No Image, No Problem: End-to-End Multi-Task Cardiac Analysis from Undersampled k-Space

2026-03-10

AI Summary

This work proposes k-MTR, a novel framework that circumvents the conventional “reconstruct-then-analyze” pipeline in cardiac magnetic resonance imaging, which is prone to artifacts and information bottlenecks. By aligning the semantics of undersampled k-space data and fully sampled images directly in a latent space, k-MTR enables end-to-end recovery of anatomical structures and concurrent multi-task analysis from raw frequency-domain data—without explicit image reconstruction. Integrating k-space representation learning, latent semantic alignment, and multi-task learning, the method demonstrates strong performance across 42,000 simulated cases, achieving accuracy on par with state-of-the-art image-domain approaches in tasks including phenotypic regression, disease classification, and anatomical segmentation. These results establish that high-fidelity diagnostic inference can be performed directly from k-space, eliminating the need for intermediate image reconstruction.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data

Application Category

• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Conventional clinical CMR pipelines rely on a sequential "reconstruct-then-analyze" paradigm, forcing an ill-posed intermediate step that introduces avoidable artifacts and information bottlenecks. This creates a fundamental mathematical paradox: it attempts to recover high-dimensional pixel arrays (i.e., images) from undersampled k-space, rather than directly extracting the low-dimensional physiological labels actually required for diagnosis. To unlock the direct diagnostic potential of k-space, we propose k-MTR (k-space Multi-Task Representation), a k-space representation learning framework that aligns undersampled k-space data and fully-sampled images into a shared semantic manifold. Leveraging a large-scale controlled simulation of 42,000 subjects, k-MTR forces the k-space encoder to restore anatomical information lost to undersampling directly within the latent space, bypassing the explicit inverse problem for downstream analysis. We demonstrate that this latent alignment enables the dense latent space embedded with high-level physiological semantics directly from undersampled frequencies. Across continuous phenotype regression, disease classification, and anatomical segmentation, k-MTR achieves highly competitive performance against state-of-the-art image-domain baselines. By showcasing that precise spatial geometries and multi-task features can be successfully recovered directly from the k-space representations, k-MTR provides a robust architectural blueprint for task-aware cardiac MRI workflows.

Problem

• undersampled k-space
• cardiac MRI
• image reconstruction
• multi-task analysis
• diagnostic bottleneck

Innovation

• k-space representation learning
• end-to-end cardiac analysis
• multi-task learning
• undersampled MRI
• latent semantic alignment