Room: AAPM ePoster Library
Purpose: In conventional phase-contrast MRI (PC-MRI), phase unwrapping of pixels with phase shifts exceeding p is essential for accurate estimation of velocity. Here we propose, implement and test the performance of a dual VENC phase unwrapping (UDV) scheme, and compare its performance against a multi-VENC (MV) method that is known to require a longer acquisition time.
MRI acquisition: Fluid flow through a 1.59 cm inner diameter tube with mean and maximum fluid velocity of 45.7 cm/s and 66.7 cm/s respectively was measured 10 times using PC-MRI with (VENC=100 and 20 cm/s) at 3.0 T, and the images were unwrapped using the UDV scheme. In order to compare the UDV and MV schemes, PC-MRI measurements at VENC=100, 50, 20 and 10 cm/s were also acquired and combined using the MV scheme yielding an overlapped velocity map. The scan time of the UDV and MV acquisitions were 198 s and 396 s respectively. The UDV scheme was also tested on in vivo cardiac short-axis PC-MRI (VENC=150 and 20 cm/s) acquisitions containing through-plane left ventricular myocardial and blood velocities.
Data Analysis: VNR was calculated as the pixel wise ratio of temporal mean to standard deviation. Bland-Altman analyses was used to compare the VNRs of the DV and MV schemes.
Results: (1) The UDV scheme produces fully unaliased velocity maps, and reduces the number of images (or acquisition time) needed for VNR optimization in both phantom and in vivo PC-MRI acquisitions. (2) The VNR of the UDV and MV velocity maps for the flow phantom experiment were in good agreement (p<0.0005 and r²=0.95), and showed a negligible bias (p=0.05).
Conclusion: The UDV scheme is effective in unwrapping aliased pixels from a PC-MRI velocity map, and hence optimizes the VNR of the acquisition without prohibitively elongating the acquisition time of the PC-MRI experiment.