Development of a Novel, Low-Power, Ultrasound Algorithm for the Detection of Pneumothorax Using a Large Animal Model
Steven G Schauer1,2,3, Kendall S Hunter4, Joseph K Maddry5
1US Army Medical Center of Excellence, JBSA Fort, Sam Houston, TX 78234, USA.
Military Medicine
|August 28, 2025
Summary
This study explored using raw radio frequency (RF) ultrasound data to detect pneumothorax, a leading cause of death in combat settings. Findings suggest RF data shows potential for diagnosing pneumothorax, aiding future device development.
Area of Science:
- Medical Imaging
- Ultrasound Technology
- Emergency Medicine
Background:
- Tension pneumothorax is a critical prehospital and combat casualty threat.
- Early detection of pneumothorax before tension physiology is challenging.
- This study investigated raw radio frequency (RF) data from ultrasound for improved detection.
Purpose of the Study:
- To determine if unprocessed raw RF data from a single-crystal ultrasound array can aid in pneumothorax detection.
- To explore the utility of RF data features for classifying pneumothorax in a preclinical model.
Main Methods:
- Prospective enrollment of sus scrofa models with induced pneumothorax.
- Acquisition of thoracic imaging using Verisonics research devices capable of providing raw RF data.
- Feature extraction using principal components analysis (PCA) and classification via linear discriminant analysis (LDA) and logistic regression.
Main Results:
- RF data from Verisonics systems yielded usable data, unlike Clarius systems.
- Principal component 0 (PC0) mean was statistically significant between pre- and post-pneumothorax groups (P=.0472).
- Logistic regression and LDA models achieved 83.3% prediction accuracy for pneumothorax detection.
Conclusions:
- Raw RF ultrasound data shows potential as a diagnostic signal for pneumothorax.
- Findings support the development of low-power devices for improved pneumothorax detection in critical settings.
