Susceptivity weighted MRI for iron deposition in multiple sclerosis : validation of filtering methods for phase drift removal

BACKGROUND AND PURPOSE Brain iron deposition has been implicated in several neurodegenerative disorders such as multiple sclerosis (MS). Several magnetic resonance imaging (MRI) techniques are being exploited to measure brain iron levels Susceptibility weighted imaging (SWI) is based on phase processing revealing local iron-induced magnetic fields anomalies. The purpose of this study was to evaluate the background phase removal effects on iron quantification accuracy. MATERIAL AND METHODS Thirty one patients with relapsing remitting MS and six healthy controls (HC) were analyzed on 1.5T scanner by anatomical MRI and gradient echo (GRE) SWI. The GRE phase was preprocessed to eliminate phase wrapping and phase background variations. Two phase drift removal methods were compared: 1) the classical homodyne low-pass (LP) filtering; ii) a novel homomorphic filtering. LP filtering by the Hanning window in the k-space was tested various window bandwidths. RESULTS We found that phase angle value changes as a function of Hanning window size. We also found that phase angle values coming from the 128x128 Hanning window in our algorithm correlated with the clinical standard obtained using 64x64 SIEMENS window size. Significant improvement were displayed by the logarithmic homomorphic filtering in place of the linear homodyne filtering. DISCUSSION For the background field removal techniques investigated, it can be easily assessed that homodyne method failed in the midbrain regions near air-tissue interfaces as sinuses. The rapid phase variations were insufficiently cancelled by Hanning window at 32x32, 64x64 and 96x96. Improved stability of results of provided by homomorphic filtering. CONCLUSION This study demonstrates that background phase removal can drastically affect iron quantification. For that reason, further assessment is necessary to confirm the accuracy of filtering techniques.