Department of Electronic and Electrical Engineering, Imperial College, London, UK.
Faculty of Mathematics and Computer Science, Weizmann Institute of Science, Rehovot, Israel.
Med Phys. 2019 Nov;46(11):4951-4969. doi: 10.1002/mp.13727. Epub 2019 Sep 10.
Magnetic resonance fingerprinting (MRF) methods typically rely on dictionary matching to map the temporal MRF signals to quantitative tissue parameters. Such approaches suffer from inherent discretization errors, as well as high computational complexity as the dictionary size grows. To alleviate these issues, we propose a HYbrid Deep magnetic ResonAnce fingerprinting (HYDRA) approach, referred to as HYDRA.
HYDRA involves two stages: a model-based signature restoration phase and a learning-based parameter restoration phase. Signal restoration is implemented using low-rank based de-aliasing techniques while parameter restoration is performed using a deep nonlocal residual convolutional neural network. The designed network is trained on synthesized MRF data simulated with the Bloch equations and fast imaging with steady-state precession (FISP) sequences. In test mode, it takes a temporal MRF signal as input and produces the corresponding tissue parameters.
We validated our approach on both synthetic data and anatomical data generated from a healthy subject. The results demonstrate that, in contrast to conventional dictionary matching-based MRF techniques, our approach significantly improves inference speed by eliminating the time-consuming dictionary matching operation, and alleviates discretization errors by outputting continuous-valued parameters. We further avoid the need to store a large dictionary, thus reducing memory requirements.
Our approach demonstrates advantages in terms of inference speed, accuracy, and storage requirements over competing MRF methods.
磁共振指纹识别（MRF）方法通常依赖于字典匹配，将时间 MRF 信号映射到定量组织参数上。这种方法存在固有的离散化误差，并且随着字典大小的增加，计算复杂度也很高。为了解决这些问题，我们提出了一种混合深度磁共振指纹识别（HYDRA）方法，称为 HYDRA。
HYDRA 包括两个阶段：基于模型的特征恢复阶段和基于学习的参数恢复阶段。信号恢复是使用基于低秩的去混淆技术实现的，而参数恢复是使用深度非局部残差卷积神经网络实现的。所设计的网络是在使用 Bloch 方程和稳态进动预饱和（FISP）序列模拟的合成 MRF 数据上进行训练的。在测试模式下，它将时间 MRF 信号作为输入，并生成相应的组织参数。
我们在合成数据和来自健康受试者的解剖数据上验证了我们的方法。结果表明，与传统的基于字典匹配的 MRF 技术相比，我们的方法通过消除耗时的字典匹配操作显著提高了推断速度，并通过输出连续值参数减轻了离散化误差。我们还避免了存储大型字典的需要，从而降低了内存需求。
与竞争的 MRF 方法相比，我们的方法在推断速度、准确性和存储要求方面具有优势。
