Department of Orthodontics, Cranial-Facial Growth and Development Center, Peking University School and Hospital of Stomatology, 22 Zhongguancun South Avenue, Haidian District, Beijing, 100081, People's Republic of China.
First Clinical Division, Peking University School and Hospital of Stomatology, 37A Xishiku Street, Xicheng District, Beijing, 100034, People's Republic of China.
Head Face Med. 2022 Feb 14;18(1):5. doi: 10.1186/s13005-022-00309-2.
To establish the digital diagnostic templates by cluster analysis based on a set of cephalometric films and evaluate the outcome of the different treatment methods in the patients affiliated to the same cephalometric morphology template (CMT). These templates could be used for the automatic diagnosis of dentofacial deformities and prediction of treatment outcomes in the future.
In this study, we assessed the coordinates of 60 different landmarks on the cephalograms of 2249 patients (14.35 ± 4.99 years, range from 7 to 62) with dentofacial deformities. The cephalometric data were subjected to dentist for clustering without a priori pattern definitions to generate biologically informative CMTs. Three templates were selected to evaluate the treatment outcome of patients affiliated to the same CMT.
The cluster analysis yielded 21 distinct groups. The total discriminant accuracy was 89.1%, while the cross-validation accuracy was 85.0%, showing that the clusters were robust. All CMTs were automatically created and drawn using a computer, based on the average coordinates of each cluster. Individuals affiliated to the same CMT showed similar dentofacial features. We also evaluated differences in the outcomes of patients affiliated to the same CMT.
Our results demonstrated the utility of clustering methods for grouping dentofacial deformities with similar dentofacial features. Clustering methods can be used to evaluate the differences in the outcomes of patients affiliated to the same CMT, which has good clinical application value.
通过聚类分析建立基于一组头颅侧位片的数字化诊断模板，并评估具有相同头颅测量形态模板（CMT）的患者的不同治疗方法的结果。这些模板可用于未来的牙颌面畸形自动诊断和治疗效果预测。
本研究评估了 2249 例牙颌面畸形患者（年龄 14.35±4.99 岁，7~62 岁）的 60 个不同标志点的坐标。对头颅侧位片数据进行聚类分析，无需先验模式定义，以生成具有生物学意义的 CMT。选择 3 个模板来评估具有相同 CMT 的患者的治疗效果。
聚类分析产生了 21 个不同的组。总判别准确率为 89.1%，交叉验证准确率为 85.0%，表明聚类是稳健的。所有 CMT 均基于每个聚类的平均坐标，使用计算机自动创建和绘制。具有相同 CMT 的个体具有相似的牙颌面特征。我们还评估了具有相同 CMT 的患者治疗效果的差异。
本研究结果表明，聚类方法可用于分组具有相似牙颌面特征的牙颌面畸形。聚类方法可用于评估具有相同 CMT 的患者治疗效果的差异，具有良好的临床应用价值。
