Wei-Fa Yang1, Yu-Xiong Su2. 1. Division of Oral and Maxillofacial Surgery, Faculty of Dentistry, The University of Hong Kong, Hong Kong Special Administrative Region. 2. Division of Oral and Maxillofacial Surgery, Faculty of Dentistry, The University of Hong Kong, Hong Kong Special Administrative Region. Electronic address: richsu@hku.hk.
Abstract
BACKGROUND: The image segmentation of skull CT is the cornerstone for the computer-assisted craniomaxillofacial surgery in multiple aspects. This study aims to introduce an AI-enabled automatic segmentation and propose its prospect in facilitating the computer-assisted surgery. METHODS: Three patients enrolled in a clinical trial of computer-assisted craniomaxillofacial surgery were randomly selected for this study. The preoperative helical CT scans of the head and neck region were subjected to the AI-enabled automatic segmentation in Mimics Viewer. The performance of AI segmentation was evaluated based on the requirements of computer-assisted surgery. RESULTS: All three patients were successfully segmented by the AI-enabled automatic segmentation. The performance of AI segmentation was excellent regarding key anatomical structures. The overall quality of bone surface was satisfying. The median DICE coefficient was 92.4% for the maxilla, and 94.9% for the mandible, which fulfilled the requirements of computer-assisted craniomaxillofacial surgery. CONCLUSIONS: The AI-enabled automatic segmentation could facilitate the preoperative virtual planning and postoperative outcome verification, which formed a feedback loop to enhance the current workflow of computer-assisted surgery. More studies are warranted to confirm the robustness of AI segmentation with more cases.
BACKGROUND: The image segmentation of skull CT is the cornerstone for the computer-assisted craniomaxillofacial surgery in multiple aspects. This study aims to introduce an AI-enabled automatic segmentation and propose its prospect in facilitating the computer-assisted surgery. METHODS: Three patients enrolled in a clinical trial of computer-assisted craniomaxillofacial surgery were randomly selected for this study. The preoperative helical CT scans of the head and neck region were subjected to the AI-enabled automatic segmentation in Mimics Viewer. The performance of AI segmentation was evaluated based on the requirements of computer-assisted surgery. RESULTS: All three patients were successfully segmented by the AI-enabled automatic segmentation. The performance of AI segmentation was excellent regarding key anatomical structures. The overall quality of bone surface was satisfying. The median DICE coefficient was 92.4% for the maxilla, and 94.9% for the mandible, which fulfilled the requirements of computer-assisted craniomaxillofacial surgery. CONCLUSIONS: The AI-enabled automatic segmentation could facilitate the preoperative virtual planning and postoperative outcome verification, which formed a feedback loop to enhance the current workflow of computer-assisted surgery. More studies are warranted to confirm the robustness of AI segmentation with more cases.
Authors: David Steybe; Philipp Poxleitner; Marc Christian Metzger; Leonard Simon Brandenburg; Rainer Schmelzeisen; Fabian Bamberg; Phuong Hien Tran; Elias Kellner; Marco Reisert; Maximilian Frederik Russe Journal: Int J Comput Assist Radiol Surg Date: 2022-06-03 Impact factor: 3.421