Deep Learning for Chest Radiograph Diagnosis in the Emergency Department

Abstract

I n 2015, approximately 137 million patients presented to the emergency department (ED) in the United States (43.3 visits per 100 persons) (1). Respiratory diseases were the second most common primary diagnosis in these patients, accounting for 9.8% of all visits (1). Chest radiography is the first-line examination for the evaluation of various thoracic diseases (2-8). The number of chest radiographs per ED visit increased by 81% between 1994 and 2014, suggesting an increasing dependency on chest radiographs (9). The interpretation of chest radiographs is a challenging task, requiring experience and expertise. Previous studies have reported suboptimal performance in the interpretation of chest radiographs by ED physicians compared with expert radiologists (10-13). In addition, the American College of Radiology recommends that qualified radiologists be available to interpret all radiographs obtained in the ED (14). However, there is a practical limitation with regard to the full-time availability of expert radiologists, especially for after-hours coverage. In a 2014 survey (15), 73% of the academic radiology departments in the United States did not provide overnight coverage by faculty. Thus, for after-hours ED coverage, a computer-aided detection system for clinically relevant findings on chest radiographs may help improve the quality of radiographic interpretation and overall turnaround time. Recently, deep learning (DL) algorithms with medical image analysis systems have been evaluated for retinal fundus photographs (16,17), pathologic images (18), and

Implementation of a Deep Learning-Based Computer-Aided Detection System for the Interpretation of Chest Radiographs in Patients Suspected for COVID-19

“…Although we were unable to determine the baseline performance of radiologists without use of the CAD system, it may enhance sensitivity for the detection of pulmonary abnormalities. In a previous study using the same CAD system, the sensitivity of radiology residents for identification of abnormal CXR were significantly enhanced (65.6-73.4%) after review of the CAD results (28).…”

Artificial Intelligence-Based Diagnosis of Cardiac and Related Diseases

“…Automatic pulmonary disease detection using computer-aided diagnosis (CAD) is based on the correct segmentation of anatomical structures, such as the lungs, heart, and clavicle bones [2]. With the success of deep learning, artificially intelligent algorithms can help medical experts and ophthalmologists to detect and diagnose the disease and increase diagnostic throughput [14][15][16][17][18][19][20]. Semantic segmentation is a special branch of deep learning that involves pixel-wise classification of the image, which is important to accurately locate the infected areas for disease analysis [21,22].…”

Preparing Medical Imaging Data for Machine Learning

“…Workflow improvements include prioritizing worklists for radiologists (2,3), triaging screening mammograms (4), reducing or eliminating gadolinium-based contrast media for MRI (5,6), and reducing the radiation dose of CT imaging by advancing image noise reduction (7)(8)(9). Automatic lesion detection by using machine learning has been applied to many imaging modalities and includes detection of pneumothorax (10,11), intracranial hemorrhage (12), Alzheimer disease (13), and urinary stones (14). Automatic quantification of medical images includes assessing skeletal maturity on pediatric hand radiographs (15), coronary calcium scoring on CT images (16), prostate classification at MRI (17), breast density at mammography (18), and ventricle segmentation at cardiac MRI (19,20).…”