New research finds increased patient BMI can negatively impact AI performance on low-dose CT scans.
Key Details
- 1Study published in the European Journal of Radiology examines BMI impact on chest LDCTs.
- 2Higher BMI causes increased X-ray attenuation and image noise, degrading image quality.
- 3Researchers compared AI and human lung nodule detection in cohorts with highest and lowest BMI (top and bottom 1.5%).
- 4AI's diagnostic accuracy is limited by the diversity of its training datasets, particularly in BMI distribution.
- 5Prior literature shows conflicting findings about BMI influence on human performance; data on AI's vulnerability to BMI-related artifacts is growing.
Why It Matters
This study highlights a potential bias and limitation of current imaging AI tools, especially those trained on less diverse patients. Understanding performance differences across BMI ranges is essential for ensuring equitable and accurate lung cancer screening in clinical settings.
Source
Radiology Business
Related News
•Radiology Business
Meta Funds Ohio State’s Radiology AI Innovation Projects
Meta will fund Ohio State researchers to explore AI-driven advancements in radiology, including chest X-ray reporting and disease prediction.
•AuntMinnie
ChatGPT-5 Performs Poorly at Detecting Distal Radius Fractures on X-rays
ChatGPT-5 missed about half of distal radius fractures on x-rays, showing limited diagnostic capability in a real-world scenario.
•AuntMinnie
AI Triage Tool Speeds Up ICH Head CT Turnaround in ED
AI-based triage for intracranial hemorrhage on head CT reduced emergency department report turnaround by 13 minutes.