高夾角解析度擴散磁振造影之技術研究與發展：擴散加權值評估與取樣效率改進

Abstract

在過去十年中，擴散磁振造影技術藉由其計算在大腦不同功能區之間的神經纖維連結的能力，已在神經科學以及臨床研究上成為一個相當重要的工具。而相較於傳統的擴散磁振造影技術，高角解析度擴散影像技術更提供了複雜的神經纖維結構，如白質神經纖維的交叉等。然而，到目前為止，由於高角解析度擴散影像的冗長取像時間，加上大多數臨床研究者對於此技術相關知識的缺乏，造成其並未在臨床研究上被廣汎的使用。因此，本論文的主要目的是讓臨床研究者易於了解高角解析度擴散造影技術並進而使用之。 首先，我們藉由假體實驗評估了高角解析度擴散造影在計算複雜神經纖維方向的準確度以及其角度解析度。藉由闡述高角解析度擴散造影與序列參數之間的相關性，我們能幫助心理學家或臨床學家知道如何根據他們的需要來設定參數以得到好的結果。我們的結果顯示伴隨著較高的信雜比或是較低的擴散權值，高角解析度擴散造影解出的神經纖維方向有較高的準確度。而其角度解析度也隨著q值的增加而變好。而q值的選取決定於高角解析度擴散造影的準確性以及角度解析度。 另外，我們提出半球取樣的方法來縮短高角解析度擴散造影的取像時間。藉由擴散權值交叉項的修正，我們可以同時縮短時間並且得到準確得神經纖維方向估計。我們的結果顯示在使用半球取樣且沒修正擴散權值交叉項的情況下，神經纖維方向的估算有了明顯的誤差，而這誤差在經由修正擴散權值交叉項之後也的確變小。本論文提供了高角解析度擴散造影參數最佳化的決策並縮短了一半的取像時間，對於高角解析度擴散造影技術在未來臨床或是心理學上的研究有相當大的助益。

In past decade, the diffusion MRI has become an important approach to predict complex neural fiber connectivity between different brain regions in neuroscience and clinical fields. Comparing with conventional diffusion MRI techniques, high angular resolution diffusion imaging (HARDI) techniques provide the capability of resolving complex neural fiber architectures, such as intravoxel fiber crossing of white matter tracts. However, so far the HARDI techniques have not been extensively used in clinical application due to excessive scan time and the lack of the knowledge on such techniques for clinicians. Therefore, the main object of this dissertation is to facilitate the application of HARDI for clinicians. First, we evaluated the q-related accuracy and angular resolution of estimating the complex fiber orientations for HARDI via phantom experiments. By illustrating the relationship between the inherent property of HARDI and the sequence parameters, we could help the psychologists or clinicians to understand how to get a better HARDI result or to set the parameters according to their demands. In our results, a more accurate estimation of fiber orientations was obtained when a higher SNR or lower q value was achieved. The angular resolution was also highly correspondent with the q value, i.e. a higher q value yielded a better angular resolution. It is a trade-off between the accuracy and the angular resolution to optimize the q value. Second, a hemi-spherical encoding scheme was proposed based on the spherically symmetric property of diffusion signals. With correcting for the cross-term, we can get accurate fiber orientation estimate and halve the scan time of HARDI simultaneously. Our results showed an obvious estimation error when using a hemi-spherical encoding scheme without correcting for the cross-term in comparison with the cross-term-free results. This estimation error could be reduced after correcting for the cross-term. Conclusively, the application in clinical and psychological filed would benefit from our studies, which provide the optimization strategy of HARDI and halve the total scan time.