Purpose: To develop a fast high-resolution magnetization-prepared (MP) gradient echo (GRE) MRI sequence using complementary acquisitions of both positive and negative phase-cycling (PC) in one single 3D sequence before compressed sensing reconstruction. This inherent PC strategy eliminates imaging ghosting/blurring in the phase encoding directions due to T1 relaxation during the long GRE readout train, without doubling the scan time.
Methods: Experiments were performed on a 3.0T Philips MRI scanner using a large ACR phantom in a 12ch head coil. Complementary acquisitions (“inherent phase-cycling”) were implemented in a 3D MP-GRE sequence which first acquired a random half of the phase encoding profiles with +PC, and then the other half with -PC. The RF receiver phases were also inverted for all -PC acquisitions so that MP signal had the same phase as in the +PC acquisitions. RF flip angle sweeping from 4° to 30° was applied to maintain constant signal from the MP contrast along the GRE readout train. For comparison, a traditional MP-GRE sequence was scanned with the same imaging parameters except that a constant flip angle of 6° was employed. Both sequences had the same scan duration. Both raw k-space datasets were exported and reconstructed with the same total variation regularization lambda of 0.02 using the BART toolbox.
Results: Both imaging sequences resulted in high-resolution along the frequency-encoding directions. For the phase encoding directions, images from the complementary MP-GRE suffered much less from global ghosting and blurring, as well as local ringing artifacts compared to the traditional MP-GRE sequence.
Conclusion: Complementary phase-cycling acquisitions in combination with a long-echo-train MP-GRE has high imaging efficiency without suffering from ghosting/blurring artifacts along the phase-encoding directions. We demonstrated that fast high spatial-resolution acquisition in 3D MP-GRE can be achieved without addition of scan time using this complementary acquisition in combination with compressed sensing reconstruction.
Not Applicable / None Entered.