Accelerated coronary MRI with sRAKI: A database-free self-consistent neural network k-space reconstruction for arbitrary undersampling
Accelerated coronary MRI with sRAKI: A database-free self-consistent neural network k-space reconstruction for arbitrary undersampling
复制标题
DOI：
10.1371/journal.pone.0229418
复制
发表时间：
2020-02-21
期刊：
影响因子：
3.7
通讯作者：
Akcakaya, Mehmet
中科院分区：
文献类型：
--
作者：
Hosseini, Seyed Amir Hossein;Zhang, Chi;Akcakaya, Mehmet
PurposeTo加速冠状动脉MRI采集与任意欠采样模式，通过使用一种新的重建算法，适用于线圈自我一致性使用subject-specific neural networks.MethodsSelf-consistent robust artificial-neural-networks for k-space interpolation（sRAKI）执行迭代并行成像重建线圈之间的自我一致性。该方法与SPIRiT相似，但将SPIRiT中的线性卷积扩展到使用卷积神经网络（CNN）的非线性插值。这些CNN使用扫描特定的自动校准信号（ACS）数据针对每次扫描单独训练。通过施加学习的自一致性和数据一致性来执行重建，这使得sRAKI能够支持随机欠采样模式。在6名健康受试者中采集全采样靶向右冠状动脉MRI。对数据进行回顾性欠采样，并使用SPIRiT、I（1）-SPIRiT和sRAKI对2至5的加速率进行重建。此外，前瞻性欠采样的全心脏冠状动脉MRI被收购，以进一步评估reconstruction performance. ResultsRAKI减少噪声放大和模糊伪影相比，SPIRiT和l（1）-SPIRiT，特别是在高加速率在有针对性的冠状动脉MRI。定量分析显示，sRAKI在标准化均方误差（与SPIRiT和I（1）-SPIRiT相比，在速率5下，与44%相似，与21%相似）和血管锐度（与SPIRiT和I（1）-SPIRiT相比，在速率5下，与10%相似，与20%相似）方面优于这些技术。全心脏数据显示，最尖锐的冠状动脉时，使用sRAKI解决，与11%和15%的改善，在SPIRiT和l（1）-SPIRiT，分别在血管锐度。ConclusionsRAKI是一个无数据库的神经网络为基础的重建技术，可以进一步加速冠状动脉MRI与任意欠采样模式，同时提高噪声弹性线性并行成像和图像清晰度超过l（1）正则化技术。
PurposeTo accelerate coronary MRI acquisitions with arbitrary undersampling patterns by using a novel reconstruction algorithm that applies coil self-consistency using subject-specific neural networks.MethodsSelf-consistent robust artificial-neural-networks for k-space interpolation (sRAKI) performs iterative parallel imaging reconstruction by enforcing self-consistency among coils. The approach bears similarity to SPIRiT, but extends the linear convolutions in SPIRiT to nonlinear interpolation using convolutional neural networks (CNNs). These CNNs are trained individually for each scan using the scan-specific autocalibrating signal (ACS) data. Reconstruction is performed by imposing the learned self-consistency and data-consistency, which enables sRAKI to support random undersampling patterns. Fully-sampled targeted right coronary artery MRI was acquired in six healthy subjects. The data were retrospectively undersampled, and reconstructed using SPIRiT, l(1)-SPIRiT and sRAKI for acceleration rates of 2 to 5. Additionally, prospectively undersampled whole-heart coronary MRI was acquired to further evaluate reconstruction performance.ResultssRAKI reduces noise amplification and blurring artifacts compared with SPIRiT and l(1)-SPIRiT, especially at high acceleration rates in targeted coronary MRI. Quantitative analysis shows that sRAKI outperforms these techniques in terms of normalized mean-squared-error (similar to 44% and similar to 21% over SPIRiT and l(1)-SPIRiT at rate 5) and vessel sharpness (similar to 10% and similar to 20% over SPIRiT and l(1)-SPIRiT at rate 5). Whole-heart data shows the sharpest coronary arteries when resolved using sRAKI, with 11% and 15% improvement in vessel sharpness over SPIRiT and l(1)-SPIRiT, respectively.ConclusionsRAKI is a database-free neural network-based reconstruction technique that may further accelerate coronary MRI with arbitrary undersampling patterns, while improving noise resilience over linear parallel imaging and image sharpness over l(1) regularization techniques.