ABSTRACT
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. 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 datasets were split into training, validation, and test. Both untrained and pre-trained deep neural networks were employed, including AlexNet and GoogLeNet classifiers, using the Caffe framework. Data augmentation was performed for the presence/absence and low/normal ET tube datasets. Receiver operating characteristic (ROC), area under the curves (AUC), and 95% confidence intervals were calculated. Statistical differences of the AUCs were determined using a non-parametric approach. The pre-trained AlexNet and GoogLeNet classifiers had perfect accuracy (AUC 1.00) in differentiating chest vs. abdominal radiographs, using only 45 training cases. 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. The best-performing network for classifying presence vs. absence of an ET tube was still very accurate with an AUC of 0.99. However, for the most difficult dataset, such as low vs. normal position of the endotracheal tube, DCNNs did not perform as well, but achieved a reasonable AUC of 0.81.
Subject(s)
Intubation, Intratracheal/methods , Neural Networks, Computer , Radiography, Abdominal/classification , Radiography, Thoracic/classification , Area Under Curve , Datasets as Topic , Humans , Intubation, Intratracheal/instrumentation , ROC CurveABSTRACT
Advances in image quality produced by computed tomography (CT) and the growth in the number of image studies currently performed has made the management of incidental pulmonary nodules (IPNs) a challenging task. This research aims to identify IPNs in radiology reports of chest and abdominal CT by Natural Language Processing techiniques to recognize IPN in sentences of radiology reports. Our preliminary analysis indicates vastly different pulmonary incidental findings rates for two different patient groups.
Subject(s)
Decision Support Systems, Clinical/organization & administration , Diagnosis, Computer-Assisted/methods , Machine Learning , Natural Language Processing , Radiography, Abdominal/statistics & numerical data , Radiology Information Systems/supply & distribution , Data Mining/methods , Humans , Illinois/epidemiology , Incidental Findings , Pilot Projects , Radiography, Abdominal/classification , Radiology Information Systems/classification , Reproducibility of Results , Sensitivity and Specificity , Terminology as Topic , Vocabulary, ControlledABSTRACT
The management of follow-up recommendations is fundamental for the appropriate care of patients with incidental pulmonary findings. The lack of communication of these important findings can result in important actionable information being lost in healthcare provider electronic documents. This study aims to analyze follow-up recommendations in radiology reports containing pulmonary incidental findings by using Natural Language Processing and Regular Expressions. Our evaluation highlights the different follow-up recommendation rates for oncology and non-oncology patient cohorts. The results reveal the need for a context-sensitive approach to tracking different patient cohorts in an enterprise-wide assessment.