Multichannel Phase Reconstruction

Phase plays an important role in magnetic resonance imaging(MRI).MRI techniques based on phase images can provide some useful information.For example,susceptibility weighted imaging(SWI)and quantitative susceptibility mapping(QSM)can reveal the deposition of iron and calcification of tissues;Water-fat separation(WFS)can be used to detect some tissue steatosis;Temperature mapping technique can measure the local temperature in ultrasound ablation;etc.All these techniques need phase images with high signal-to-noise ratio(SNR)and free of artifact.To combine multichannel phase images,traditional method will perform reference scan using the body coil.And the sensitivity information generated from the reference scan will serve as a weighting function to combine multichannel complex images.This method is time-consuming and sometimes cannot be used since not all scanners have integrated body coils.There are some existing phase-sensitive channel combination methods that don't rely on extra reference scans,but these methods are problematic in some specific applications.For example,weighted mean method will get images with low SNR;Hermitian product method is only suitable for multiecho phase-contrast reconstruction,it is not applicable for single-echo acquisition,and is not compatible with WFS;Adaptive reconstruction(AR)will get magnitude images with high SNR,but will lead to an uncertainty when reconstructing phase images,and may suffer from local variations in phase which are introduced by water and fat components;etc.Therefore,there are still not standard channel combination methods for SWI,QSM and WFS.In this thesis,an improved AR channel combination method is proposed for QSM and SWI applications,and another new channel combination method is also proposed for WFS application.There are three steps in the improved AR method,including estimating and removing channel-dependent phase offsets,data quality judgement and channel rearrangement,and the final AR combination.Phantom study and in-vivo QSM and SWI experiments show that this method can get more robust results,compared with the traditional AR method.For the new channel combination method for WFS,there arealso two steps.In the first step,original multichannel data are combined with a weighted mean method and perform WFS to get rough water and fat images.In the second step,original data incorporate with the rough water and fat images,are used to estimate channel-dependent phase offsets.These offsets are removed from original data and weighted mean is used again to get combined complex images.A final WFS will produce water and fat images with high accuracy.Simulation and in-vivo studies have validated this accuracy.