Search papers, labs, and topics across Lattice.
This study introduces HypOProto, a novel hyperbolic ordinal prototype-based framework designed for the classification of Left Ventricular Filling Pressure (LVFP) directly from B-mode echocardiography, addressing the limitations of traditional Doppler-derived methods. By leveraging a frozen, explainable foundation model and a unique Hyperbolic Prototype Angular Separation (HyperPAS) loss, the framework enhances interpretability while achieving state-of-the-art performance in LVFP classification. The hyperbolic arrangement of prototypes not only captures clinically meaningful ordinal relationships but also provides visualizations that highlight critical diagnostic regions, making it a significant advancement in cardiac imaging analysis.
Achieving state-of-the-art LVFP classification while enhancing interpretability, HypOProto transforms how we leverage echocardiography in heart failure diagnostics.
Echocardiography (echo) is a widely used imaging modality for assessing cardiac function, with Left Ventricular Filling Pressure (LVFP) serving as a critical physiological marker for conditions such as heart failure. Standard LVFP classification into normal \emph{vs} elevated categories relies on the Doppler-derived $E/e'$ ratio, which is operator-dependent and often unavailable in resource-limited settings, motivating methods that infer LVFP directly from B-mode echo. Existing deep learning approaches achieve high performance but remain largely black-box, limiting clinical interpretability. We propose HypOProto, a hyperbolic, ordinal prototype-based framework for interpretable LVFP classification using a frozen, explainable foundation model backbone. HypOProto arranges prototypes along the physiological $E/e'$ scale, placing borderline cases near the hyperboloid root where small angular differences separate similar cases, while normal and elevated cases occupy outward positions reflecting increasing diagnostic certainty. This hyperbolic geometry encodes clinically meaningful ordinal relationships and improves interpretability. We also introduce a novel Hyperbolic Prototype Angular Separation (HyperPAS) loss, enforcing inter-class prototype separation in hyperbolic space. HypOProto achieves SOTA performance while maintaining transparency, and highlights clinically relevant regions in visualizations. This work represents the first prototype-based framework for LVFP classification in echo. Our code can be found at https://github.com/DeepRCL/HypOProto.
