Promotion of Reconstruction Performance for Equally Spaced Under-sampled Signals Using Amplitude Modulation Pulses

Motivation: In compressed sensing, it is impossible to separate folded-over images using equally spaced under-sampled signal when the subject is a real-valued image. Goal(s): Our goal was to enable image reconstruction using equally spaced under-sampled signal, and to generalize it to phase-varied images. Approach: Amplitude modulation in the phase-encoding direction makes the folded image complex-valued and thus facilitates separation of overlapping images in the image space. Results: The image reconstruction was successful up to a speedup factor of 4 with U-Net. Phase images were successfully reconstructed by partial continuous signal acquisition and phase correction. Impact: Reconstruction performance is improved by applying specific amplitude-modulated pulses prior to signal acquisition in equally spaced under-sampling CS, which is advantageous for sharp image reconstruction.