Deep Learning Method for Nonstationary Image Artifact Correction

Nonstationary image artifacts frequently arise in MRI from off-resonance and motion. Current methods to correct these nonstationary effects are computationally expensive. Stanford researchers have developed a new deep learning framework to improve image quality in minutes. This method allows radiologists to evaluate image quality while the patient is still in the scanner. This technique also allows for shorter scans by allowing for longer readouts and correcting the artifacts that may arise.