Magnetic resonance imaging (MRI) is a widely used medical imaging modality. However, due to the limitations in hardware, it is often clinically challenging to obtain high-quality MR images. Super-resolution (SR) is potentially promising to improve MR image quality without any hardware upgrade. Instead of the classical SR reconstruction method enhance the spatial resolution via utilizing the spatial information itself, in this work, we propose a novel SR method via learning channel information in virtual parallel imaging. Using auxiliary variable technology to make the channel number of network output to be equal to the network input, thereby increasing the number of channels information to achieve SR reconstruction. Compared with state-of-the-art SR methods, the present approach is advantageous in suppressing artifacts and keeping more image details.

virtual parallel imaging,Super-Resolution imaging,reversible network