Direct Cardiac T1 Mapping with Subspace Modeling and Free-breathing Data Acquisition

Objective: Cardiac T₁ mapping is an emerging magnetic resonance imaging (MRI) tool in the quantitative assessment of cardiomyopathies. Conventional techniques necessitate breath-holding during each scan, which can be challenging for certain patients. This work presents a novel T₁ estimation framework based on subspace modeling and direct T₁ estimation for cardiac T₁ mapping with free-breathing data acquisition.

Sparse (k,t)-space data is collected along a random radial trajectory using a free-breathing, ECG-gated, inversion-recovery fast low-angle shot (FLASH) sequence. The T₁ relaxation model is incorporated into a direct reconstruction framework to perform end-to-end T₁ mapping directly from the undersampled (k,t)-space data. The resultant image reconstruction problem is solved using the alternating direction method of multipliers (ADMM) algorithm, which decomposes the optimization problem into subproblems, such as image reconstruction with a low-rank constraint, parametric fitting with a sparsity constraint, and total variation (TV) based image denoising problems.

The performance of the proposed method is evaluated by comparisons with the two-step, indirect approaches through numerical simulations and in-vivo experiments. Results demonstrate the benefits of the proposed direct approach in terms of T₁ estimation bias and variance Conclusions: Direct cardiac T₁ mapping with subspace modeling enables free-breathing cardiac T₁ mapping with improved accuracy.

This study demonstrates the potential utility of incorporating MR physics knowledge and subspace modeling into a direct estimation framework for improved cardiac T₁ mapping with free-breathing acquisition.