- 學位論文
擴散張量造影之參數的準確性及重複性：單極梯度與雙極梯度之比較
Reproducibility and Accuracy of Diffusion Tensor Imaging-derived Parameters: A Comparison between Unipolar and Bipolar Gradient
摘要
目的： 擴散張量造影中的擴散不等向性(fractional anisotropy, FA)、平均擴散度(mean diffusivity, MD)和主特徵向量(principal eigenvector, PEV)可描述水分子擴散程度、大腦纖維結構和纖維走向，然而，強大擴散梯度磁場在開關過程中會產生渦電流扭曲。目前單極梯度配合仿射影像對位和雙極梯度技術是減低渦電流扭曲良好方式，但是修正後的擴散影像分析結果並未與雙極梯度有量化的比較。因此，本研究目的為藉由分析FA、MD和PEV的重複性與準確性，來比較單極梯度配合仿射影像對位與雙極梯度技術的表現。 材料與方法： 本研究使用3T Siemens Skyra MRI掃描儀進行影像收集，十位健康受試者(男:女=5:5，平均年齡22.1±2.7歲)在TE為71ms(單極梯度可用之低限值)、84ms(雙極梯度可用之低限值)和100ms時，分別使用單極梯度與雙極梯度擴散脈衝序列掃描腦部影像，並分別重複收集五次以進行統計分析。透過仿射影像對位修正移動影響及渦電流扭曲，接著計算修正後影像的FA、MD和PEV，最後以變異係數、角度變異、角度差異、"∆(" ‾("FA" ) ")和∆(" ‾("MD" ) ")" 比較兩項技術結果的重複性與準確性並進行統計分析。 結果與討論： 當兩項技術選擇最短TE時，單極梯度的變異係數不論在白質或灰質皆低於雙極梯度，因為其可選擇較雙極梯度短的TE而有較高的SNR，另外，相同TE時單極梯度的變異係數也普遍低於雙極梯度。在準確性的實驗中，FA與MD數值可能因為擴散時間不同導致兩項技術有顯著差異，而數值差異分別介於1.54%~4.46%和1.19%~2.61%，但是角度變異則顯示PEV的方向性差別不大。 結論： 根據重複性結果發現單極梯度的變異係數普遍較雙極梯度低，尤其當兩者使用最短TE掃描時更為明顯，準確性的實驗則發現兩項技術的FA與MD數值有些微不同，但是PEV的方向性差異不大。不過就FA、MD與PEV的重複性方面而言，我們發現單極梯度擴散脈衝序列配合仿射影像對位是優於雙極梯度技術，因此，此方式是一項可提升擴散張量造影影像品質的方法，並且可以應用於臨床診斷來增加病變的確診率。
並列摘要
Purpose: The quantitative diffusion tensor indices, such as mean diffusivity (MD), fractional anisotropy (FA) and principal eigenvector (PEV), can describe the diffusion of water and fiber integrity of brain tissue in vivo. However, the diffusion gradients are known to cause eddy-current distortions in diffusion-weighted images. For this problem, unipolar gradient with affine image registration and bipolar gradient technique are both feasible methods for correcting eddy-current distortion, but their performances have not been compared quantitatively. Therefore, the purpose of this study is to compare the reproducibility and accuracy of DTI-derived indices between unipolar and bipolar gradients at different TEs. Materials and Methods: All imaging data were acquired at a 3T MRI scanner (Siemens Skyra). Ten healthy volunteers (male/female=5/5, age: 22.1±2.7 years) were scanned with unipolar and bipolar DTI at TE=71ms (the minimum for unipolar), 84ms (the minimum for bipolar) and 100ms, respectively, and each acquisition was repeated five times for statistical analysis. The eddy-current distortion and motion effects were corrected by affine image registration, followed by FA, MD and PEV calculation. Afterwards, the reproducibility and accuracy of DTI-derived indices were evaluated by calculating their coefficient of variation (CV), angular variation (AV), angular difference (AD), "∆(" ‾("FA" ) ") and ∆(" ‾("MD" ) ")" . Afterwards, the results were compared statistically by using paired Student’s t-test. Results and discussion: This study found that the bipolar DTI had larger CV(MD) and CV(FA) than unipolar DTI at minimum TEs (= 71ms vs 84ms). The finding indicates that the reproducibility of unipolar DTI is better than bipolar DTI due to the fact that unipolar DTI can achieve shorter TE than bipolar DTI. Of particular interest, unipolar DTI generally exhibited lower CV(FA) and CV(MD) than those of bipolar DTI even with same TE. However, CV(MD) did not have statistically significant differences at TE=84 and 100ms between unipolar and bipolar DTI. In accuracy, the FA and MD values were significantly different between unipolar and bipolar DTI due likely to their different diffusion time and signal-to-noise ratio, but their differences were only 1.54%~4.46% and 1.19%~2.61%, respectively. The unipolar DTI and bipolar DTI had similar PEV in white matter. Conclusion: According to the results, the reproducibility analysis found that the unipolar DTI generally had lower CV(FA), CV(MD) and AV(PEV) than bipolar DTI, especially at their minimal TE. In the results of accuracy, the unipolar DTI and bipolar DTI had significantly different FA and MD values but have similar PEV. Therefore, we concluded that the unipolar DTI with affine image registration correction is superior to bipolar DTI technique in terms of reproducibility of DTI-derived indices and had a potential of improving clinical diagnosis.