Li Mengxun, Xu Xiangyang, Punithakumar Kumaradevan, Le Lawrence H, Kaipatur Neelambar, Shi Bin
Department of Implantology, School and Hospital of Stomatology, Wuhan University, Wuhan, China.
School of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China.
Comput Biol Med. 2020 Jul;122:103794. doi: 10.1016/j.compbiomed.2020.103794. Epub 2020 May 23.
Digital smile design is the technique that dentists use to analyze, design, and visualize therapeutic results on a computing workstation prior to actual treatment. Despite it being a crucial step in digital smile design, the process of labeling and integrating the information in facial and intra-oral images is laborious. Therefore, this study aims to develop an automated photo integrating system to facilitate this process.
The teeth in intra-oral images were distinguished by their curvature and finely segmented using an active contour model. The facial keypoints were detected by a sophisticated facial landmark detector algorithm; these keypoints were then overlaid on the corresponding intra-oral image by extracting the contour of the teeth in the facial and intra-oral photographs. With this system, the tooth width-to-height ratios, smile line, and facial midline were automatically marked in the intra-oral image. The accuracy of the proposed segmentation algorithm was evaluated by applying it to 50 images with 274 maxillary anterior teeth.
The proposed algorithm recognized 96.0% (263/274) of teeth in our selected image set. The results were then compared to those obtained by applying manual segmentation to the remaining 263 recognized teeth. With a 95% confidence interval, a Jaccard index of 0.928 ± 0.081, average distance of 0.128 ± 0.109 mm, and Hausdorff distance between the results and ground truth of 0.461 ± 0.495 mm were achieved.
The results of this study show that the proposed automated system can eliminate the need for dentists to employ a laborious image integration process. It also has the potential for broad applicability in the field of dentistry.
数字微笑设计是一种技术，牙医可在实际治疗前，在计算机工作站上分析、设计并可视化治疗效果。尽管这是数字微笑设计中的关键步骤，但对面部和口腔内图像中的信息进行标注和整合的过程却很繁琐。因此，本研究旨在开发一种自动照片整合系统，以简化这一过程。
利用主动轮廓模型，根据口腔内图像中牙齿的曲率对其进行区分并精细分割。通过一种复杂的面部标志点检测算法来检测面部关键点；然后通过提取面部和口腔内照片中牙齿的轮廓，将这些关键点叠加到相应的口腔内图像上。借助该系统，可在口腔内图像中自动标记出牙宽高比、微笑线和面部中线。通过将所提出的分割算法应用于50张包含274颗上颌前牙的图像，评估该算法的准确性。
在所选择的图像集中，所提出的算法识别出了96.0%（263/274）的牙齿。然后将结果与对其余263颗识别出的牙齿进行手动分割所获得的结果进行比较。在95%的置信区间下，杰卡德指数为0.928±0.08（此处原文有误，应为0.081），平均距离为0.128±0.109毫米，结果与真实值之间的豪斯多夫距离为0.461±0.495毫米。
本研究结果表明，所提出的自动化系统可消除牙医进行繁琐图像整合过程的需求。它在牙科领域也具有广泛应用的潜力。
