Echocardiography
Overview
Echocardiography research develops methods to acquire and interpret heart ultrasound, including identifying standard views, segmenting anatomy, tracking cardiac motion, and estimating measures such as ejection fraction and strain. Deep learning approaches range from convolutional and video models to self-supervised foundation models, often incorporating anatomical or physiological constraints; newer systems also combine multiple views, clinical measurements, language models, or robotic probe guidance. These tools could make cardiac assessment more consistent and accessible, but performance can change across hospitals, scanners, patient groups, and measurement conventions. Key challenges include learning from limited or imperfect labels, quantifying uncertainty, and ensuring models attend to clinically meaningful anatomy and motion rather than relying on misleading visual correlations.
Latest papers
59 papers7 papers in the last four weeks, up 40% on the four weeks before. 0.0% of all new papers.