文献速递口腔修复与种植口腔颌面外科

基于解剖的多任务深度学习影像组学列线图预测上颌窦底提升种植失败风险

Anatomically Based Multitask Deep Learning Radiomics Nomogram Predicts the Implant Failure Risk in Sinus Floor Elevation.
PubMed 文献 · Clin Oral Implants Res · 2025-07-23

本研究开发并评估了基于解剖的多任务深度学习影像组学列线图(AMDRN)系统,用于在上颌窦底提升(MSFE)术前预测种植失败风险,并实现关键解剖结构的自动分割。回顾性收集患者术前锥形束CT(CBCT)和电子病历(EMR),优化nn-UNet v2分割上颌窦(MS)、施耐德膜(SM)和剩余牙槽骨(RAB),基于分割掩膜构建深度学习模型(3D-Attention-ResNet)和影像组学模型提取3D特征生成DL Score和Rad Score,结合临床特征构建临床模型,通过逻辑回归(LR)整合为AMDRN模型。测试集上MS、RAB、SM的DICE系数分别为99.50%±0.84%、92.53%±3.78%、91.58%±7.16%;独立测试集上临床模型、影像组学模型、3D-DL模型和AMDRN模型的预测准确率分别为60%、76%、82%和90%,AMDRN的AUC最高为93%。结论:AMDRN系统能够高效术前预测MSFE种植失败风险并准确分割关键解剖结构,支持个性化治疗规划和临床风险管理。

要点

  • AMDRN模型在独立测试集预测MSFE种植失败风险的准确率达90%,AUC为93%。
  • nn-UNet v2分割MS、RAB、SM的DICE系数分别为99.50%±0.84%、92.53%±3.78%、91.58%±7.16%。
  • 系统整合深度学习、影像组学和临床特征,并提供可视化模块辅助临床决策。

核心结论

  • AMDRN系统可高效术前预测MSFE种植失败风险并准确分割关键解剖结构,支持个性化治疗规划和临床风险管理。

研究设计与样本

  • 回顾性收集患者术前CBCT图像和电子病历,优化nn-UNet v2分割MS、SM和RAB,基于分割掩膜构建3D-Attention-ResNet深度学习模型和影像组学模型提取3D特征生成DL Score和Rad Score,结合临床特征构建临床模型,通过逻辑回归整合为AMDRN模型。

主要发现

  • 测试集上MS、RAB、SM的DICE系数分别为99.50%±0.84%、92.53%±3.78%、91.58%±7.16%。
  • 独立测试集上临床模型、影像组学模型、3D-DL模型和AMDRN模型的预测准确率分别为60%、76%、82%和90%。
  • AMDRN模型在独立测试集上AUC最高为93%。

临床怎么用

  • AMDRN系统可在MSFE术前预测种植失败风险,辅助个性化治疗规划和临床风险管理。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

OBJECTIVES: To develop and assess the performance of an anatomically based multitask deep learning radiomics nomogram (AMDRN) system to predict implant failure risk before maxillary sinus floor elevation (MSFE) while incorporating automated segmentation of key anatomical structures. MATERIALS AND METHODS: We retrospectively collected patients' preoperative cone beam computed tomography (CBCT) images and electronic medical records (EMRs). First, the nn-UNet v2 model was optimized to segment the maxillary sinus (MS), Schneiderian membrane (SM), and residual alveolar bone (RAB). Based on the segmentation mask, a deep learning model (3D-Attention-ResNet) and a radiomics model were developed to extract 3D features from CBCT scans, generating the DL Score, and Rad Score. Significant clinical features were also extracted from EMRs to build a clinical model. These components were then integrated using logistic regression (LR) to create the AMDRN model, which includes a visualization module to support clinical decision-making. RESULTS: Segmentation results for MS, RAB, and SM achieved high DICE coefficients on the test set, with values of 99.50% ± 0.84%, 92.53% ± 3.78%, and 91.58% ± 7.16%, respectively. On an independent test set, the Clinical model, Radiomics model, 3D-DL model, and AMDRN model achieved prediction accuracies of 60%, 76%, 82%, and 90%, respectively, with AMDRN achieving the highest AUC of 93%. CONCLUSION: The AMDRN system enables efficient preoperative prediction of implant failure risk in MSFE and accurate segmentation of critical anatomical structures, supporting personalized treatment planning and clinical risk management.

来源:Clin Oral Implants Res,查看原文。版权归原作者所有。

引用

  1. Zhu Y, Liu Y, Zhao Y, et al. Anatomically Based Multitask Deep Learning Radiomics Nomogram Predicts the Implant Failure Risk in Sinus Floor Elevation. Clin Oral Implants Res. 2025. PMID: 40702787. doi:10.1111/clr.70011

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。