Super resolution of dynamic MRI using deep learning, enhanced by prior-knowledge

摘要

Introduction: Dynamic MRI suffers from the spatial-temporal resolution trade-off of MRI, making the acquisition of 3D Dynamic MRI a challenging task. Deep Learning has been proven [1-3] to be a successful tool for performing super-resolution on MRIs. The fast speed of inference of deep learning based models makes them perfect for real-time dynamic MRI for interventional purposes. But typically, deep learning based methods need large training sets similar to the actual real-time acquisition; using training set significantly different than the test set can produce results of poor quality. This research tries to address this problem by training on a publicly available dataset, acquired using completely different sequences than the dynamic MRI used here for testing; and then fine-tuning using a prior static scan of the same subject. Before any intervention, such scans are taken for planning. During intervention that model can be used to perform super-resolution reconstruction of the real-time dynamic MRI. Subjects/methods: In this work, a 3D version of the U-Net [4] has been used for super-resolution. 3D abdominal MR volumes were artificially downsampled [5] to simulate low resolution datasets, by taking the center of the k-space of each slice. For the main training of the model, T1-Dual (in- and opposed-phase) images from the training set of CHAOS [6] were used. The training was performed using 3D patches of the volumes, with a patch size of 243 and a stride of 6 for the slice dim and 12 for the other dims. The loss was calculated using SSIM and was minimized using Adam with a learning rate of 1e-4. Furthermore, 2D Dynamic MRIs of the abdomen were acquired at 3T (Siemens Skyra) [T1 GRE, TR: 4.1, TE: 1.60] and then stacked together to create 3D Dynamic MRI, by syncing the breathing phases using an optical tracker. This dataset was also artificially downsampled as the training set. Finally, this was split into different timepoints and the first time point was considered