Detection of optical coherence tomography-defined thin-cap fibroatheroma in the coronary artery using deep learning
Detection of optical coherence tomography-defined thin-cap fibroatheroma in the coronary artery using deep learning
复制标题
DOI：
10.4244/eij-d-19-00487
复制
发表时间：
2020-08-01
期刊：
影响因子：
6.2
通讯作者：
Park, Seung-Jung
中科院分区：
文献类型：
--
作者：
Min, Hyun-Seok;Yoo, Ji Hyeong;Park, Seung-Jung
目的：本研究的目的是开发一种深度学习模型，用于对使用或不使用光学相干断层扫描 (OCT) 衍生的薄帽纤维粥样斑块 (TCFA) 的帧进行分类。方法和结果：将 602 名心绞痛患者的总共 602 个冠状动脉病变按 4:1 的比例随机分为训练组和测试组。开发了 DenseNet 模型来对带有或不带有 OCT 派生 TCFA 的 OCT 帧进行分类。梯度加权类激活映射用于可视化关注区域。在训练样本（480 个病变的 35,678 帧）中，五重交叉验证的模型在预测 TCFA 存在时的总体准确度为 91.6 +/- 1.7%，敏感性为 88.7 +/- 3.4%，特异性为 91.8 +/- 2.0%（平均 AUC=0.96 +/- 0.01）。在测试样本中（122 个病灶的 9,722 帧），病灶内帧级别的总体准确率为 92.8%（AUC=0.96），整个 OCT 回撤的准确率为 91.3%。模型预测的每艘船 TCFA 负荷百分比与专家确定的百分比之间的相关性显着（r=0.87，p
Aims: The aim of this study was to develop a deep learning model for classifying frames with versus with -out optical coherence tomography (OCT)-derived thin-cap fibroatheroma (TCFA).Methods and results: A total of 602 coronary lesions from 602 angina patients were randomised into training and test sets in a 4:1 ratio. A DenseNet model was developed to classify OCT frames with or without OCT-derived TCFA. Gradient-weighted class activation mapping was used to visualise the area of attention. In the training sample (35,678 frames of 480 lesions), the model with fivefold cross-validation had an overall accuracy of 91.6 +/- 1.7%, sensitivity of 88.7 +/- 3.4%, and specificity of 91.8 +/- 2.0% (averaged AUC=0.96 +/- 0.01) in predicting the presence of TCFA. In the test samples (9,722 frames of 122 lesions), the overall accuracy at the frame level was 92.8% within the lesion (AUC=0.96) and 91.3% in the entire OCT pullback. The correlation between the %TCFA burden per vessel predicted by the model compared with that identified by experts was significant (r=0.87, p