Numerical and experimental study of a novel phase contrast magnetic resonance (PC-MR) imaging technique: Sparse Interleaved Referencing PC-MR imaging
Journal
Journal of Magnetic Resonance Imaging
Date Issued
March 2008
DOI
10.1002/jmri.21323
Abstract
Purpose: To use numerical simulation and experimental approaches to introduce a novel phase contrast magnetic resonance (PC-MR) data processing technique termed Sparse Interleaved Referencing PC-MR, with potential to improve accuracy, temporal resolution, and signal-to-noise ratio (SNR) of PC-MR data. Materials and Methods: Computational fluid dynamics data were generated for a two-chamber orifice flow model simulating valvular regurgitation. The numerical results were validated and used to simulate conventional and Sparse Interleaved Referencing PC-MR data acquisitions. Common data sets were processed using conventional and Sparse Interleaved Referencing approaches and quantitative errors in velocity-time waveforms were measured and compared. In vitro phantom jet flow data and in vivo ascending aorta data were acquired and used to simulate Sparse Interleaved Referencing PC-MR. Results: The Sparse Interleaved Referencing PC-MR data showed significantly better representation of the velocity-time waveform in three areas: (i) lower root-mean-square errors (9.0 ± 1.0% versus 24.0 ± 0.2%; P < 0.005), (ii) simulation of conventionally processed data showed a pattern of peak velocity overestimation, which was experimentally demonstrated in in vitro data, whereas overestimation of peak velocity was dramatically attenuated using Sparse Interleaved Referencing (2.8 ± 0.4% versus 16.9 ± 6.4%, P < 0.005), and (iii) compared with the conventional scan, an average of 119.4 ± 26.6% (P < 0.005) SNR was realized in in vitro and in vivo Sparse Interleaved Referencing PC-MR data. Conclusion: Simulation and in vitro/in vivo results show that Sparse Interleaved Referencing PC-MR processed data in pulsatile and jet flow showed higher accuracy, better peak velocity representation, and improved SNR compared with the data processed using the conventional PC-MR method.