Automated identification of retinopathy of prematurity by image-based deep learning
Automated identification of retinopathy of prematurity by image-based deep learning
复制标题
DOI：
10.1186/s40662-020-00206-2
复制
发表时间：
2020-08-01
期刊：
影响因子：
4.2
通讯作者：
Shen, Yin
中科院分区：
文献类型：
--
作者：
Tong, Yan;Lu, Wei;Shen, Yin
早产儿视网膜病变(ROP)是世界范围内导致儿童失明的主要原因之一，但只要得到适当和及时的诊断，它是一种可以治疗的视网膜疾病。本研究旨在开发一个基于深度学习的健壮智能系统，从眼底图像中自动对ROP的严重程度进行分类，并检测ROP的分期和PLUS疾病的存在，从而实现自动化诊断和进一步的治疗。训练了一个101层卷积神经网络(ResNet)和一个更快的基于区域的卷积神经网络(FASTER-RCNN)用于图像分类和识别。我们应用了10次交叉验证的方法来训练和优化我们的算法。在四级分类任务中评估了准确性、灵敏度和特异度，以评估智能系统的性能。将该系统的性能与两位视网膜专家的结果进行了比较。此外，该系统还设计了基于FAST-RCNN的目标检测网络来检测ROP的分期和是否存在PLUS疾病，以及突出病变区域。结果该系统对ROP的严重程度分类的准确率为0.903。具体来说，区分正常、轻度、半紧急和紧急的准确率分别为0.883、0.900、0.957和0.870；两位专家的相应准确率分别为0.902和0.898。此外，我们的模型对ROP分期的准确率为0.957，对PLUS疾病的准确率为0.896，对I期到V期的判别准确率分别为0.876、0.942、0.968、0.998和0.999。结论我们的系统能够检测ROP并区分四级分类眼底图像，具有较高的准确性和特异性。该系统的性能与人类专家相当或更好，表明该系统可以用于支持临床决策。
BackgroundRetinopathy of prematurity (ROP) is a leading cause of childhood blindness worldwide but can be a treatable retinal disease with appropriate and timely diagnosis. This study was performed to develop a robust intelligent system based on deep learning to automatically classify the severity of ROP from fundus images and detect the stage of ROP and presence of plus disease to enable automated diagnosis and further treatment.MethodsA total of 36,231 fundus images were labeled by 13 licensed retinal experts. A 101-layer convolutional neural network (ResNet) and a faster region-based convolutional neural network (Faster-RCNN) were trained for image classification and identification. We applied a 10-fold cross-validation method to train and optimize our algorithms. The accuracy, sensitivity, and specificity were assessed in a four-degree classification task to evaluate the performance of the intelligent system. The performance of the system was compared with results obtained by two retinal experts. Moreover, the system was designed to detect the stage of ROP and presence of plus disease as well as to highlight lesion regions based on an object detection network using Faster-RCNN.ResultsThe system achieved an accuracy of 0.903 for the ROP severity classification. Specifically, the accuracies in discriminating normal, mild, semi-urgent, and urgent were 0.883, 0.900, 0.957, and 0.870, respectively; the corresponding accuracies of the two experts were 0.902 and 0.898. Furthermore, our model achieved an accuracy of 0.957 for detecting the stage of ROP and 0.896 for detecting plus disease; the accuracies in discriminating stage I to stage V were 0.876, 0.942, 0.968, 0.998 and 0.999, respectively.ConclusionsOur system was able to detect ROP and differentiate four-level classification fundus images with high accuracy and specificity. The performance of the system was comparable to or better than that of human experts, demonstrating that this system could be used to support clinical decisions.