A new artificial intelligence (AI) system using deep learning can detect airway nodules in chest CT scans with high accuracy. It identified 75.1% of nodules while keeping false positives low at 0.25 per scan in patients without nodules. The system showed 79% sensitivity in spotting tumorous nodules, even in cases that radiologists often miss.
How the Study Was Done
Researchers analyzed 320 chest CT scans, including 160 from patients with airway nodules and 160 without. The average age was 64 years for patients with nodules and 60 years for those without. They used a deep learning-based AI system, called DL-CAD, and tested its performance using a method called 10-fold cross-validation.
Key Findings
- The AI system detected 75.1% of all airway nodules.
- It had 0.25 false positives per scan in patients without nodules.
- For tumorous nodules, it had 79% sensitivity.
- It showed 78.1% sensitivity for malignant tumors.
- It detected 71.4% of non-tumorous nodules.
- It successfully identified subtle nodules that radiologists may miss.
What This Means for Healthcare
The study shows that the AI system can help radiologists find both benign and cancerous airway nodules in routine chest CT scans. While these nodules are rare, the AI tool could improve early detection and patient care.
Source: medscape
