Letter by Goto and Goto Regarding Article, "Fully Automated Echocardiogram Interpretation in Clinical Practice: Feasibility and Diagnostic Accuracy"
Letter by Goto and Goto Regarding Article, "Fully Automated Echocardiogram Interpretation in Clinical Practice: Feasibility and Diagnostic Accuracy"
复制标题
Goto 和 Goto 关于文章“临床实践中的全自动超声心动图解读：可行性和诊断准确性”的信函
DOI：
10.1161/circulationaha.118.038451
复制
发表时间：
2019
期刊：
影响因子：
37.8
通讯作者：
Goto S and Goto S
中科院分区：
文献类型：
--
作者：
S. Goto;H. Oka;K. Ayabe;H. Yabushita;M. Nakayama;T. Hasebe;H. Yokota;S. Takagi;M. Sano;and A. Tomita;Goto Shinya;Goto S and Goto S
影像研究是用于诊断的强大工具；然而，对这些模式的解释需要特殊的培训。在大多数情况下，如果没有训练有素的物理学家的解释，就不可能做出诊断或预测预后。涉及二维卷积神经网络的深度学习方法已经成为从原始图像中提取有意义数据的强大工具。关于这种方法在医学领域的成功应用已有多篇报道。1、2然而，该应用的目标主要限于静止图像。这主要是因为二维卷积神经网络的性质，它需要二维数据作为输入。然而，临床上可用的超声心动图数据不是二维图像，而是具有时间序列的视频。因此，二维卷积神经网络在超声心动图中的应用受到了限制。
Imaging studies are powerful tools to be used in making a diagnosis; however, the interpretation of these modalities requires special training. In most cases, it is not possible to make a diagnosis or predict prognosis without interpretations by trained physicians.The deep-learning method involving 2-dimensional convolutional neural networks has emerged as a powerful tool to extract meaningful data from raw images. There have been multiple reports of the successful application of this method to medical fields. 1, 2 However, the targets of the application have mainly been restricted to still images. This is mainly because of the nature of the 2-dimensional convolutional neural networks, which requires 2-dimensional data as input. However, clinically available echocardiography data are not 2-dimensional images but videos with time sequence. Therefore, application of 2-dimensional convolutional neural networks in echocardiography has been limited.