Deep learning with domain adaptation for accelerated projection-reconstruction MR
Deep learning with domain adaptation for accelerated projection-reconstruction MR
复制标题
DOI：
10.1002/mrm.27106
复制
发表时间：
2018-09-01
影响因子：
3.3
通讯作者：
Ye, Jong Chul
中科院分区：
文献类型：
--
作者：
Han, Yoseob;Yoo, Jaejun;Ye, Jong Chul
目的径向k空间轨迹是与磁共振成像结合使用的一种成熟的采样轨迹。然而，径向k空间轨迹需要大量的径向线进行高分辨率重建。增加径向线的数量导致采集时间延长，使常规临床应用更加困难。另一方面，如果我们减少径向线的数量，条纹伪影图案是不可避免的。为了解决这一问题，我们提出了一种具有域自适应的新颖深度学习方法，用于从欠采样k空间数据中恢复高分辨率MR图像。方法提出的深度网络从伪图像中去除条纹伪图像。为了解决现有数据有限的情况，我们提出了一种领域自适应方案，该方案采用使用大量x射线计算机断层扫描（CT）或合成的径向磁共振数据集的预训练网络，然后仅使用少量径向磁共振数据集进行微调。结果该方法优于现有的压缩感知算法，如total variation和pr - focus方法。计算时间比总变分法和pr - focus法快几个数量级。此外，我们发现使用来自相似器官的CT或MR数据进行预训练比使用来自相同模态的数据进行不同器官的预训练更重要。结论：当只有有限数量的MR数据可用时，我们证明了域适应的可能性。该方法在图像质量和计算时间方面优于现有的压缩感知算法。
PurposeThe radial k-space trajectory is a well-established sampling trajectory used in conjunction with magnetic resonance imaging. However, the radial k-space trajectory requires a large number of radial lines for high-resolution reconstruction. Increasing the number of radial lines causes longer acquisition time, making it more difficult for routine clinical use. On the other hand, if we reduce the number of radial lines, streaking artifact patterns are unavoidable. To solve this problem, we propose a novel deep learning approach with domain adaptation to restore high-resolution MR images from under-sampled k-space data.MethodsThe proposed deep network removes the streaking artifacts from the artifact corrupted images. To address the situation given the limited available data, we propose a domain adaptation scheme that employs a pre-trained network using a large number of X-ray computed tomography (CT) or synthesized radial MR datasets, which is then fine-tuned with only a few radial MR datasets.ResultsThe proposed method outperforms existing compressed sensing algorithms, such as the total variation and PR-FOCUSS methods. In addition, the calculation time is several orders of magnitude faster than the total variation and PR-FOCUSS methods. Moreover, we found that pre-training using CT or MR data from similar organ data is more important than pre-training using data from the same modality for different organ.ConclusionWe demonstrate the possibility of a domain-adaptation when only a limited amount of MR data is available. The proposed method surpasses the existing compressed sensing algorithms in terms of the image quality and computation time.