Quantitative susceptibility mapping (QSM) has shown significant clinical potential for studying neurological disorders, but its acquisitions are relatively slow, e.g. 5-10 mins. Compressed sensing (CS) undersampling and reconstruction techniques have been used to accelerate the magnitude-based MRI acquisitions; however, most of them are ineffective to phase signal due to its non-convex nature. In this study, we propose a deep neural network “CANet” using complex attention modules to recover both the magnitude and phase images from the CS-undersampled data, enabling substantial acceleration of phase-based QSM.
This abstract and the presentation materials are available to members only; a login is required.