Radiology Artificial Intelligence Lab (RAIL), Malone Center for Engineering in Healthcare, Johns Hopkins University Whiting School of Engineering, Baltimore, MD, USA.
The Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
Emerg Radiol. 2020 Aug;27(4):367-375. doi: 10.1007/s10140-020-01767-4. Epub 2020 Jul 8.
To (1) develop a deep learning system (DLS) using a deep convolutional neural network (DCNN) for identification of pneumothorax, (2) compare its performance to first-year radiology residents, and (3) evaluate the ability of a DLS to augment radiology residents by detecting missed pneumothoraces.
This was a retrospective study performed in September 2018. We obtained 112,120 chest radiographs (CXRs) from the NIH ChestXray14 database, of which 4360 cases (4%) were labeled as pneumothorax by natural language processing. We utilized 111,518 CXRs to train and validate the ResNet-152 DCNN pretrained on ImageNet to identify pneumothorax. DCNN testing was performed on a hold-out set of 602 CXRs, whose groundtruth was determined by a cardiothoracic radiologist. Two first-year radiology residents evaluated the test CXRs for presence of pneumothorax. Receiver operating characteristic (ROC) curves were generated for each evaluator with area under the curve (AUC) compared using the DeLong parametric method.
The DCNN achieved AUC of 0.841 for identification of pneumothorax at a rate of 1980 images/min. In contrast, both first-year residents achieved significantly higher AUCs of 0.942 and 0.905 (p < 0.01 for both compared to DCNN), but at a slower rate of two images/min. The DCNN identified 3 of 31 (9.7%) additional pneumothoraces missed by at least one of the residents.
A DLS for pneumothorax identification had lower AUC than 1st-year radiology residents, but interpreted images > 1000× as fast and identified 3 additional pneumothoraces missed by the residents. Our findings suggest that DLS could augment radiologists-in-training to identify potential urgent findings.
（1）开发一种基于深度卷积神经网络（DCNN）的深度学习系统（DLS），用于识别气胸，（2）将其性能与第一年的放射科住院医师进行比较，（3）评估 DLS 通过检测漏诊气胸来增强放射科住院医师的能力。
这是一项 2018 年 9 月进行的回顾性研究。我们从 NIH ChestXray14 数据库中获得了 112120 张胸部 X 光片（CXR），其中 4360 例（4%）通过自然语言处理标记为气胸。我们利用 111518 张 CXR 训练和验证了在 ImageNet 上预训练的 ResNet-152 DCNN 以识别气胸。DCNN 测试是在一组 602 张 CXR 的保留集上进行的，其金标准由一名心胸放射科医生确定。两名第一年的放射科住院医师评估了测试 CXR 中气胸的存在。生成了每个评估者的接收器工作特征（ROC）曲线，并使用 DeLong 参数方法比较了曲线下面积（AUC）。
DCNN 识别气胸的 AUC 为 0.841，图像识别速度为 1980 张/分钟。相比之下，两名第一年的住院医师的 AUC 明显更高，分别为 0.942 和 0.905（与 DCNN 相比，两者均 p < 0.01），但速度较慢，为 2 张/分钟。DCNN 识别出至少一名住院医师漏诊的 31 例气胸中的 3 例。
用于气胸识别的 DLS 的 AUC 低于第一年的放射科住院医师，但可以快速解释 > 1000 张图像，并识别出住院医师漏诊的 3 例气胸。我们的研究结果表明，DLS 可以增强放射科住院医师识别潜在紧急发现的能力。
