Average Error of Key Landmarks Within 1 mm
Comparable Accuracy With Manual Planning
Overall Virtual Surgical Planning Time Reduced From 17–18 Minutes to 7–9 Minutes

A research team from Hallym University’s Graduate School of Clinical Dentistry has reported that by utilizing artificial intelligence (AI), the planning time for three-dimensional virtual orthognathic surgery can be reduced by more than half. The AI-based method also demonstrated a similar level of accuracy in skeletal positioning when compared to actual surgical results, matching that of traditional manual approaches.

From the left: alumnus Sehoon Baek, department head Suhwan Byun, and graduate school dean Byeongeun Yang. Provided by Hallym University

From the left: alumnus Sehoon Baek, department head Suhwan Byun, and graduate school dean Byeongeun Yang. Provided by Hallym University

View original image

This study involved Sehoon Baek, an alumnus of the Department of Dental AI & Robotics at Hallym University’s Graduate School of Clinical Dentistry; department head Suh Whan Byun; and graduate school dean Byungun Yang. A professor from the Department of Oral and Maxillofacial Surgery at Karolinska University Hospital in Sweden also participated as a co-researcher.


The results received publication approval in September from 'Scientific Reports,' an SCIE-indexed international journal in the top 15% of its field, with a citation index of 4.9.


Orthognathic surgery corrects the positions of the upper and lower jaws to treat facial asymmetry or severe malocclusion. Since each patient’s jawbone shape and dental alignment are different, precise preoperative planning is required to determine the direction and extent of jaw movement for each case.


While 3D CT-based surgical planning has recently become standard, clinicians previously had to manually identify and mark key anatomical landmarks of the jawbones and teeth on the images.


The research team compared three different virtual surgical planning methods for orthognathic surgery patients: manual planning using conventional software, manual planning using the ON3D program, and the use of ON3D’s AI-powered automatic measurement function.


They then compared each virtual surgery plan with CT images taken one week after surgery, analyzing how closely the positions of key skeletal and dental points in the upper and lower jaws matched actual surgical outcomes.

Image explaining the reduction of orthognathic surgery planning time using AI. Provided by Hallym University

Image explaining the reduction of orthognathic surgery planning time using AI. Provided by Hallym University

View original image

The results showed that when comparing AI-based and manual planning using the same ON3D program, the differences between the virtual plan and actual surgical results at most major skeletal points were not statistically significant. At the primary reference points of the upper jaw, differences typically averaged around just 1 mm.


The most pronounced impact, however, was in time savings. While it took clinicians an average of about 10.5 minutes to mark key points on CT images and construct the 3D structures manually, the AI-based method reduced this to approximately 4.2 minutes. Overall virtual surgery planning time was also cut by more than half, from the previous 17–18 minutes down to 7–9 minutes.


The research team anticipates that automating these repetitive landmarking tasks with AI could ease clinicians’ workloads and significantly improve the efficiency of surgical planning. This technology is expected to be especially useful at institutions that handle a high volume of surgeries or require collaborative planning across departments.



Looking ahead, the team plans to conduct additional multicenter studies and long-term follow-up involving a larger cohort of patients to further validate the accuracy and safety of AI-based three-dimensional surgical planning.


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing