Time-Varying Under-sampling Trajectory Design for Compressed Sensing Based DCE-MRI

Abstract

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) needs multiple images for achieving high temporal resolution in a very short time period. During the circulation of contrast agent (CA), scans characterizing signal intensity change of Region of Interest (ROI) highly determine the accuracy of parameter estimation. Traditional design is always bothered with parameter and reduction factor adjustment. Therefore, specific under-sampling trajectory design strategy should be applied for the scans with apparent intensity change in ROI. In this paper, we study how to design a time varying under-sampling trajectory considering specific characteristics in different stages of DCE-MRI. Scans are segmented into two stages with apparent and faint intensity change in ROI. For scans with apparent intensity change in ROI, low and high frequency domain is segmented considering different characteristics of images and phase encodings in localized low frequency domain are full-sampled. For the remainder phase encodings localized in high-frequency domain, the k-space error replace PDF to generate random sampling trajectory with fixed reduction factor and without any parameter adjustment. VD design with given reduction factor is still applied for the remainder scans with faint intensity change in ROI. Simulation results demonstrate high accuracy and robustness.