NLCG-Net: A Model-Based Zero-Shot Learning Framework for Undersampled Quantitative MRI Reconstruction
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918264207048704 |
|---|---|
| author | Jiang, Xinrui Jun, Yohan Cho, Jaejin Gao, Mengze Yong, Xingwang Bilgic, Berkin |
| author_facet | Jiang, Xinrui Jun, Yohan Cho, Jaejin Gao, Mengze Yong, Xingwang Bilgic, Berkin |
| contents | Typical quantitative MRI (qMRI) methods estimate parameter maps in a two-step pipeline that first reconstructs images from undersampled k-space data and then performs model fitting, which is prone to biases and error propagation. We propose NLCG-Net, a model-based nonlinear conjugate gradient (NLCG) framework for joint T2/T1 estimation that incorporates a U-Net regularizer trained in a scan-specific, zero-shot fashion. The method directly estimates qMRI maps from undersampled k-space using mono-exponential signal modeling with scan-specific neural network regularization, enabling high-fidelity T1 and T2 mapping. Experimental results on T2 and T1 mapping demonstrate that NLCG-Net improves estimation quality over subspace reconstruction at high acceleration factors. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_12004 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | NLCG-Net: A Model-Based Zero-Shot Learning Framework for Undersampled Quantitative MRI Reconstruction Jiang, Xinrui Jun, Yohan Cho, Jaejin Gao, Mengze Yong, Xingwang Bilgic, Berkin Image and Video Processing Machine Learning Signal Processing Typical quantitative MRI (qMRI) methods estimate parameter maps in a two-step pipeline that first reconstructs images from undersampled k-space data and then performs model fitting, which is prone to biases and error propagation. We propose NLCG-Net, a model-based nonlinear conjugate gradient (NLCG) framework for joint T2/T1 estimation that incorporates a U-Net regularizer trained in a scan-specific, zero-shot fashion. The method directly estimates qMRI maps from undersampled k-space using mono-exponential signal modeling with scan-specific neural network regularization, enabling high-fidelity T1 and T2 mapping. Experimental results on T2 and T1 mapping demonstrate that NLCG-Net improves estimation quality over subspace reconstruction at high acceleration factors. |
| title | NLCG-Net: A Model-Based Zero-Shot Learning Framework for Undersampled Quantitative MRI Reconstruction |
| topic | Image and Video Processing Machine Learning Signal Processing |
| url | https://arxiv.org/abs/2401.12004 |