利用相位重建改良技術擷取磁場分佈圖: 空間頻率能量能譜的進階應用

Abstract

磁共振造影中，磁場梯度會改變影像上不同空間位置的共振頻率進而累積出相位角。藉由不同時間得到的相位影像得知其累積的相位角，便可計算得到磁場分佈圖，此即為磁場分佈擷取技術的概念。而其中，相位累積時間的長短會影響相位重建技術的結果，同時也就決定了磁場分佈擷取技術成功與否。當採用小的時間差，所累積出來的相位值易受到雜訊的影響而造成磁場分佈估算錯誤；然而，當利用長時間去累積相位時，磁場所造成的訊號衰減和相位折疊效應都會降低磁場分佈圖的準確度。另一方面，空間頻率能量能譜演算法是另一種可以用來擷取磁場分佈圖的技術，然而其所採用的積分技巧會是此技術的誤差來源。本篇論文將空間頻率能量能譜演算法所得到的磁場分佈圖視作一個估算值，進一步套用至時間域上和空間域上的相位重建技術，來取得更準確的磁場分佈圖。

B0 inhomogeneity mapping, recording spatially-dependent frequency offset, provides off-resonance information for MR imaging applications that need high accuracy. Mapping of B0 is often achieved through an extraction of phase variation versus time interval. Since phase values are restricted in the interval of , the accuracy of B0 mapping shows dependence on the success of phase unwrapping, which is in turn dominated by the decision of the phase accumulation time (ΔTE). Smaller accumulation time generates results that are prone to noise influences; while the wrap-around effects and the susceptibility-induced signal loss lead to degradation of B0 map with large ΔTE. On the other hand, KESA measuring the susceptibility field gradients has been shown as an effective alternative for EPI-based B0 mapping studies. However, the integration nature limits the accuracy of B0 mapping. In this thesis, two improved phase unwrapping procedures for mapping B0 field inhomogeneity are proposed with prior knowledge obtained from KESA processing. For temporal dimension phase unwrapping, the KESA method is used for an initial estimation of the phase evolution as a function of TE. The accurate phase values are then derived on a pixel-by-pixel basis from the multi-TE data, using sparse TE spacing for scan efficiency and large TE coverage. For dual-TE data, the spatial domain phase unwrapping for phase accumulation map is accomplished by exploiting field gradients information from KESA, accompanying with the appropriate use of the mask in the initial KESA estimation to further address the signal dropout effect.