SubZero: Subspace Zero-Shot MRI Reconstruction
This work addresses the need for high-quality MRI reconstructions without large training datasets, but it is incremental as it builds on existing subspace-based zero-shot methods.
The paper tackled the problem of improving zero-shot self-supervised learning for accelerated MRI reconstruction by proposing a parallel network framework with an attention mechanism, achieving improved performance in T1 and T2 mapping acquisitions compared to current methods.
Recently introduced zero-shot self-supervised learning (ZS-SSL) has shown potential in accelerated MRI in a scan-specific scenario, which enabled high-quality reconstructions without access to a large training dataset. ZS-SSL has been further combined with the subspace model to accelerate 2D T2-shuffling acquisitions. In this work, we propose a parallel network framework and introduce an attention mechanism to improve subspace-based zero-shot self-supervised learning and enable higher acceleration factors. We name our method SubZero and demonstrate that it can achieve improved performance compared with current methods in T1 and T2 mapping acquisitions.