Comparison of Accuracy of Maxilla Between Virtual and Conventional Surgical Planning in Bimaxillary Orthognathic Surgery

Materials and Methods Study Design This study was conducted as a prospective, single-center, randomized, and blinded case-controlled trial. The research protocol was reviewed and approved by the Ethics Committee of the University of Medicine and Pharmacy at Ho Chi Minh City (Approval No. 647/HĐĐĐ-ĐHYD). All participants were fully informed of their treatment options and provided written informed consent prior to enrollment. The study adhered to ethical guidelines for clinical research, ensuring the protection of participants' rights, safety, and confidentiality.

Study Subjects The study was carried out at the National Hospital of Odonto-Stomatology in Ho Chi Minh City between August 2023 and February 2025. Eligible participants were aged 18 to 30 years and diagnosed with malocclusion requiring bimaxillary orthognathic surgery.

Exclusion criteria included:

1. Cleft lip and/or palate or craniofacial syndromes
2. Facial deformities resulting from trauma, tumors, or iatrogenic causes
3. Temporomandibular joint disorders
4. Previous history of orthognathic surgery
5. Indication for multipiece Le Fort I osteotomy A total of 20 patients met the inclusion criteria and agreed to participate. All patients had completed presurgical orthodontic treatment prior to surgery.

Randomization and Blinding Each patient underwent both two-dimensional (2D) lateral cephalometric analysis and three-dimensional (3D) virtual surgical planning preoperatively. For each case, both a conventional surgical splint and a 3D-printed splint were fabricated.

Randomization was performed intraoperatively by an operating room nurse through a simple draw to determine which splint would be used. Allocation was recorded accordingly. The surgical and research teams were blinded to group assignment, with participants identified only as "Group 1" or "Group 2." Group identities (VSP or CSP) were only disclosed after data collection and analysis were complete.

Patients were assigned to two groups:

• Test group (VSP): 3D imaging, virtual osteotomy simulations, and 3D-printed splints
• Control group (CSP): 3D imaging, virtual osteotomy simulations, model surgery, and hand-fabricated resin splints Methods Preoperative Examination and Surgical Planning All patients underwent a full preoperative assessment, including clinical examination, radiographic imaging (panoramic, lateral and posteroanterior cephalograms, and CT scans), 2D cephalometric analysis, intraoral and facial photography, and dental impressions.

Facebow registration and intermaxillary relationship assessment were performed, followed by mounting on a semi-adjustable articulator using the standard Frankfort horizontal plane. Final occlusion was determined by the orthodontist. Dental models were digitized using the Autoscan-DS-EX Pro (Shining 3D) scanner.

Surgical planning was initially based on clinical findings and 2D cephalometric analysis using WebCeph software. Parameters included maxillary dental midline, anteroposterior and vertical positioning of maxillary incisors, and occlusal plane canting determined by the canine and first molar reference points.

3D Simulation

A 3D virtual skull model was constructed using anatomical reference planes for consistent orientation:

• Horizontal plane: through Nasion, parallel to the Frankfort horizontal plane
• Midsagittal plane: through Nasion and Basion, perpendicular to the horizontal plane
• Coronal plane: through Nasion, perpendicular to both horizontal and midsagittal planes The mandible was repositioned into the final occlusion, and the maxillomandibular complex was aligned according to P2D (Planning 2D) parameters derived from 2D analysis. Maxillary landmarks (A point, ANS, U1, U1L, U1R, U3L, U3R, U6L, U6R) were recorded.

Next, refinements were applied using VSP to adjust occlusal plane canting, midline deviation, yaw rotation, and anteroposterior and vertical positioning based on Vietnamese normative standards. These refined landmark positions were recorded as P3D (Planning 3D).

Comparison Between P2D and P3D Differences in landmark movements between P2D and P3D (A point, ANS, U1, U1L, U1R, U3L, U3R, U6L, U6R) were analyzed to assess the discrepancies between 2D and 3D planning methods.

Group Allocation and Splint Fabrication

After determining the final maxillary and mandibular positions, surgical splints were fabricated as follows:

• CSP Group (Conventional Resin Occlusal Splint): Splints were manually fabricated based on P3D values. Landmarks were marked, and the maxillary model was segmented and repositioned to approximate the P3D position based on anterior-posterior and superior-inferior displacements. The actual repositioning distances of U1L, U1R, U3L, U3R, U6L, and U6R were recorded as PCM (Planning Cast Model).
• VSP Group (Virtual 3D-Printed Splint): Surgical splints were digitally generated using Dolphin software and printed using a Formlabs 3D printer.

Surgical Procedures All surgeries were performed by the same experienced maxillofacial surgeon using a standardized maxilla-first approach.

• Maxillary osteotomy was performed first (LeFort I), and the maxilla was positioned using the intermediate splint, with position verified by simulation images. Rigid fixation was performed using four miniplates and screws.
• Mandibular osteotomy followed (BSSO technique), and repositioning was guided by the final splint. Fixation was achieved with two miniplates per side.
• Final verification of skeletal alignment and occlusion was conducted before layered soft tissue closure.
• Intermaxillary fixation was not routinely used, but light elastics were applied as needed.

Postoperative Skeletal Accuracy Analysis Two weeks post-surgery, all patients underwent CT imaging in occlusion with the final splint and before initiation of postoperative orthodontic treatment.

Pre- and postoperative scans were superimposed using voxel-based registration (Invivo 7.0, Anatomage, CA), aligned to non-surgical cranial reference areas using the same reference planes:

• X-axis: medio-lateral (positive = left)
• Y-axis: antero-posterior (positive = forward)
• Z-axis: superior-inferior (positive = downward) Landmark movements (A point, ANS, U1, U1L, U1R, U3L, U3R, U6L, U6R) were recorded as Actual values.

Surgical accuracy was evaluated through:

• Intra-group comparisons: P3D vs. Actual positions for each landmark
• Inter-group comparisons: Discrepancies between VSP and CSP groups Measurement Methods for Accuracy Assessment The intraclass correlation coefficient (ICC) was used to assess measurement reliability. Ten patients were randomly selected and remeasured after two weeks.

Statistical Analysis

• Normality testing was performed for all variables.
• If normally distributed, paired t-tests were used for P3D vs. Actual, and independent t-tests for CSP vs. VSP comparisons.
• If non-normally distributed, Wilcoxon signed-rank and Mann-Whitney U tests were applied.
• A p-value of < 0.05 was considered statistically significant.