Evaluation of Deep Learning Model Architectures for Point-of-Care Ultrasound Diagnostics

• Published in: Bioengineering (Basel) Vol. 11; no. 4; p. 392
• Main Authors: Hernandez Torres, Sofia I, Ruiz, Austin, Holland, Lawrence, Ortiz, Ryan, Snider, Eric J
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 18.04.2024
• Subjects: Abdomen; Animal welfare; Artificial intelligence; Automation; Bayesian analysis; Bladder; Catheters; Data augmentation; Decision making; Deep learning; Hemorrhage; Hemothorax; Hospitals; image interpretation; Injuries; Injury prevention; Laboratory animals; Mass casualty incidents; Mathematical models; Medical imaging; medical triage; Military medicine; Model accuracy; Pneumothorax; Trauma; Ultrasonic imaging; Ultrasound; ultrasound imaging; United States > US; ultrasound imaging; deep learning; medical triage; artificial intelligence; medical imaging; image interpretation
• ISSN: 2306-5354; 2306-5354
• DOI: 10.3390/bioengineering11040392

Point-of-care ultrasound imaging is a critical tool for patient triage during trauma for diagnosing injuries and prioritizing limited medical evacuation resources. Specifically, an eFAST exam evaluates if there are free fluids in the chest or abdomen but this is only possible if ultrasound scans can be accurately interpreted, a challenge in the pre-hospital setting. In this effort, we evaluated the use of artificial intelligent eFAST image interpretation models. Widely used deep learning model architectures were evaluated as well as Bayesian models optimized for six different diagnostic models: pneumothorax (i) B- or (ii) M-mode, hemothorax (iii) B- or (iv) M-mode, (v) pelvic or bladder abdominal hemorrhage and (vi) right upper quadrant abdominal hemorrhage. Models were trained using images captured in 27 swine. Using a leave-one-subject-out training approach, the MobileNetV2 and DarkNet53 models surpassed 85% accuracy for each M-mode scan site. The different B-mode models performed worse with accuracies between 68% and 74% except for the pelvic hemorrhage model, which only reached 62% accuracy for all model architectures. These results highlight which eFAST scan sites can be easily automated with image interpretation models, while other scan sites, such as the bladder hemorrhage model, will require more robust model development or data augmentation to improve performance. With these additional improvements, the skill threshold for ultrasound-based triage can be reduced, thus expanding its utility in the pre-hospital setting.