Enhancing Implicit Neural Representations with Image Feature Embedding for Unsupervised Cardiac Cine MRI Reconstruction

2026-07-04

AI Summary

This work addresses the degradation in image quality caused by k-space undersampling in unsupervised cine cardiac MRI reconstruction by proposing I-FP-INR, a novel image-domain dual-branch implicit neural representation framework. For the first time, an image feature embedding mechanism is integrated into implicit neural representations, enabling a backbone branch and a feature-processing branch to collaboratively learn complementary representations. This design enhances model expressiveness without requiring fully sampled reference data. By incorporating coil sensitivity encoding and an unsupervised learning strategy, I-FP-INR consistently outperforms existing baseline methods across multiple public and internal datasets, achieving superior reconstruction quality and demonstrating robust performance under varying sampling rates and diverse clinical scenarios.

Abstract

Cardiac cine Magnetic Resonance Imaging (MRI) is a critical diagnostic tool that provides dynamic insights for radiologists. To accelerate acquisition, under-sampled k-space data is often used, requiring reconstruction methods that combine coil sensitivity encoding with prior information to recover missing data. Deep learning approaches have gained more attention for leveraging data-adaptive priors. While supervised learning approaches are a common choice, they depend on fully sampled reference data, which is not always available. Unsupervised methods eliminate the need for fully sampled reference data, which can be advantageous in cardiac cine MRI reconstruction. Among them, implicit neural representations (INRs) have shown great potential due to their simple architecture and good quality reconstructions. In this work, we propose an image-domain dual-branch INR framework, termed I-FP-INR, which extends the original INR design by introducing an additional feature-processing branch. This design aims to extract complementary feature embeddings to enhance the overall representation, thereby benefiting reconstruction. Extensive evaluations on both public datasets and in-house data show consistent improvements over baseline methods in reconstruction quality, with strong robustness across varied scenarios.

Problem

• unsupervised
• cardiac cine MRI
• reconstruction
• implicit neural representations
• under-sampled k-space

Innovation

• Implicit Neural Representations
• Image Feature Embedding
• Unsupervised MRI Reconstruction
• Cardiac Cine MRI
• Dual-Branch Framework