HERA W10 Diagnostic Ultrasound System; HERA W9 Diagnostic Ultrasound System

The HERA W9 and HERA W10 Diagnostic Ultrasound Systems provide versatile, multi-modal imaging for diverse clinical applications by employing established ultrasound technology coupled with advanced software functionalities. These systems are enhanced with a new EV2-12 transducer and innovative AI-driven features, BiometryAssist and ViewAssist, which have been validated through non-clinical testing to automate and improve the accuracy of fetal biometry and anatomical view recognition.

The HERA W9 and HERA W10 are mobile, software-controlled diagnostic ultrasound systems designed for a comprehensive range of clinical applications, including fetal/obstetrics, abdominal, gynecological, pediatric, cardiac, and vascular imaging, among others. These systems acquire and display ultrasound data using multiple modes such as 2D, M-mode, various Doppler modalities (Pulsed Wave, Continuous Wave, Color, Power, Tissue), 3D/4D imaging, and ElastoScan+™, to facilitate the measurement of anatomical structures and provide analysis packages that aid healthcare professionals in clinical diagnosis. The devices operate on established ultrasound principles, incorporating real-time display of mechanical and thermal indices for patient safety monitoring. This iteration of the HERA W9/W10 systems introduces key software enhancements, notably an expanded BiometryAssist AI for improved fetal biometry measurements (achieving ≤8% error for circumference) and an enhanced ViewAssist AI for automated view recognition (94.70% accuracy) and anatomical annotation. Non-clinical validation studies, utilizing deep learning algorithms on diverse datasets from South Korea and the United States, demonstrated high performance for these AI features, with average segmentation dice scores of 0.91 for BiometryAssist and 0.875 for ViewAssist. Hardware updates include the introduction of a new EV2-12 transducer, validated for biocompatibility and image performance, and the option of a 27-inch OLED monitor, further broadening the systems' capabilities and adherence to safety and performance standards.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the AI performance (BiometryAssist, ViewAssist) vary with image quality, operator experience, and in pregnancies with multiple fetuses or rare anomalies not extensively represented in the training data?; What are the measured improvements in examination time or inter-observer variability for fetal biometry and cardiac view assessment when using the AI-assisted features compared to manual methods?; Can the raw data or confidence scores from the AI segmentation and classification (ViewAssist, BiometryAssist) be accessed or visualized by the clinician to understand the basis of the AI's output, especially in ambiguous cases?; What is the performance and clinical utility of the new EV2-12 transducer across its specified applications, particularly in terms of image resolution, penetration, and diagnostic confidence compared to existing Samsung transducers?; How are the AI models (BiometryAssist, ViewAssist) maintained and updated over time to account for new imaging guidelines, evolving clinical practices, or to improve performance on underrepresented populations or conditions?