Spatio-Temporal Fusion Model for Standard View Classification of Echocardiographic Videos

The authors released the Echocardiographic Videos of Nine Views (EV9V) dataset, the largest publicly available dataset for echocardiographic view classification, containing 5,138 videos and 910,579 frames across 9 standard views. They introduced a Spatio-Temporal Fusion Model (STFM), a dual-stream CNN-LSTM framework that integrates spatial and temporal features while employing uncertainty-aware learning to enhance robustness against frame quality variations. This work is significant for practitioners as it provides a comprehensive dataset and a novel model architecture that can improve classification accuracy in clinical echocardiography applications.