Cedars-Sinai Medical Center (Smidt Heart Institute) / Ouyang Lab
Automated deep learning workflow that detects and grades tricuspid regurgitation (TR) severity from full transthoracic echocardiography studies. The pipeline first identifies apical-4-chamber (A4C) video clips with color Doppler across the tricuspid valve from a full echo study, then applies a dedicated R(2+1)D video classifier to grade TR severity. Trained on over 2 million echo videos from 47,312 studies at Cedars-Sinai Medical Center and externally validated at Stanford Healthcare, the model identified color-Doppler A4C views with AUC ≥0.999 and detected clinically significant (moderate-or-severe) TR with AUC 0.951 and severe TR with AUC 0.980. Code and trained model weights are released to support prospective evaluation of AI-assisted TR screening.
Architecture
CNN (3D)
R(2+1)D 3D-CNN pipeline: a view/Doppler-quality classifier first identifies A4C color-Doppler tricuspid clips from a full echo study, followed by a video classifier that grades TR severity (moderate-or-severe vs. severe) at the study level
Framework
PyTorch
Added to catalog
2026-08-13
Cedars-Sinai Medical Center Echocardiography Dataset
CSMC (Cedars-Sinai Medical Center) clinical echocardiography cohort: 877,983 individual sonographer measurements spanning 9 B-mode and 9 Doppler measurement types, drawn from 155,215 studies.
Stanford Healthcare Echocardiography
Clinical transthoracic echocardiography cohort from Stanford Healthcare (SHC); used to train EchoNet-Labs (70,066 videos / 39,460 patients) and as an external validation set for several EchoNet valvular models. Detailed demographics not reported.
Tricuspid regurgitation severity classification (moderate-or-severe vs. not; severe vs. not) from A4C color-Doppler echocardiography, preceded by automated view/Doppler-quality identification