For each case, a resident drafted the report based on the randomly assigned captions and an experienced radiologist finalized the report blinded to the original captions. The participants in the prospective test group were randomly assigned to 1 of 3 different caption generation models: a normal template, NLP-generated captions, and rule-based captions based on convolutional neural networks. The prospective test data set included consecutive images and reports from May 1 to September 30, 2021.Įxposures A bidirectional encoder representation from a transformers model was used to extract language entities and relationships from unstructured CXR reports to establish 23 labels of abnormal signs to train convolutional neural networks. The retrospective test data set included consecutive images and reports from April 1 to July 31, 2019. The training data set included CXR images and reports retrospectively collected from February 1, 2014, to February 28, 2018. Objective To prospectively compare natural language processing (NLP)-generated CXR captions and the diagnostic findings of radiologists.ĭesign, Setting, and Participants A multicenter diagnostic study was conducted. Importance Artificial intelligence (AI) can interpret abnormal signs in chest radiography (CXR) and generate captions, but a prospective study is needed to examine its practical value. Shared Decision Making and Communication.Scientific Discovery and the Future of Medicine.Health Care Economics, Insurance, Payment.Clinical Implications of Basic Neuroscience.Challenges in Clinical Electrocardiography.Multiple Regression Analysis on the Significance of Reporting Time and BLEU Score Among Three Models Classification Performance of Convolutional Neural Networks of Asymptomatic Screening Participants (n=916) in the Prospective Test Using Board Reading as the ReferenceĮTable 7. Classification Performance of Convolutional Neural Networks of Symptomatic Patients (n=4,175) in the Prospective Test Dataset Using Board Reading as the ReferenceĮTable 6. Disease Prevalence in the Prospective Test Dataset (n=5,091) Based on Board ReadingĮTable 5. Classification Performance of Convolutional Neural Networks of Asymptomatic Screening Participants (n=2,130) in the Retrospective Test Dataset Using Board Reading as the ReferenceĮTable 4. Classification Performance of Convolutional Neural Networks of Symptomatic Patients (n=5,996) in the Retrospective Test Dataset Using Board Reading as the ReferenceĮTable 3. Three Representative Cases of Different Report Generation Models and Two Cases in Which Errors Occur in the Prospective Test DatasetĮTable 2. Precision-Recall Curves of Convolutional Neural Network Classification in the Prospective Test DatasetĮFigure 2.

Normal chest x ray findings report