Deep learning to detect acute respiratory distress syndrome on chest radiographs: a retrospective study with external validation

• Published in: The Lancet. Digital health 2021-06, Vol.3 (6), p.e340-e348
• Main Authors: Sjoding, Michael W, Taylor, Daniel, Motyka, Jonathan, Lee, Elizabeth, Co, Ivan, Claar, Dru, McSparron, Jakob I, Ansari, Sardar, Kerlin, Meeta Prasad, Reilly, John P, Shashaty, Michael G S, Anderson, Brian J, Jones, Tiffanie K, Drebin, Harrison M, Ittner, Caroline A G, Meyer, Nuala J, Iwashyna, Theodore J, Ward, Kevin R, Gillies, Christopher E
• Format: Article
• Language: English
• ISSN: 2589-7500; 2589-7500
• DOI: 10.1016/S2589-7500(21)00056-X

Acute respiratory distress syndrome (ARDS) is a common, but under-recognised, critical illness syndrome associated with high mortality. An important factor in its under-recognition is the variability in chest radiograph interpretation for ARDS. We sought to train a deep convolutional neural network (CNN) to detect ARDS findings on chest radiographs. CNNs were pretrained on 595 506 radiographs from two centres to identify common chest findings (eg, opacity and effusion), and then trained on 8072 radiographs annotated for ARDS by multiple physicians using various transfer learning approaches. The best performing CNN was tested on chest radiographs in an internal and external cohort, including a subset reviewed by six physicians, including a chest radiologist and physicians trained in intensive care medicine. Chest radiograph data were acquired from four US hospitals. In an internal test set of 1560 chest radiographs from 455 patients with acute hypoxaemic respiratory failure, a CNN could detect ARDS with an area under the receiver operator characteristics curve (AUROC) of 0·92 (95% CI 0·89–0·94). In the subgroup of 413 images reviewed by at least six physicians, its AUROC was 0·93 (95% CI 0·88–0·96), sensitivity 83·0% (95% CI 74·0–91·1), and specificity 88·3% (95% CI 83·1–92·8). Among images with zero of six ARDS annotations (n=155), the median CNN probability was 11%, with six (4%) assigned a probability above 50%. Among images with six of six ARDS annotations (n=27), the median CNN probability was 91%, with two (7%) assigned a probability below 50%. In an external cohort of 958 chest radiographs from 431 patients with sepsis, the AUROC was 0·88 (95% CI 0·85–0·91). When radiographs annotated as equivocal were excluded, the AUROC was 0·93 (0·92–0·95). A CNN can be trained to achieve expert physician-level performance in ARDS detection on chest radiographs. Further research is needed to evaluate the use of these algorithms to support real-time identification of ARDS patients to ensure fidelity with evidence-based care or to support ongoing ARDS research. National Institutes of Health, Department of Defense, and Department of Veterans Affairs.