Yale School of Medicine (CarDS Lab)
Model weights not public. Contact creators for more information.
Ensemble of 3D convolutional neural networks that detects severe aortic stenosis directly from single-view 2D parasternal long-axis (PLAX) transthoracic echocardiogram videos, without requiring Doppler imaging. Representations are first pretrained with patient-level contrastive self-supervised learning on PLAX clips, then fine-tuned for binary AS classification; the ensemble is externally validated across a temporally-distinct cohort and geographically-distinct cohorts in California and New England. No pretrained weights are released; only the training/evaluation pipeline is public.
Architecture
CNN (3D)
Ensemble of 3D CNN video encoders (self-supervised contrastive pretraining compared against Kinetics-400-initialized and randomly-initialized baselines) fine-tuned for binary classification on PLAX echo clips
Added to catalog
2026-08-10
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.
New England Hospitals Echocardiography Cohort (non-YNHH)
External, geographically-distinct validation cohort of echocardiography studies from New England hospitals outside the Yale-New Haven Health network, used to test generalization of severe aortic-stenosis detection.
Yale-New Haven Hospital PLAX Echocardiography Cohort
5,257 transthoracic echo studies (17,570 PLAX videos) from 2016-2020 used for training/internal validation of severe aortic-stenosis detection, plus a temporally-distinct 2021 test set of 2,040 studies from the same Yale-New Haven Health network.
Severe aortic stenosis presence, from single-view 2D PLAX echo without Doppler