- Medical Records (Postnatal subject medical record data)
- Inference Hole (The position information of postnatal cardiac hole)
- AI-assisted Postnatal Cardiac Septal Defects Prediction (CardiaCare is capable of predicting postnatal cardiac septal defects)
The ultrasonography video-recording dataset, comprising both normal and cardiac septal defects conditions—including atrial septal defects (ASDs), ventricular septal defects (VSDs), and atrioventricular septal defects (AVSDs) on apical four-chamber (A4CH), apical five-chamber (A5CH), parasternal long axis (PLAX), parasternal short axis (PSAX), and subcostal (SC).
The framework of CardiaCare is as follows:
- Postnatal cardiac normal-abnormal classification: With AI-assisted, CardiaCare is trained to distinguish between normal and abnormal postnatal cardiac anatomy.
- Standard view classification: With AI-assisted, CardiaCare is trained to classify images based on standard echocardiographic views.
- Cardiac septal defect detection: After a cardiac abnormality is detected and the view is classified, CardiaCare detects specific cardiac defects.
- Decision-making algorithm: The final step integrates medical knowledge to make a precise decision based on the presence, position, and type of defect, as well as the echocardiographic view.