For the s-CAIS group, Cone Beam Computed Tomography (CBCT) and full-arch optical scan were performed to provide digital information for implant planning software (coDiagnostix 9, Dental Wings GmbH, Chemnitz, Germany). The virtual implant was set on a three dimensional (3D) virtual jaws according to the prosthetically driven protocol by one postgraduate dentist and was confirmed by one experienced dentist. Static surgical template covering on occlusal part of 4 teeth anteroposteriorly was then fabricated via 3D printing machine (surgical guide resin, Form 2, Formlabs, Somerville, Massachusetts, USA). This surgical template would be used during surgery.

For the c-LIS group, a radiographic template (ORTHO Plast, prominent®, Chonburi, Thailand) was fabricated covering on occlusal part of 4 teeth anteroposteriorly according to diagnostic wax-up on the study model. A radiographic marker (gutta percha) was then filled in the created hole of the template and for used while taking CBCT image to verify marker position.

Next, a study model was scanned by laboratory surface scan (D900m, 3Shape, Copenhagen, Denmark). The STL file was imported to 3D printing devices and the resin model was fabricated (Dental LT clear resin, Form 2, Formlabs, Somerville, Massachusetts, USA).

Next, the same template that had been used for CBCT was used to place implant replicas in a resin model. The position of the implant replicas in the models were assumed as pre-operative planned implant position.

Three months following the implant placement, patients were called back to the implant clinic to record the actual implant position with the digital impression technique. Full mouth scans were done by using the intraoral scanner (Trios 3, 3Shape, Copenhagen, Denmark). The surface scans were then exported as an STL file and were imported to coDiagnostiX software. The "Treatment evaluation tool" function tool was used to measure the accuracy of the implant placement which measured the amount of deviation of the placed implant from the planned position. The outcomes were generated into three main parameters.

Patients were called back to record the actual implant position with the the intraoral scanner (Trios 3, 3Shape, Copenhagen, Denmark). The surface scans were then exported as an STL file and were imported to coDiagnostiX software. Those postoperative surface scans were merged with the preoperative surface scan in each group protocols. The "Treatment evaluation tool" function tool was used to measure the accuracy of the implant placement which measured the amount of deviation of the placed implant from the planned position. The outcomes were generated into three main parameters which were Angular deviation, Coronal global deviation and Apical global deviation.

Angular deviation = the amount of angle(°), in which the actually placed implant deviated from the virtually planned implant in 3D.

Coronal/ Apical global deviation = the amount of distance(mm), in which the actually placed implant deviated from the virtually planned implant at the coronal/ apical position in 3D.

Various factors influencing the accuracy of implant placement were also analyzed including the type of arch (Maxilla vs. Mandible), side of the arch (Left vs. Right), implant location (Premolar vs. Molar), implant diameter (WN vs. RN), implant length (8 vs. 10 mm), ridge morphology (Broad (≥8mm) vs. Narrow (˂8mm), cortical bone thickness and cortical interference. The thickness of the cortical bone was evaluated at the center of the implant after the planned implant position was completely set in the coDiagnostiX program. If any parts of the planned implant involving against any cortical bone walls, it would count as having a cortical interference.