Scholarlink

The significance of 3D medical image segmentation in contemporary artificial intelligence-assisted diagnosis is growing. Accurate teeth segmentation can provide strong evidence for disease diagnosis. However, the annotation cost of medical images such as cone beam computed tomography (CBCT) is relatively high, and many unlabeled data have not been fully utilized. In this paper, we propose a training strategy for the 3D CBCT segmentation task and design a semi-supervised learning method to optimize the model using unlabeled data. Our method achieved an average score of 0.7743, an average Dice coefficient of 0.7750, and an average Intersection over Union (IoU) of 0.8167 for teeth segmentation on the validation set of CTooth+.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Automated Dental CBCT Segmentation Using Pseudo Labeling Method

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Weiyan Feng

摘要

The significance of 3D medical image segmentation in contemporary artificial intelligence-assisted diagnosis is growing. Accurate teeth segmentation can provide strong evidence for disease diagnosis. However, the annotation cost of medical images such as cone beam computed tomography (CBCT) is relatively high, and many unlabeled data have not been fully utilized. In this paper, we propose a training strategy for the 3D CBCT segmentation task and design a semi-supervised learning method to optimize the model using unlabeled data. Our method achieved an average score of 0.7743, an average Dice coefficient of 0.7750, and an average Intersection over Union (IoU) of 0.8167 for teeth segmentation on the validation set of CTooth+.