文献速递儿童口腔牙体牙髓病

深度学习检测混合牙列根尖片邻面龋的诊断准确性

Diagnostic accuracy of deep learning for detection of incipient and advanced radiographic interproximal caries in mixed dentition periapical radiographs.
PubMed 文献 · J Dent · 2026-09-23

本研究基于1838张混合牙列儿童根尖片,开发并比较Faster R-CNN、YOLOv11x和DEIMv2-X深度学习模型检测邻面龋。内部测试中DEIMv2-X多尺度集成加TTA获得最高总体Macro-F1(0.538)和mAP50(0.486),YOLOv11x召回最高(0.561);外部验证性能下降,早期龋检测仍具挑战。

要点

  • DEIMv2-X在内部测试中精确率与召回率平衡最佳,Macro-F1为0.538。
  • YOLOv11x外部验证Macro-F1与精确率最高,但所有架构性能均下降。
  • 早期邻面龋检测难度高于进展期龋,临床实施前需多人群验证。

核心结论

  • 深度学习模型检测混合牙列根尖片邻面龋的性能因架构、推理配置和数据集而异。
  • DEIMv2-X在内部评估中精确率与召回率平衡最佳,YOLOv11x在外部数据集上Macro-F1和精确率最高。
  • 早期龋检测仍具挑战,临床实施前需跨人群和临床环境进一步验证。

研究设计与样本

  • 回顾性诊断准确性研究,纳入1838张混合牙列儿童数字根尖片。
  • 两名校准检查者独立标注邻面病变(检查者间κ=0.89;检查者内κ=0.92),分歧通过共识和专家裁决解决。
  • 图像按患者水平分为训练集70%、验证集15%和保留内部测试集15%。
  • 评估Faster R-CNN、YOLOv11x和基于Transformer的DEIMv2-X,采用单尺度、多尺度和测试时增强(TTA)配置。

主要发现

  • 内部测试集中,DEIMv2-X多尺度集成加TTA获得最高总体Macro-F1(0.538)和mAP50(0.486)。
  • YOLOv11x相同配置下获得最高总体召回(0.561)。
  • 进展期龋检测性能 consistently 高于早期龋。
  • 二分类病变检测中,DEIMv2-X多尺度集成加TTA获得最高F1-score(0.630)。
  • 外部验证显示所有架构性能下降,YOLOv11x总体Macro-F1和精确率最高。

关键数值与推荐

  • 病变按影像范围分为早期或进展期。
  • 评估指标包括病变级精确率、召回率、F1-score、mAP和AUPRC。
  • 内部测试中DEIMv2-X多尺度集成加TTA的Macro-F1为0.538,mAP50为0.486。
  • 内部测试中YOLOv11x多尺度集成加TTA的召回为0.561。
  • 二分类病变检测中DEIMv2-X多尺度集成加TTA的F1-score为0.630。

临床怎么用

  • 常规牙科护理中获取的根尖片可提供邻面龋的额外诊断信息,无需额外辐射暴露。
  • 架构和推理配置的差异为开发更准确的AI辅助检测系统提供基础。
  • 临床检查仍是确定病变活动性、成洞状态和治疗决策的必要手段。

局限性

  • 外部验证中所有架构性能下降,提示模型泛化能力有限。
  • 早期邻面龋检测仍具挑战性。
  • 临床实施前需在不同人群和临床环境中进一步验证。

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: This study aimed to develop and compare deep-learning models for detecting proximal carious lesions of different radiographic severity on mixed-dentition periapical radiographs. METHODS: A retrospective diagnostic accuracy study included 1,838 digital periapical radiographs from pediatric patients in the mixed dentition stage. Two calibrated examiners independently annotated proximal radiographic lesions using bounding boxes (inter-examiner κ=0.89; intra-examiner κ=0.92), with disagreements resolved by consensus and specialist adjudication. Lesions were classified as incipient or advanced based on radiographic extent. Images were partitioned at the patient level into training (70%), validation (15%), and held-out internal test (15%) sets. Faster R-CNN, YOLOv11x, and transformer-based DEIMv2-X were evaluated using lesion-level precision, recall, F1-score, mAP, and AUPRC. Single-scale, multi-scale, and test-time augmentation (TTA) configurations were evaluated. RESULTS: On the held-out internal test set, DEIMv2-X achieved the highest overall Macro-F1 (0.538) and mAP50 (0.486) using the multi-scale ensemble with TTA, whereas YOLOv11x achieved the highest overall recall (0.561) using the same configuration. Detection performance was consistently higher for advanced than incipient radiographic lesions. In binary lesion detection, DEIMv2-X multi-scale ensemble with TTA achieved the highest F1-score (0.630). External validation showed reduced performance across all architectures, with YOLOv11x demonstrating the highest overall Macro-F1 and precision. CONCLUSIONS: Performance varied across architectures, inference configurations, and datasets. DEIMv2-X achieved the highest overall balance between precision and recall in the internal evaluation, whereas YOLOv11x achieved the highest overall Macro-F1 and precision on the independent external dataset. Incipient lesions remained challenging to detect, highlighting the need for further validation across diverse populations and clinical settings before clinical implementation. CLINICAL SIGNIFICANCE: Periapical radiographs acquired during routine dental care may provide additional diagnostic information regarding proximal carious lesions without additional radiation exposure. The observed differences among detection architectures and inference configurations provide a foundation for developing more accurate AI-assisted detection systems, although clinical examination remains essential for determining lesion activity, cavitation status, and treatment decisions.

来源:J Dent,查看原文。版权归原作者所有。

引用

  1. Baraka M, Elsayed A, Issa A, et al. Diagnostic accuracy of deep learning for detection of incipient and advanced radiographic interproximal caries in mixed dentition periapical radiographs. J Dent. 2026. PMID: 42778128. doi:10.1016/j.jdent.2026.107071

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。