欢迎来到《四川大学学报（医学版）》
高级检索
赵雨暄, 张啸云, 杨必, 等. 深度学习算法在角膜荧光染色分级评估中的应用[J]. 四川大学学报（医学版）, 2023, 54(5): 908-914. DOI: 10.12182/20230960104
引用本文: 赵雨暄, 张啸云, 杨必, 等. 深度学习算法在角膜荧光染色分级评估中的应用[J]. 四川大学学报（医学版）, 2023, 54(5): 908-914. DOI: 10.12182/20230960104
shu
ZHAO Yuxuan, ZHANG Xiaoyun, YANG Bi, et al. Application of Deep Learning Algorithm in the Grading Assessment of Corneal Fluorescein Staining[J]. Journal of Sichuan University (Medical Sciences), 2023, 54(5): 908-914. DOI: 10.12182/20230960104
Citation: ZHAO Yuxuan, ZHANG Xiaoyun, YANG Bi, et al. Application of Deep Learning Algorithm in the Grading Assessment of Corneal Fluorescein Staining[J]. Journal of Sichuan University (Medical Sciences), 2023, 54(5): 908-914. DOI: 10.12182/20230960104
shu

深度学习算法在角膜荧光染色分级评估中的应用

Application of Deep Learning Algorithm in the Grading Assessment of Corneal Fluorescein Staining

    摘要
  • 摘要:
      目的  探讨深度学习算法在角膜荧光染色分级评估中的应用价值。
      方法  采用横断面研究，共纳入四川大学华西医院隐形眼镜门诊就诊患者的角膜染色图片600张。其中500张用于构建算法，其余100张用于验证算法有效性并比较人工智能（artificial intelligence, AI）与视光初学者间评级准确性（accuracy, ACC）及诊断时长的差异。在完成初次评级1个月后，进行第二次评估，比较两次评估的一致性指数（kappa值）。3位经验丰富的视光师综合分级的结果作为本研究的“金标准”。
      结果  通过全集、训练集、测试集交叉验证4种深度学习模型，ResNet34模型预测准确性最高。ResNet34深度学习模型在角膜染色分级上的准确性为93.0%，敏感性为89.5%，特异性为89.6%。AI与两位初学者比较，AI的准确性较高（ACCAI=87.0%、ACCS1=78.0%、ACCS2=52.0%，PACC=0.001），同时AI的平均诊断时长短于视光初学者（tAI=1.00 s，tS1=11.86 s，tS2=13.25 s，Pt=0.001）。在两次评级的一致性比较中，AI（kappaAI=0.658，PAI=0.001）的一致性程度高于视光初学者（kappaS1=0.575, PS1=0.001; kappaS2=0.609，PS2=0.001）。
      结论  将深度学习算法应用于角膜染色分级评估具有一定的可行性及临床价值。在与初学者的比较中，AI表现出了较高的准确性和较好的一致性，能够指导视光师在临床工作中快速准确地评定角膜染色分级。

     

    Abstract:
      Objective  To explore the application value of applying deep learning (DL) algorithm in the grading assessment of corneal fluorescein staining.
      Methods  A cross-sectional study was carried out, covering 600 corneal fluorescein staining photos acquired in the Contact Lens Clinic, West China Hospital, Sichuan University between 2020 and 2022. Out of the 600 photos, 500 were used to construct the algorithm and the remaining 100 were used for the validation of the algorithm and a comparative analysis of the difference in grading accuracy (ACC) and the length of diagnostic time between artificial intelligence (AI) and optometry students. One month after finishing the first grading analysis, assessment by AI and optometry students was conducted for a second time and results from the two rounds of assessment were compared to examine the intrarater agreement (kappa value) of the two analyses. The grading analysis results of 3 experienced optometrists were used as the gold standard in the study.
      Results   Findings of the cross validation with the complete dataset, the training dataset, and the test dataset showed that ResNet34 had the highest predictive accuracy among four DL models. ResNet34 DL model achieved an accuracy of 93.0%, sensitivity of 89.5%, and specificity of 89.6% in the grading of corneal staining. In the comparison of the grading accuracy of AI and two optometry students, AI showed better accuracy, with the respective grading accuracy being 87.0%, 78.0%, and 52.0% for AI, student 1, and student 2 (PACC=0.001). In addition, the average diagnostic time of AI was shorter than that of optometry students (tAI=1.00 s, tS1=11.86 s, tS2=13.25 s, Pt=0.001). In the comparative analysis of the intrarater agreement between the two assessments, AI (kappaAI=0.658, PAI=0.001) achieved better consistency than the two optometry students did （kappaS1=0.575, PS1=0.001; kappaS2=0.609, PS2=0.001).
      Conclusion  Applying deep learning algorithms in the grading assessment of corneal fluorescein staining has considerable feasibility and clinical value. In the performance comparison between AI and optometry students, AI achieved higher accuracy and better consistency, which indicates that AI has potential application value for assisting optometrists to make clinical decisions with speed and accuracy.

     

    • HTML全文
    • 参考文献(26)
    • 相关文章
    • 施引文献
  • 资源附件(0)

/

返回文章
返回