Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD, USA.
Faculty of Medicine, University of Ottawa, Ontario, Canada.
Skeletal Radiol. 2022 Nov;51(11):2121-2128. doi: 10.1007/s00256-022-04077-7. Epub 2022 May 28.
Deep learning has the potential to automatically triage orthopedic emergencies, such as joint dislocations. However, due to the rarity of these injuries, collecting large numbers of images to train algorithms may be infeasible for many centers. We evaluated if the Internet could be used as a source of images to train convolutional neural networks (CNNs) for joint dislocations that would generalize well to real-world clinical cases.
We collected datasets from online radiology repositories of 100 radiographs each (50 dislocated, 50 located) for four joints: native shoulder, elbow, hip, and total hip arthroplasty (THA). We trained a variety of CNN binary classifiers using both on-the-fly and static data augmentation to identify the various joint dislocations. The best-performing classifier for each joint was evaluated on an external test set of 100 corresponding radiographs (50 dislocations) from three hospitals. CNN performance was evaluated using area under the ROC curve (AUROC). To determine areas emphasized by the CNN for decision-making, class activation map (CAM) heatmaps were generated for test images.
The best-performing CNNs for elbow, hip, shoulder, and THA dislocation achieved high AUROCs on both internal and external test sets (internal/external AUC): elbow (1.0/0.998), hip (0.993/0.880), shoulder (1.0/0.993), THA (1.0/0.950). Heatmaps demonstrated appropriate emphasis of joints for both located and dislocated joints.
With modest numbers of images, radiographs from the Internet can be used to train clinically-generalizable CNNs for joint dislocations. Given the rarity of joint dislocations at many centers, online repositories may be a viable source for CNN-training data.
深度学习有可能自动分诊骨科急症，如关节脱位。然而，由于这些损伤的罕见性，许多中心可能难以收集大量图像来训练算法。我们评估了互联网是否可以用作训练用于关节脱位的卷积神经网络（CNN）的图像来源，这些网络可以很好地推广到实际临床病例。
我们从四个关节的在线放射学存储库中收集了每个数据集 100 张射线照片（50 张脱位，50 张在位）：原生肩、肘、髋和全髋关节置换术（THA）。我们使用即时和静态数据增强训练了各种 CNN 二进制分类器，以识别各种关节脱位。每个关节表现最佳的分类器在来自三个医院的 100 张对应射线照片（50 张脱位）的外部测试集中进行了评估。使用 ROC 曲线下的面积（AUROC）评估 CNN 性能。为了确定 CNN 用于决策的强调区域，为测试图像生成了类激活图（CAM）热图。
在内部和外部测试集上，表现最佳的肘部、髋部、肩部和 THA 脱位 CNN 都获得了较高的 AUROC（内部/外部 AUC）：肘部（1.0/0.998），髋部（0.993/0.880），肩部（1.0/0.993），THA（1.0/0.950）。热图显示了位于和脱位关节的关节的适当强调。
使用少量图像，来自互联网的射线照片可以用于训练用于关节脱位的具有临床通用性的 CNN。鉴于许多中心关节脱位的罕见性，在线存储库可能是 CNN 训练数据的可行来源。
