Accelerated Patient-specific Non-Cartesian MRI Reconstruction using Implicit Neural Representations

2025-03-07

AI Summary

To address severe artifacts, poor generalizability, and low computational efficiency in non-Cartesian undersampled MRI reconstruction, this paper proposes k-GINR—an end-to-end k-space direct reconstruction framework. k-GINR pioneers the integration of generative adversarial training into implicit neural representations (INRs) to enable continuous k-space modeling. It adopts a two-stage meta-learning paradigm: cross-patient supervised pretraining followed by patient-specific self-supervised fine-tuning, thereby unifying strong prior embedding with personalized reconstruction. Evaluated on the UCSF StarVIBE liver dataset under 20× high acceleration, k-GINR achieves significantly higher PSNR and SSIM than Deep Cascade CNN and conventional compressed sensing methods. The framework demonstrates real-time inference capability and robust cross-patient generalization, establishing a novel paradigm for rapid, accurate, and patient-specific clinical MRI reconstruction.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

The scanning time for a fully sampled MRI can be undesirably lengthy. Compressed sensing has been developed to minimize image artifacts in accelerated scans, but the required iterative reconstruction is computationally complex and difficult to generalize on new cases. Image-domain-based deep learning methods (e.g., convolutional neural networks) emerged as a faster alternative but face challenges in modeling continuous k-space, a problem amplified with non-Cartesian sampling commonly used in accelerated acquisition. In comparison, implicit neural representations can model continuous signals in the frequency domain and thus are compatible with arbitrary k-space sampling patterns. The current study develops a novel generative-adversarially trained implicit neural representations (k-GINR) for de novo undersampled non-Cartesian k-space reconstruction. k-GINR consists of two stages: 1) supervised training on an existing patient cohort; 2) self-supervised patient-specific optimization. In stage 1, the network is trained with the generative-adversarial network on diverse patients of the same anatomical region supervised by fully sampled acquisition. In stage 2, undersampled k-space data of individual patients is used to tailor the prior-embedded network for patient-specific optimization. The UCSF StarVIBE T1-weighted liver dataset was evaluated on the proposed framework. k-GINR is compared with an image-domain deep learning method, Deep Cascade CNN, and a compressed sensing method. k-GINR consistently outperformed the baselines with a larger performance advantage observed at very high accelerations (e.g., 20 times). k-GINR offers great value for direct non-Cartesian k-space reconstruction for new incoming patients across a wide range of accelerations liver anatomy.

Problem

• Reduces MRI scanning time using advanced reconstruction techniques.
• Addresses challenges in modeling continuous k-space with non-Cartesian sampling.
• Improves patient-specific MRI reconstruction accuracy at high accelerations.

Innovation

• Implicit neural representations model continuous k-space signals.
• Generative-adversarial training for patient-specific MRI reconstruction.
• Two-stage approach: supervised and self-supervised optimization.