemark System RP, Nobel Biocare, Kloten, Switzerland) at positions #15, #12, #22, and #25. All digital impressions were made by a dentist having more than 3 years of experience with digital impressions in routine practice.
Fabrication of assistive device
Scan bodies (Position Locator Multiple Nobel Biocare Multi-unit Abutment, Nobel Biocare, Kloten, Switzerland) were connected to the implant analogs in the reference model (Fig. 1a). The model was scanned using a dental laboratory scanner (D810, 3Shape, Copenhagen, Denmark, D810). On this surface data, the assistive device was designed using a CAD software (3Shape, Copenhagen, Denmark).
The Type 1 assistive device connected only the neighboring scan bodies, namely #12-15, #12-22, and #22-25, respectively (Fig.1b).
The Type 2 assistive device connected the neighboring scan bodies, similar to the Type 1 device. However, it also had a connecting bar between the two posterior implants (#15-25), and two perpendicular branches from this bar extended towards the two anterior implants (Fig. 1c). These two types of designs were input to a CAM device (250i, imes-icore GmbH) and milled from a Polymethyl methacrylate disk (M-PM disk, Shofu, Tokyo, Japan).
Impression making
Conventional impression
Conventional silicone impressions were made from the reference model using impression copings and an open tray at 24 °C. Impression copings were connected using a 2.35 mm diameter cobalt–chromium metal wire and fixed by self-curing acrylic resin (FIXPEED, GC, Tokyo, Japan). Twenty-four hours after making the impression, implant analogs were connected to the impression copings on impressions, followed by fabrication of the plaster models. This procedure was repeated and five plaster models were prepared. To acquire the surface data of these models, scan bodies were connected to the implant analogs, and titanium oxide powder (CEREC Opti Spray, Sirona, Long Island City, NY, USA) was sprayed on the surface to inhibit light reflection. They were then scanned using a dental laboratory scanner (D810, 3Shape, Copenhagen, Denmark) and five STL datasets of the conventional impression group were acquired (hereon referred to as “CON”).
Digital impression
Three different IOSs were evaluated in this study: Primescan (PS; Dentsply Sirona, USA), 3M True Definition scanner (TDS; 3M ESPE, Seefeld, Germany), and TRIOS scanner 3 (TR; 3Shape, Copenhagen, Denmark). When using TDS and PS, titanium dioxide powder was sprayed on the reference model according to the manufacturers’ recommendations. The digital impression was made according to the manufacturers’ instructions, except for the scan path, which was kept consistent irrespective of the type of IOS used.
This procedure was conducted for models without an assistive device, or “Type 0,” and on models with “Type 1” and “Type 2” assistive devices. The assistive devices were fixed to scan bodies with self-curing acrylic resin (UNIFAST III, GC) to prevent dislodgement by IOSs during the scan.
For Type 0 and Type 1, the digital impression started from the bucco-occlusal surface of the #15 scan body, continued along with the bucco-occlusal surface on the alveolar ridge until the #25 scan body, and turned to the palatal and the palato-occlusal surface of the alveolar ridge; scan bodies were scanned continuously until the #15 scan body (Fig. 2a).
For the Type 2 assistive device, branches running over the palate were scanned after scanning along the arch as for Types 0 and 1 (Fig. 2b). Although the operator tried to perform a continuous scan, rescanning was done to complement the missing dataset captured during the first scan. This procedure was repeated five times for each IOS, and all the captured images were converted to the STL format.
Conventional plaster model via verification jigs
Five verification jigs made of cobalt–chromium alloy were fabricated to fit the reference model. The impression copings were connected to the reference model, and the verification jig was fixed with a self-curing acrylic resin. The verification jig was removed after 30 min, and the laboratory analogs to the impression copings were connected to the impression coping, and the verification plaster models were fabricated, which are commonly considered the most accurate three-dimensional reference in the conventional analogue workflow. The scan bodies were connected to these verification models and scanned using a non-contact 3D measuring instrument (COMET5, Stanbekeley, Germany; hereon referred to as “VJ”) with a reported precision of 1.5 μm.
Data analysis and evaluation of precision
The overall workflow of this study is presented in Fig. 3. The STL datasets obtained from each group were imported into the measurement software (PolyWorks Inspector, PolyWorks Japan, Tokyo, Japan). Subsequently, only the surface data of the four scan bodies were retained, and the other data were removed.
To evaluate the precision, all combinations of the five STL datasets obtained from each IOS were superimposed using the least-squares best-fit method. Absolute values of the closest distance between the pairs of polygon data were computed for all the surfaces and averaged to determine the surface deviation of the two surface data [6, 9, 16]. To qualitatively evaluate the surface deviation, a color-coded gradient was generated for each situation.
Statistical analysis
Two types of statistical analyses were performed: analysis 1 compared the type of IOS (PS, TR, and TDS) and types of assistive devices (Type 0, 1, and 2) in relation to their effects on digital impression precision as evaluated by the average deviation (surface deviation) for each situation. Analysis 2 compared the average surface deviation between CON, VJ, and the digital impression group that was the most precise in Analysis 1.
For analysis 1, a two-way analysis of variance was performed, and for analysis 2 a one-way analysis of variance was performed. Tukey's multiple comparison test was used for the post-hoc tests. All statistical analyses software (JMP, SAS Institute Japan, Tokyo, Japan). The level of statistical significance was set at 5%.