Contrast-enhanced MR angiography with frequency-dependent mask subtraction

One of the main challenges for high-resolution contrast-enhanced magnetic resonance angiography (CE-MRA) is the limited signal-to-noise ratio (SNR). In conventional CE-MRA, spoiled gradient-recalled echo sequence is typically used to acquire raw data before and after contrast material injection, followed by mask subtraction to improve vessel contrast at the cost of increased image noise. We reported here on a frequency-dependent mask subtraction technique where only low spatial frequency data were subject to mask subtraction to improve image contrast and reduce motion artifact while high spatial frequency data were not subtracted to preserve image noise level. The feasibility of this technique was demonstrated through phantom, animal, and human volunteer studies. The lowest half, quarter, 1/8 and 1/16 of all spatial frequencies were subject to mask subtraction, respectively. These partial subtraction techniques were compared with conventional mask subtraction in terms of SNR and artifact power. An SNR gain up to 37% was achieved without significant artifact power increase for quarter-subtraction where only the central one quarter k-space data were subject to mask subtraction.