擴散資料組合對於擴散峰值造影參數之重複性與平均值的影響

Effects of diffusion-weighting scheme on the reproducibility and mean values of diffusion kurtosis imaging-derived parameters

夏邦凱(Pang-Kai Hsia)

指導教授 ： 周銘鐘

高雄醫學大學/健康科學院/醫學影像暨放射科學系碩士在職專班/碩士(2015年)

https://doi.org/10.6832/KMU.2015.00039

摘要

目的：
擴散峰值造影（DKI）有別以往所使用的擴散造影方式，其利用多個擴散權重係數（b值）及多個擴散方向（dir）以觀察水分子擴散時的非高斯分布情況。雖然這樣的造影方式目前已被廣泛使用在臨床疾病研究方面，但此造影技術可能會受到造影參數的調整而影響係數的準確性及重複性，有鑒於此，為減少影像誤差對於診斷上誤診的機會，本研究探討擴散資料組組合對於擴散峰值造影的重複性及平均值的影響，藉此尋找出最佳表現之造影參數。
材料與方法：
本研究使用1.5T GE Signa HDxt MRI儀器進行所有擴散影像資料的擷取，並由十位健康自願的受試者（平均年齡介於20至22歲間）參與擴散峰值造影掃描，使用擴散權重b值為0、1000、3000 s/mm2；擴散方向為30個方向，並重複進行三次資料擷取。分離各個擴散方向所獲取的個別擴散權重影像後，再組合出不同的擴散峰值造影資料以及被定義出的參考標準組影像資料，所有資料進行曲線擬合後，藉由圈選出特定腦部之感興趣區的擴散不等向性（FA）、平均擴散度（MD）以及平均擴散峰值（MK），觀察不同擴散影像資料對於上述係數的重複性及平均值的影響。
結果與討論：
本實驗研究發現，不同擴散資料組合對於不同腦部感興趣區的FA、MD以及MK皆有不同特性的影響。全腦掃描綜觀結果來看，再重複性方面，b值＝1000 s/mm2, NEX＝2, dir＝6與b值＝3000 s/mm2, NEX＝3, dir＝15的資料組合有較低的變異係數；而在平均值方面，b值＝0, 1000, 3000 s/mm2，平均次數（NEX）＝3及dir＝15的資料組合，與參考標準影像所獲得的係數相比，有較低的誤差比例。若排除本實驗不可抗拒之誤差因素，如受試者本質結構上的差異或感興趣區圈選差異等等，特定幾組實驗參數仍有最佳的準確度以及重複性。
結論：
擴散峰值造影所獲得的FA、MD及MK參數，其準確性與重複性在不同的腦區會受到的擴散資料組合的影響

關鍵字

擴散峰值造影 ； 準確度 ； 重複性

並列摘要

Purpose:
Diffusion Kurtosis Imaging (DKI) is a new imaging method and is different from previous techniques. It utilizes several diffusion weighting factors (b value) and multiple diffusion directions (dir) to observe the Non-Gaussian distribution of water diffusion .Although this imaging method has been widely used in clinical research, its accuracy and reproducibility may be affected by imaging parameters Therefore, to reduce the likelihood of misdiagnosis for image diagnosis, this study investigated the effects of diffusion weighting scheme on the accuracy and reproducibility of DKI-derived indices and searched for the suitable imaging parameters for DKI acquisition.
Materials and Methods:
In this study, all diffusion image data were acquired in a 1.5T GE Signa HDxt MRI scanner from ten healthy volunteers (the average age of 20~22) who participated in DKI scan. The DKI data were acquired with b-value = 0, 1000, 3000 s/mm2 in 30 non-collinear diffusion directions and were repeated three times for reproducibility test. All of DKI data were firstly separated to generate different DKI data with different b-value, number of exciations (NEX), and number of directions, and were combined together as a reference DKI dataset. After curve fitting on individual DKI subsets, the fractional anisotropy (FA), mean diffusivity (MD), and mean kurtosis (MK) were calculated from multiple region-of-interest (ROI) to assess the impacts of different diffusion-weighting scheme on the reproducibility and mean value of DKI-derived indices.
Results and discussion:
This study found that different diffusion data combination impacts FA, MD, and MK differently in different brain regions. Overall, in terms of reproducibility, the data set of b value=1000 s/mm2, NEX＝2, dir＝6 and the set of b value＝3000 s/mm2, NEX＝3, dir＝15 have lower coefficient of variation. However, in terms of mean values , the data sets of b value＝0, 1000, 3000 s/mm2,NEX＝3, and dir＝15 have lower fractional errors , as compared with those indices obtained from the reference DKI dataset. Although there are some inevitable factors , such as structural differences of the subject or difference of selecting ROI and etc, that may influence the results, this study found that some DKI subsets with specific imaging parameters have both high accuracy and reproducibility.
Conclusion:
The accuracy and reproducibility of DKI-derived FA, MD, and MK indices were significantly influenced by the diffusion weighting scheme in different brain regions.

並列關鍵字

DKI ； Accuracy ； Reproducibility

參考文獻

1. Pauling, L. and C.D. Coryell, The magnetic properties and structure of hemoglobin, oxyhemoglobin and carbonmonoxyhemoglobin. Proceedings of the National Academy of Sciences of the United States of America, 1936. 22(4): p. 210.

2. Damadian, R., Tumor detection by nuclear magnetic resonance. Science, 1971. 171(3976): p. 1151-1153.

3. Lauterbur, P.C., Image formation by induced local interactions: examples employing nuclear magnetic resonance. Nature, 1973. 242(5394): p. 190-191.

4. Le Bihan, D. and E. Breton, Imagerie de diffusion in-vivo par résonance magnétique nucléaire. Comptes-Rendus de l'Académie des Sciences, 1985. 93(5): p. 27-34.

5. Merboldt, K.-D., W. Hanicke, and J. Frahm, Self-diffusion NMR imaging using stimulated echoes. Journal of Magnetic Resonance (1969), 1985. 64(3): p. 479-486.