Related Experiment Video
Updated: Jan 18, 2026
06:15
Protocol and Guidelines for Point-of-Care Lung Ultrasound in Diagnosing Neonatal Pulmonary Diseases Based on International Expert Consensus
Published on: March 6, 2019
51.9K
Automatic approach for B-lines detection in lung ultrasound images using You Only Look Once algorithm.
Alberto Bottino1, Chiara Botrugno2,3, Ernesto Casciaro2
1Department of Innovation Engineering, University of Salento, Lecce, Italy.
Journal of Ultrasound
|September 11, 2025
Summary
A new AI algorithm accurately detects B-lines in Lung Ultrasound (LUS) images, improving diagnostic support. This tool assists clinicians, especially non-experts, in interpreting LUS for better respiratory management.
Area of Science:
- Medical imaging analysis
- Artificial intelligence in healthcare
- Pulmonary diagnostics
Background:
- B-lines are crucial ultrasound artifacts for diagnosing lung conditions.
- Manual detection and quantification of B-lines are challenging and time-consuming.
- Automated tools are needed to enhance Lung Ultrasound (LUS) interpretation consistency and efficiency.
Purpose of the Study:
- To evaluate a YOLO-based algorithm for automated B-line detection in LUS.
- To assess the algorithm's performance against expert sonographer annotations.
- To provide a novel tool for supporting clinical decision-making in respiratory care.
Main Methods:
- An observational agreement study using 386 LUS images from 46 patients.
- Ground truth established by blinded expert sonographer identification of B-lines.
- Algorithm performance measured by Precision, Recall, F1-score, and weighted kappa (kw) statistics.
Main Results:
- The YOLO-based algorithm achieved high performance metrics: Precision 0.92, Recall 0.81, F1-score 0.86.
- Weighted kappa of 0.68 indicated substantial agreement between the algorithm and expert annotations.
- The algorithm demonstrates reliable B-line detection capabilities.
Conclusions:
- The developed algorithm shows significant potential for enhancing diagnostic support in LUS.
- Accurate B-line detection by the algorithm can improve LUS interpretation efficiency and consistency.
- This AI tool can aid clinicians, particularly non-experts, in respiratory management.
