Cardiovascular AI
Clinical Trial
Grant Funded ($1.8M)
CardioVision-Net: Real-Time Echocardiogram Segmentation
Lead Investigator: Dr. Sarah Chen
Executive Summary
Developing lightweight neural networks deployable on point-of-care cardiac ultrasound devices for low-resource clinics and emergency response teams.
Technology Stack & Domain Frameworks
PyTorch
Ultrasound
Real-Time Inference
Methodology & Clinical Validation
Project Overview:
CardioVision-Net addresses the acute challenge of rapid ejection fraction estimation and cardiac chamber segmentation directly on portable ultrasound hardware.
Key Innovations:
1. Lightweight 2D+T temporal convolution modules operating at 60 FPS on edge hardware.
2. Anatomical shape priors enforcing valid left ventricle chamber contours under noisy ultrasound acoustic shadows.
3. Multi-center clinical validation across 4 tertiary hospitals.