K-numberK251456
Device nameBrightHeart View Classifier
ApplicantBrightheart
Product codeQIH
Device classClass II
Decision dateJun 5, 2025
DecisionSubstantially Equivalent
Regulation892.2050
AI Summary extracted from FDA summary PDF · never regenerated
Intended use

BrightHeart View Classifier is a cloud-based software device that uses artificial intelligence and neural networks to automatically detect and classify standard views of the fetal heart in 2D ultrasound images and video clips. It is intended to assist qualified healthcare professionals (sonographers, physicians, fetal surgeons) during fetal anatomic ultrasound examinations in the second or third trimester of pregnancy by automatically extracting example frames, classifying views into standard categories, and assessing whether documentation satisfies an acquisition protocol.

Technological characteristics

Both subject and predicate devices are cloud-based, software-only medical image management systems operating on DICOM files using neural networks trained to classify fetal heart ultrasound images into standard views. The primary difference is that subject device v1.1 adds an optional web interface for output display, whereas predicate device v1.0 displayed outputs only within a DICOM viewer. Both use identical algorithms and serve the same fundamental purpose of assisting physicians performing fetal ultrasound examinations.

Test standards cited

IEC 62304:2016 (Medical device software — Software life cycle processes). FDA guidance documents on software validation, cybersecurity management, and changes to existing devices were followed; however, no other specific consensus standards (ISO, ASTM) are cited in this summary.

Substantial equivalence argument

Substantial equivalence is based on: (1) identical core algorithm and intended use—both classify fetal heart views from ultrasound images; (2) identical technology platform (cloud-based neural network on DICOM); (3) identical performance metrics validated on 2,290 clinically acquired images (sensitivity 0.939, specificity 0.984); and (4) the addition of a web interface output option does not raise different safety or effectiveness questions because the underlying algorithmic functionality and clinical application remain unchanged. The predicate device's bench testing remains valid because the classification algorithm is identical between versions.

Extracted by AI from the official FDA summary PDF →
Source

View the full FDA submission: accessdata.fda.gov

Researching this as a predicate?
Want a transparent AI-ranking score, AI-discovered related predicates, ongoing safety and warning-letter monitoring, full predicate chain lineage, and a drafted SE rationale — all saved to your own project? That's what an account adds.
Start free trial →

Everything you need for a 510(k) submission. Nothing you don't.

14-day free trial. No setup. Cancel anytime.

Start free trial →
Building an AI or ML-enabled device? Predicate search, PCCP tracking, and AI-specific FDA intelligence — built exclusively for AI/ML devices. Try AIFDA Intel →