Fast data acquisition in Magnetic Resonance Imaging (MRI) is vastly in demand and scan time directly dependson the number of acquired k-space samples. Recently, the deep learning-based MRI reconstruction techniqueswere suggested to accelerate MR image acquisition. The most common issues in any deep learning-based MRIreconstruction approaches are generalizability and transferability. For different MRI scanner configurations us-ing these approaches, the network must be trained from scratch every time with new training dataset, acquiredunder new configurations, to be able to provide good reconstruction performance. Here, we propose a newgeneralized parallel imaging method based on deep neural networks called NLDpMRI to reduce any structuredaliasing ambiguities related to the different k-space undersampling patterns for accelerated data acquisition.Two loss functions including non-regularized and regularized are proposed for parallel MRI reconstruction us-ing deep network optimization and we reconstruct MR images by optimizing the proposed loss functions overthe network parameters. Unlike any deep learning-based MRI reconstruction approaches, our method doesn'tinclude any training step that the network learns from a large number of training samples and it only needs thesingle undersampled multi-coil k-space data for reconstruction. Also, the proposed method can handle k-spacedata with different undersampling patterns, and the different number of coils. Experimental results show thatthe proposed method outperforms the current state-of-the-art GRAPPA method and the deep learning-basedvariational network method.
展开▼
