Edge detection using sub-sampled k-space data: application to upper airway MRI

Introduction: The detection of tissue borders is of great importance in several MRI applications. Edge detection is typically performed as a post-processing step, using magnitude images that are reconstructed from fully-sampled k-space data. In dynamic imaging (e.g. of human speech, ventricular function, and joint kinematics), tissue borders often comprise the primary information of interest. In such cases, full and uniform k-space sampling may not be the most time efficient approach. In real-time MRI of human speech, high temporal resolution and accurate edge information are both required for a complete understanding of the production dynamics [1]. We propose a rotationally symmetric partial sampling scheme and modified edge detection approach that produces comparable edge maps relative to full sampling.