Related Experiment Video
Updated: Jun 26, 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
An End-to-end Interactive Software for Pediatric Pneumonia Severity Assessment from Lung Ultrasound Images
Shuo Li1, Xue Yang2, Yue Jiang2
1Department of Ultrasound MedicineBeijing Chaoyang Hospital, Capital Medical UniversityBeijingChina.
Ultrasound International Open
|June 25, 2026
Summary
This study developed interactive software integrating lung ultrasound radiomics and machine learning for objective pediatric pneumonia severity assessment. The tool achieved high accuracy, aiding clinical decision-making at the point of care.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Pediatric Pulmonology
Background:
- Pediatric pneumonia diagnosis relies on clinical and imaging findings, often subjective.
- Objective assessment tools are needed for accurate severity grading and timely treatment.
Purpose of the Study:
- To develop and validate an interactive software for objective pediatric pneumonia severity assessment.
- Integrate lung ultrasound radiomics and machine learning at the point of care.
Main Methods:
- Retrospective analysis of 293 pediatric lung ultrasound images.
- Extracted 104 radiomics features, selected using LASSO regression.
- Developed and evaluated 10 machine learning algorithms, with optimal model interpretation via SHapley Additive exPlanations.
Main Results:
- The Light Gradient Boosting Machine classifier achieved 89.8% accuracy, 91.7% sensitivity, and 88.6% specificity.
- SHapley Additive exPlanations identified morphological characteristics as key predictive features.
- The finalized classifier was integrated into user-friendly interactive software.
Conclusions:
- An end-to-end interactive software successfully uses lung ultrasound radiomics and machine learning for objective pediatric pneumonia severity assessment.
- This tool offers potential for standardized diagnosis and improved clinical decision-making in real-world settings.
