Purpose: To develop and analyze the performance of the variation correction algorithm (VCA), a phase correction technique that mitigates the contribution of background phase variations by combining accurate alignment of echoes, K-space-based phase correction (as opposed to spatial polynomials), and extraction of alias-free phase difference images.
Materials and methods: A series of echo-shifted gradient-recalled echo (GRE) images was processed with K-space alignment and phase corrected with increasing sizes of M x M masks of central K-space coefficients. The extent of background phase variation suppression due to magnet field drift was assessed. Further, a simulated thermal profile was superimposed on the same data in a related experiment. Residual errors in reconstructed simulated thermal profiles were quantitatively characterized to estimate algorithm performance.
Results: Using a 3 x 3 K-space mask, the VCA was able to 1) maintain the typical mean background error in a 35 x 35 pixel region of interest (ROI) at -0.1 degrees C; and 2) reconstruct, relative to the applied thermal profile, a phase-corrected profile that typically contains a 1.7 degrees C underestimation of peak temperature difference and a mean error along the 60 degrees C line of -0.8 degrees C.
Conclusion: The results suggest that thermal profiles can be accurately reconstructed at 0.2 T using the VCA, even in the presence of over 1 ppm spatially and temporally dependent field drift over a 1-hour time frame.
Copyright 2003 Wiley-Liss, Inc.