New model unifies shape and texture for improved cardiac video classification

Researchers have developed a new model for cardiac video classification that integrates deformable shape and texture representations. This model uses a bi-directional cross-attention mechanism in its latent space to fuse these features, allowing each modality to adaptively weight the other based on spatio-temporal correspondence. Unlike previous methods that applied uniform weighting across all cardiac phases, this new approach dynamically adjusts the contributions of shape and texture representations over time. The model achieved state-of-the-art performance on a cine cardiac magnetic resonance (CMR) video dataset and improved interpretability through attention mechanisms that identify critical cardiac phases and modality contributions. AI

IMPACT This research advances medical imaging analysis by improving the accuracy and interpretability of cardiac video classification models.

RANK_REASON The cluster contains an academic paper detailing a novel model for cardiac video classification. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• arXiv
• Attention Mechanism
• Bi-directional Cross-Attention
• cardiac image classification
• cardiac video classification
• cine cardiac magnetic resonance (CMR) video dataset
• computer science
• Computer vision and pattern recognition
• Deep Neural Networks
• deformable shape representations
• Texture Features and PDL1 in CT-PET 18 FDG

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →