上海口腔医学

上海口腔医学 ›› 2025, Vol. 34 ›› Issue (2): 119-125.doi: 10.19439/j.sjos.2025.02.002

• 论著 • 上一篇    下一篇

基于深度学习的CBCT中下颌管及分支的分割与验证

叶页1,*, 房硕博1,*, 路惠童2, 刘鸣谦2, 邬雪颖1   

  1. 1.上海市口腔医院 修复科,上海 200001;
    2.卫宁健康科技集团股份有限公司,上海 200072
  • 收稿日期:2023-11-24 修回日期:2024-01-24 出版日期:2025-04-25 发布日期:2025-05-15
  • 通讯作者: 邬雪颖,E-mail: xueyingwu_kq@fudan.edu.cn
  • 作者简介:叶页（1991-）,女,硕士,主治医师,E-mail:yeye_91@126.com;房硕博（1995-）,男,博士,住院医师,E-mail:fsbm1995@163.com。*并列第一作者
  • 基金资助:
    上海市卫生健康委员会卫生行业临床研究专项（202040494）; 上海市口腔医院课题（sh-2022-yj-a04）

Segmentation and validation of mandibular canal and its bifurcation on cone beam CT based on deep learning

Ye Ye1, Fang Shuobo1, Lu Huitong2, Liu Mingqian2, Wu Xueying1   

  1. 1. Department of Prosthodontics, Shanghai Stomatological Hospital & School of Stomatology, Fudan University. Shanghai 200001;
    2. Winning Health Technology Group Co., Shanghai 200072, China
  • Received:2023-11-24 Revised:2024-01-24 Online:2025-04-25 Published:2025-05-15

PDF (PC)

4

摘要： 目的：通过训练卷积神经网络中的U形网络架构（U-Net）,构建识别和分割下颌管及其分叉的方法。以专家标注结果作为标准,评估方法的准确性。方法：收集2022年1月—2022年12月于上海市口腔医院就诊的290例患者的CBCT资料,其中,训练集200例,测试集90例。模型训练分为三步,第一步由研究人员在3D Slicer软件中标注50例CBCT的双侧下颌管及其分叉。第二步采用伪标签法,基于人工标注的50例数据,结合数据增强方法,初步训练U-net的三维分割模型,对分割结果进行形态学后处理;应用初步训练的模型,对150例数据进行智能标注,人工审校后纳入训练集。第三步基于人工标注与审校的共200例数据训练三维U-net识别与分割模型。评估阶段,由2名医师与U-net模型分别标注90例测试集CBCT数据的双侧下颌管及其分叉,检验2名医师标注结果的一致性,计算人工标注与模型标注结果的Dice相似性系数（Dice similarity coefficient, DSC）与豪斯道夫距离（Hausdorff distance, HSD）,计算双下颌管检出率。采用SPSS 20.0软件包对数据进行统计学分析。结果：人工标注的90个测试集中,医师之间的Kappa值为 0.667。模型标注与人工标注相比,DSC为（0.739±0.068）, HSD为（0.988±1.14） mm。双下颌管的检出率达到91.30%。结论：本模型具有较高的分割精度和预测精确性,是一种可靠实用的CBCT下颌管分割方法。

关键词: 深度学习, 下颌管, 双下颌管, CBCT

Abstract: PURPOSE: To train the U-net of convolutional neural network to establish a method for detecting and segmenting the mandibular canal and its bifurcation, and validate its accuracy based on the ground truth labeled by experts. METHODS: A total of 290 CBCT scans were collected from Shanghai Stomatological Hospital from Jan. 2022 to Dec. 2022, which were divided into training set of 200 scans and test set of 90 scans. Model training included two steps. In the first step, bilateral mandibular canals and its bifurcation of 50 CBCT scans were labeled in 3D Slicer image computing platform by investigators. Three dimensional U-net segmentation model were trained initially with data enhancement. A morphological post-processing method was applied to the predicted results. In the second step, pseudo label method was employed to help annotating the mandibular canal and corresponding bifurcations on remaining 150 CBCTs, which would be included in training set after revision. Three dimensional U-net model was trained based on these 200 data. During test phase, totally 90 scans were labeled by two doctors and U-net model respectively. Consistency check was conducted to evaluate the labels between two doctors. Dice similarity coefficient and Hausdorff distance were calculated to evaluate the labels between doctors and the model. The detection rate of bifurcation was calculated. SPSS 20.0 software package was used for data analysis. RESULTS: In 90 CBCT test set, the Kappa value between two dentists' annotations was 0.667. The average Dice and Hausdorff distance between predictions and labels of doctors were (0.739±0.068) and (0.988±1.14) mm. In bifurcation detection, the detection rate was 91.30% on scans with clear bifurcations. CONCLSIONS: The dentification and segmentation U-net model of mandibular canal on dental CBCT can be reliable and practical for its high segmentation precision and predicting speed.

Key words: Deep learning, Mandibular canal, Bifid mandibular canal, CBCT

中图分类号: