Skip to main content
All evidence

Using AI to improve radiologist performance in detecting abnormalities on chest radiographs


500 chest X-rays with CT scanner performed within 72 hours from Hôpital Cochin (AP-HP)


  • Gold standard: CT-based annotation by chest radiologist

  • Readers: 4 chest radiologists, 4 general radiologists, 4 radiology residents

  • MRMC study design: reading with and without AI


  • AI-assisted chest radiography interpretation resulted in an increased sensitivity of 5.9 to 26.2 % (P<.001) for all readers including thoracic radiologists, general radiologists, and radiology residents.

Capture Decran 2023 12 18 A 1840 59
  • General radiologists and radiology residents assisted by AI achieved the performance of chest radiologists without AI

  • Mean reading time was 81s without AI vs 56s with AI (-31%, P<.001), with a 17% reduction for radiographs with abnormalities vs 38% for no abnormalities.


AI assistance can improve the detection accuracy of thoracic abnormalities on chest radiographs across radiologists of varying expertise, leading to marked improvements in sensitivity and a reduction in interpretation time.


ChestView AI enhances the detection of urgent findings like pneumothorax, pleural effusion, and consolidation, as well as early cancer indicators such as nodules, and mediastinal mass. Co-developed with AP-HP and grounded on a robust database partly cross-referenced with CT-scan, it is now widely used in private and public facilities worldwide.

Learn more
Chestview V2