Automatic virtual reduction of unilateral zygomatic fractures based on ICP algorithm: A preliminary study - 08/01/25

Doi : 10.1016/j.jormas.2025.102220

Runqi Liu ^a,^b,^c,^d,^{^{1
Runqi Liu and Bimeng Jie contributed equally to this work and share first authorship.}}, Bimeng Jie ^a,^b,^c,^d,^{^{1
Runqi Liu and Bimeng Jie contributed equally to this work and share first authorship.}}, Yanhang Tong ^a,^b,^c,^d, Junchen Wang ^e, Yang He ^a,^b,^c,^d,^⁎
^a Department of Oral and Maxillofacial Surgery, Peking University School and Hospital of Stomatology, PR China
^b National Engineering Laboratory for Digital and Material Technology of Stomatology, PR China
^c Beijing Key Laboratory of Digital Stomatology, PR China
^d National Clinical Research Center for Oral Diseases, Beijing, PR China
^e School of Mechanical Engineering and Automation, Beihang University, Beijing, China

^⁎Corresponding author at: Department of Oral and Maxillofacial Surgery, Peking University School and Hospital of Stomatology, 22 Zhongguancun South Road, Beijing 100081, PR China.Department of Oral and Maxillofacial SurgeryPeking University School and Hospital of Stomatology22 Zhongguancun South RoadBeijing100081PR China

Sous presse. Épreuves corrigées par l'auteur. Disponible en ligne depuis le Wednesday 08 January 2025

Abstract

Objective

To establish an automatic reduction method for unilateral zygomatic fractures based on Iterative Closes Point (ICP) algorithm.

Material and methods

60 patients with unilateral type B zygomatic fractures were included. After acquiring CT images, zygomatic fragments were segmented using self-developed software MICSys. Mid-Sagittal-Plane (MSP) was manually defined using anatomical skull landmarks. Surface of zygoma on the healthy side was then “mirrored” according to MSP. Referring to mirror image, the fragments were reduced by both automatic and manual methods. In automatic group, fragments were registered onto mirror images by ICP algorithm in MICSys. In manual group, an experienced maxillofacial surgeon translated and rotated fragments until coincided with mirror images. Operating time of each group was recorded. RMSE between reduced fragment and mirror image was calculated to evaluate accuracy. Operating time and accuracy between the two groups were compared using T-test.

Results

Virtual bone reduction was conducted for all 60 patients by the two methods. Operating time of automatic group and manual group were 3.06 ± 1.93 s and 65.45 ± 32.19 s, with significant difference (P < 0.0001). RMSE of automatic group and manual group were 1.94 ± 0.59 mm and 2.33 ± 0.57 mm, with significant difference (P < 0.0001).