Fetal EchoScan (v1.1)

Fetal EchoScan v1.1 is a cloud-based machine learning software device that analyzes fetal heart ultrasound images from pregnant women (18+ years old) during second-trimester anatomy scans (18-24 weeks gestation). It detects eight types of suspicious cardiac findings (e.g., overriding artery, septal defects, valve abnormalities) and displays results as annotated images in DICOM viewers or an optional web interface to aid interpreting physicians (OB-GYNs, maternal-fetal medicine specialists) in diagnosis.

The subject device v1.1 differs from predicate v1.0 only in output display: v1.1 adds an optional web interface display of results in addition to annotated DICOMs in a PACS viewer, while v1.0 outputs only annotated DICOMs. Both use identical machine learning algorithms, accept the same ultrasound video inputs (4-chamber, left and right ventricular outflow tract views), categorize findings into classification and measurement features, and produce frame-level and summary-level outputs evaluating findings as present, absent, or inconclusive.

IEC 62304:2016 (Medical device software — Software life cycle processes); FDA Guidance for Content of Premarket Submissions for Software Contained in Medical Devices; FDA Guidance for Management of Cybersecurity in Medical Devices.

The core argument is that v1.1 shares the exact same detection algorithm as predicate v1.0, making bench testing data from v1.0 fully applicable to v1.1. The only change is the optional addition of a web interface for result display, which uses similar output presentation compared to the annotated DICOMs. Since both devices employ identical machine learning models, identical input requirements, and identical clinical outputs (presence/absence/inconclusive determinations), the web interface—a presentation-layer enhancement with no impact on the algorithm's diagnostic logic—does not raise different safety or effectiveness questions. Reader studies and standalone validation testing confirmed diagnostic performance improvements for physicians using the device versus unaided interpretation.