Fetal Imaging Demonstration of a Real-Time, Autonomous Acoustic Output Selection Method on a Clinical Ultrasound System
Monday, April 8, 2024
5:27pm – 5:37pm
Location: 406
Authors: Matthew Huber, Duke University David Bradway, Duke University Patricia McNally, Duke Sarah Ellestad, Duke University Gregg Trahey, Duke University ,
This work showcases a potential clinical tool that automatically selects recommended acoustic output levels, minimizing the visual impact of noise in ultrasound acquisitions. During live imaging, the system adapts the Mechanical Index (MI) in real-time as the target conditions or view change. This tool would streamline the highly user-dependent ultrasound imaging process. Further, the acoustic output automatically follows the ALARA (As Low As Reasonably Achievable) principle, balancing image quality considerations with acoustic exposure concerns. During fetal abdomen imaging tests in 11 pregnant volunteers, average recommended MI was 0.16-0.25, depending on the imaging frequency. This represents a significant reduction from the potential 0.8-1.2 MI often used on clinical scanners. During live imaging, noise conditions changed as the target structure or transducer moved, but the MI recommended by the system adapted, maintaining a stable image quality level. Different MI levels were required across volunteers to reach the target signal quality level, emphasizing the advantage of adapting to the variety of patients that might be encountered in the clinic. Altogether, this tool promises to improve the imaging experience for the sonographer, while demonstrating an approach to parameter adjustment that could lead to further automated adjustments on clinical ultrasound systems.