![]() Recent studies using feature extraction and classification with support vector machines have resulted in area under the curves (AUC) of 0.88 and 0.94, respectively, for detection of ET tubes. As such, there has been interest in using computer-aided detection (CAD) methods to facilitate detection of ET tubes. Increased mortality and pneumonia have also been reported with low positioning of the tube into the bronchi. ![]() There are important consequences of ET tube malposition-a low insertion of the tube into the main stem bronchus can lead to hyperinflation of one lung and pneumothorax, and atelectasis, and hypoxemia of the contralateral non-ventilated lung. One of the goals of a trained radiologist is to accurately describe the presence and position of an endotracheal (ET) tube on chest radiography. normal position of the endotracheal tube, DCNNs did not perform as well, but achieved a reasonable AUC of 0.81. However, for the most difficult dataset, such as low vs. absence of an ET tube was still very accurate with an AUC of 0.99. The best-performing network for classifying presence vs. For more difficult datasets, including the presence/absence and low/normal position endotracheal tubes, more training cases, pre-trained networks, and data-augmentation approaches were helpful to increase accuracy. abdominal radiographs, using only 45 training cases. The pre-trained AlexNet and GoogLeNet classifiers had perfect accuracy (AUC 1.00) in differentiating chest vs. Statistical differences of the AUCs were determined using a non-parametric approach. Receiver operating characteristic (ROC), area under the curves (AUC), and 95% confidence intervals were calculated. Data augmentation was performed for the presence/absence and low/normal ET tube datasets. Both untrained and pre-trained deep neural networks were employed, including AlexNet and GoogLeNet classifiers, using the Caffe framework. The datasets were split into training, validation, and test. Three different datasets were created, which included presence/absence of the endotracheal (ET) tube ( n = 300), low/normal position of the ET tube ( n = 300), and chest/abdominal radiographs ( n = 120). The goal of this study is to evaluate the efficacy of deep convolutional neural networks (DCNNs) in differentiating subtle, intermediate, and more obvious image differences in radiography.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |