FDA Clears New AI Algorithm for Early Pneumothorax Detection

The latest clearance marks a significant step in automated triage for emergency departments.

The U.S. Food and Drug Administration has granted 510(k) clearance to a new artificial intelligence algorithm designed to detect pneumothorax — a potentially life-threatening collapsed lung — in chest X-rays with high sensitivity and specificity.

Developed by a team at Stanford Medical AI Lab in collaboration with three Level I trauma centers, the system uses a convolutional neural network trained on over 200,000 annotated chest radiographs. In clinical validation, it achieved 94.2% sensitivity and 97.1% specificity, outperforming the average radiologist in time-critical scenarios.

The algorithm integrates directly into existing PACS workflows, flagging high-priority cases and routing them to the top of radiologist worklists. In emergency settings, where a pneumothorax can progress to tension pneumothorax within minutes, early detection is critical.

Chief Medical Officer Dr. Maya Patel explained that the tool is not designed to replace radiologists but to serve as a safety net — particularly during overnight shifts or in under-resourced facilities. The system also provides a heatmap overlay showing the regions of the image that drove the AI decision.

Following a 90-day pilot at three hospitals, the system reduced time-to-diagnosis for pneumothorax by 38%. The FDA clearance paves the way for broad commercial deployment, with the company targeting emergency departments at over 500 hospitals in the next 18 months.