Susceptibility phase imaging with improved image contrast using moving window phase gradient fitting and minimal filtering

Abstract

Purpose: To enhance image contrast in susceptibility phase imaging using a new method of background phase removal.

Materials and methods: A background phase removal method is proposed that uses the spatial gradient of the raw phase image to perform a moving window third-order local polynomial estimation and correction of the raw phase image followed by minimal high pass filtering. The method is demonstrated in simulation, 10 healthy volunteers, and 5 multiple sclerosis patients in comparison to a standard phase filtering approach.

Results: Compared to standard phase filtering, the new method increased phase contrast with local background tissue in subcortical gray matter, cortical gray matter, and multiple sclerosis lesions by 67% ± 33%, 13% ± 7%, and 48% ± 19%, respectively (95% confidence interval). In addition, the new method removed more phase wraps in areas of rapidly changing background phase.

Conclusion: Local phase gradient fitting combined with minimal high pass filtering provides better tissue depiction and more accurate phase quantification than standard filtering.