DEEPECHO

DEEPECHO is a cloud-based machine learning software designed to assist healthcare professionals in analyzing fetal ultrasound images for biometric measurements and anatomical view identification. It utilizes advanced machine learning techniques validated through non-clinical testing against expert annotations, demonstrating high accuracy and robust performance across diverse patient and equipment parameters, thus serving as a safe and effective concurrent reading and measuring aid.

Deep Echo has developed DEEPECHO, a cloud-based software designed to assist qualified healthcare professionals in analyzing fetal ultrasound images and clips. This software utilizes machine learning techniques to detect standard biometry views, such as cephalic, abdominal, and femoral, and then helps compute various fetal biometric parameters including head circumference, abdominal circumference, femur length, and bi-parietal diameter, as well as estimated fetal weight and gestational age. It is intended for use by trained professionals as a concurrent reading and measuring aid for pregnant patients aged 18 and older, between 14 and 41 weeks of pregnancy, supporting both post-exam image file analysis and real-time image streaming.To evaluate its performance, DEEPECHO underwent comprehensive non-clinical testing, comparing its analysis to ground truth annotations by expert sonographers across a diverse dataset of ultrasound studies. The primary endpoints included the accuracy of continuous measurements for key fetal parameters and the sensitivity and specificity of anatomical view identification. The results demonstrated high accuracy for the biometric measurements and robust sensitivity and specificity in detecting the required views. Subgroup analyses confirmed consistent performance across different maternal ages, gestational ages, BMI categories, and ultrasound system manufacturers, supporting the conclusion that DEEPECHO is a safe and effective tool.

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 DEEPECHO's performance compare to manual measurements by multiple sonographers in a diverse clinical setting, beyond the non-clinical testing environment?; What are the specific machine learning models and algorithms employed by DEEPECHO for view detection and biometric parameter computation, and how are they robust against variations in image quality or atypical fetal presentations?; Are there any known limitations or specific scenarios (e.g., multifetal pregnancies, certain anomalies) where DEEPECHO's accuracy might be compromised, and how are these addressed?; What kind of training or certification is required for healthcare professionals to effectively use and interpret DEEPECHO's output, and how does the software integrate into existing clinical workflows?; What is the long-term data collection strategy for DEEPECHO, and how will continuous learning or updates ensure sustained accuracy and adaptation to evolving clinical standards or new ultrasound technologies?