文献速递神经技术与安全CT

混合3D/2D卷积神经网络用于头部CT出血评估

Hybrid 3D/2D Convolutional Neural Network for Hemorrhage Evaluation on Head CT.
PubMed 文献 · AJNR Am J Neuroradiol · 2018-07-26

本研究分两阶段开发并前瞻性验证了一种定制混合3D/2D掩模ROI卷积神经网络,用于平扫CT上脑实质、硬膜外/硬膜下及蛛网膜下腔出血的检测与量化。训练队列10159例(出血901例,8.1%),前瞻性测试队列862例(出血82例,12%)。检测准确度0.970、AUC 0.981、敏感度0.951、特异度0.973;脑实质出血、硬膜外/硬膜下出血、蛛网膜下腔出血的Dice评分分别为0.931、0.863、0.772。

要点

  • 前瞻性测试集出血检测AUC 0.981,敏感度0.951、特异度0.973、阴性预测值0.993。
  • 脑实质出血量化Dice 0.931,硬膜外/硬膜下出血0.863,蛛网膜下腔出血0.772。
  • 作者认为该深度学习工具在急诊平扫CT出血检测与量化上准确,具有临床可行性。

核心结论

  • 定制的深度学习工具在平扫CT上检测和量化出血是准确的,前瞻性急诊平扫CT上的高性能提示该工具具有临床可行性。

研究设计与样本

  • 研究分两阶段:第一阶段用2017年1月1日至2017年7月31日单机构所有平扫CT作为训练队列,开发并交叉验证定制混合3D/2D掩模ROI卷积神经网络;第二阶段将训练好的网络通过自动推理流程前瞻性应用于2018年2月1日至2018年2月28日急诊开具的所有平扫CT。

关键影像表现/征象

  • 训练队列10159例检查(512598幅图像;出血901例,占8.1%),测试队列862例检查(23668幅图像;出血82例,占12%)。
  • 训练队列交叉验证的出血检测准确度0.975、AUC 0.983、敏感度0.971、特异度0.975、阳性预测值0.793、阴性预测值0.997。
  • 前瞻性测试集的出血检测准确度0.970、AUC 0.981、敏感度0.951、特异度0.973、阳性预测值0.829、阴性预测值0.993。
  • 脑实质出血、硬膜外/硬膜下出血、蛛网膜下腔出血的Dice评分分别为0.931、0.863、0.772。

诊断标准与关键数值

  • 出血检测在训练队列交叉验证的准确度0.975、AUC 0.983、敏感度0.971、特异度0.975、阳性预测值0.793、阴性预测值0.997。
  • 出血检测在前瞻性测试集的准确度0.970、AUC 0.981、敏感度0.951、特异度0.973、阳性预测值0.829、阴性预测值0.993。
  • 量化评估采用Dice评分系数和Pearson相关;脑实质出血Dice 0.931、硬膜外/硬膜下出血Dice 0.863、蛛网膜下腔出血Dice 0.772。

临床应用与阅片要点

  • 作者认为该定制深度学习工具在平扫CT出血检测与量化上准确,前瞻性急诊平扫CT的高性能提示其具有临床可行性。

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)。

英文原文(摘要)· 点击展开

BACKGROUND AND PURPOSE: Convolutional neural networks are a powerful technology for image recognition. This study evaluates a convolutional neural network optimized for the detection and quantification of intraparenchymal, epidural/subdural, and subarachnoid hemorrhages on noncontrast CT.

MATERIALS AND METHODS: This study was performed in 2 phases. First, a training cohort of all NCCTs acquired at a single institution between January 1, 2017, and July 31, 2017, was used to develop and cross-validate a custom hybrid 3D/2D mask ROI-based convolutional neural network architecture for hemorrhage evaluation. Second, the trained network was applied prospectively to all NCCTs ordered from the emergency department between February 1, 2018, and February 28, 2018, in an automated inference pipeline. Hemorrhage-detection accuracy, area under the curve, sensitivity, specificity, positive predictive value, and negative predictive value were assessed for full and balanced datasets and were further stratified by hemorrhage type and size. Quantification was assessed by the Dice score coefficient and the Pearson correlation.

RESULTS: A 10,159-examination training cohort (512,598 images; 901/8.1% hemorrhages) and an 862-examination test cohort (23,668 images; 82/12% hemorrhages) were used in this study. Accuracy, area under the curve, sensitivity, specificity, positive predictive value, and negative-predictive value for hemorrhage detection were 0.975, 0.983, 0.971, 0.975, 0.793, and 0.997 on training cohort cross-validation and 0.970, 0.981, 0.951, 0.973, 0.829, and 0.993 for the prospective test set. Dice scores for intraparenchymal hemorrhage, epidural/subdural hemorrhage, and SAH were 0.931, 0.863, and 0.772, respectively.

CONCLUSIONS: A customized deep learning tool is accurate in the detection and quantification of hemorrhage on NCCT. Demonstrated high performance on prospective NCCTs ordered from the emergency department suggests the clinical viability of the proposed deep learning tool.

来源:AJNR Am J Neuroradiol,查看原文。版权归原作者所有。

引用

  1. Chang PD, Kuoy E, Grinband J, et al. Hybrid 3D/2D Convolutional Neural Network for Hemorrhage Evaluation on Head CT. AJNR Am J Neuroradiol. 2018. PMID: 30049723. doi:10.3174/ajnr.A5742

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。