The SemanticST framework outperformed state-of-the-art methods in segmentation accuracy, temporal consistency, and ejection fraction estimation correlation on the CAMUS and EchoNet-Dynamic datasets.
Does the SemanticST framework improve left ventricular ejection fraction estimation and segmentation accuracy in echocardiography compared to existing methods?
The proposed SemanticST framework improves automated left ventricular ejection fraction estimation from echocardiograms by integrating spatio-temporal modeling with clinical semantic priors.
Estimating left ventricular Ejection Fraction (EF) from echocardiography is critical for cardiac systolic function assessment and clinical risk stratification. Unfortunately, existing EF estimation methods are limited by (i) insufficient modeling of temporal clues embedded in video frames and (ii) inadequate exploitation of clinical semantics in easily accessible textual reports. In this study, we propose a unified segmentation and EF estimation framework with Semantics-enhanced spatio-temporal modeling (named SemanticST), integrating spatio-temporal consistency modeling with structured clinical semantic priors. Our SemanticST introduces a spatio-temporal & text-guided neighborhood correlation mining (STT-NCM) encoder, which captures both short- and long-range temporal dependencies via text-modulated 3D neighborhood attention. Further, a text-guided pixel-level semantic projection module (TextSP) is designed to map the key clinical cues extracted by a large language model (LLM) into pixel-level guidance features, enabling the alignment of semantic priors with visual context for optimized EF estimation. Extensive experiments on two public datasets (CAMUS and EchoNet-Dynamic) demonstrate that our SemanticST outperforms state-of-the-art methods in segmentation accuracy, temporal consistency, and EF estimation correlation.
Zheng et al. (Thu,) conducted a other in Ejection Fraction estimation in echocardiography. SemanticST (Semantics-enhanced spatio-temporal modeling) vs. state-of-the-art methods was evaluated on segmentation accuracy, temporal consistency, and EF estimation correlation. The SemanticST framework outperformed state-of-the-art methods in segmentation accuracy, temporal consistency, and ejection fraction estimation correlation on the CAMUS and EchoNet-Dynamic datasets.