por
Gus Iversen, Editor in Chief | November 18, 2025
3D reconstruction of the airways generated from a CT scan using AI
Researchers at the University of Southampton have developed an AI tool capable of identifying hard-to-see foreign objects lodged in the airways with significantly greater accuracy than experienced radiologists.
Published in
npj Digital Medicine, the study evaluated a deep learning model trained to detect radiolucent foreign body aspiration (FBA) — objects such as plant material or shell fragments that are invisible on X-rays and only faintly detectable on CT scans. These cases are particularly dangerous because they can lead to coughing, choking, breathing difficulties, and severe complications if missed.
To address this diagnostic gap, the team combined a high-precision airway segmentation technique (MedpSeg) with a neural network trained on CT images from over 400 patients across multiple hospitals in China.
Ad Statistics
Times Displayed: 362784
Times Visited: 21094 MIT labs, experts in Multi-Vendor component level repair of: MRI Coils, RF amplifiers, Gradient Amplifiers Contrast Media Injectors. System repairs, sub-assembly repairs, component level repairs, refurbish/calibrate. info@mitlabsusa.com/+1 (305) 470-8013
In a head-to-head comparison involving 70 CT scans, including 14 confirmed FBA cases, the model caught far more FBA cases than radiologists:
Radiologists: Detected 36% of radiolucent foreign bodies
AI model: Detected 71%
F1 score: AI achieved 74% vs. radiologists’ 53%
While radiologists made no false-positive calls, the AI model produced some, as reflected in its 77% precision, yet still dramatically improved overall case detection.
“These objects can be extremely subtle and easy to miss, even for experienced clinicians,” said Zhe Chen, co-first author and Ph.D. researcher at the University of Southampton. “Our AI model acts like a second set of eyes, helping radiologists detect these hidden cases earlier and more reliably.”
Lead author Dr. Yihua Wang emphasized that the technology is designed to support, not replace, radiologists by providing added confidence in difficult or ambiguous cases.
Foreign body aspiration remains a significant clinical challenge: up to 75% of adult FBA cases involve radiolucent objects, leading to frequent misdiagnosis or delays in care.
The team plans to expand the work through larger, multicenter studies to validate performance across more diverse populations and reduce potential bias.
The paper,
“Automated Detection of Radiolucent Foreign Body Aspiration on Chest CT Using Deep Learning,” is available online. The project was supported by the U.K. Medical Research Council and the China Scholarship Council.
