Key result
SGS EchoNet matches state-of-the-art LVEF estimation without requiring a ventricular segmentation model.
Why the study?
Segmentation-dependent deep learning models for transthoracic echocardiograms can be excessively data hungry and sensitive to common data quality issues.
Does the Scaled Gumbel Softmax (SGS) EchoNet deep learning model improve ejection fraction estimation and robustness to noise in transthoracic echocardiograms compared to segmentation-based models?
Does the Scaled Gumbel Softmax (SGS) EchoNet deep learning model improve ejection fraction estimation and robustness to noise in transthoracic echocardiograms compared to segmentation-based models?
The SGS EchoNet model provides a data-efficient and robust method for estimating ejection fraction from echocardiograms without requiring labor-intensive ventricular segmentation.
Answers are grounded in this paper and cite their sources.
Supports segmentation-free LVEF estimation; leaves open prospective clinical validation before adoption.
Video-based deep learning (DL) algorithms often rely on segmentation models to detect clinically important features in transthoracic echocardiograms (TTEs). While effective, these algorithms can be too data hungry for practice and may be sensitive to common data quality issues. To overcome these concerns, we present a data-efficient DL algorithm, Scaled Gumbel Softmax (SGS) EchoNet, that is robust to these common data quality issues and, importantly, requires no ventricular segmentation model. In lieu of a segmentation model, we decompose and transform the output of an R(2 + 1)D convolutional encoder to estimate frame-level weights associated with the cardiac cycle, that are then used to obtain a video representation that can be used for estimation. We find that our transformation obviates the need for a segmentation model while improving the ability of the predictive model to handle noisy inputs. We show that our model achieves comparable performance to the state of the art, while demonstrating robustness to noise on an independent (external) validation set. Not applicable.
No takes yet. Share an insight, caveat, or question.
Artman et al. (2025) studied Cardiomyopathy (n=11,074). Scaled Gumbel Softmax (SGS) EchoNet vs. EchoNet Dynamic (with segmentation) was evaluated on Ejection fraction estimation performance (MAE, RMSE, Pearson's r). The Scaled Gumbel Softmax (SGS) EchoNet algorithm estimated left ventricular ejection fraction from echocardiograms without requiring a ventricular segmentation model, achieving comparable performance to state-of-the-art methods.
Synapse has enriched 10 related papers. Consider them for comparative context: