A new deep-learning AI algorithm significantly lowered false positives in lung nodule malignancy assessment while maintaining high detection rates.
Key Details
- 1Deep-learning algorithm estimated lung nodule malignancy risk using LDCT data.
- 2Trained on 16,077 nodules (1,249 malignant) from the NLST; validated on a pooled cohort of 4,146 participants and 7,794 nodules.
- 3AUC for AI model was equal or superior to PanCan: ~0.98 (1yr), 0.96 (2yr) in pooled cohort; up to 0.95 for indeterminate nodules.
- 4For size-matched nodules, AI model AUC was 0.79 vs. PanCan at 0.6.
- 5At 100% sensitivity for cancer, AI flagged 68.1% of benign nodules as low-risk vs. PanCan's 47.4% (39.4% relative reduction in false positives).
- 6Authors stress the need for prospective clinical validation before routine use.
Why It Matters
Reducing false positives in lung cancer screening can minimize unnecessary procedures, patient anxiety, and healthcare costs. Improved AI-based risk stratification stands to improve the efficiency and effectiveness of LDCT lung cancer screening programs.
Source
AuntMinnie
Related News
•AuntMinnie
Study: Computer Vision Models Best LLMs in Chest CT Breast Abnormality Detection
Computer vision models (CVMs) surpass large language models (LLMs) in accurately labeling incidental breast abnormalities on chest CT scans.
•Cardiovascular Business
Radiology Maintains Lead in FDA-Cleared AI Algorithms, Cardiology Follows
Radiology remains the top specialty for FDA-cleared AI, with cardiology as a strong second, particularly in cardiovascular imaging.
•AuntMinnie
Deep Learning Models Rival Radiologists for Pancreatic Cancer Detection on CT
Deep-learning models achieved comparable or superior accuracy to experienced radiologists in detecting pancreatic cancer on CT scans, especially for small tumors.