請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/43009
標題: | Q球擴散影像中缺損影像的誤差評估與資料修正 Error Evaluation and Data Correction for the Corrupted Images in Q-ball Imaging |
作者: | Yen-Wei Cheng 鄭炎煒 |
指導教授: | 鍾孝文 |
共同指導教授: | 周銘鐘 |
關鍵字: | Q球造影,擴散權重影像,方位分佈函數,缺損影像,心電圖限制延遲發,對稱補償法,鄰點內差法,擴散頻譜影像工作室, QBI,DWI,ODF,corrupted images,ECG gating trigger delay,symmetrical compensation,neighboring interpolation,DSI Studio, |
出版年 : | 2011 |
學位: | 博士 |
摘要: | 目的
Q球造影是解開像素中神經纖維交會的一種擴散造影技術。然而使用面回訊的擴散權重影像,經常會發生影像的缺損,降低神經纖維方向與追蹤的準確性。因此本實驗的目的是提出一個適當的後處理法方法來回復缺損影像,不增加掃瞄時間並且能修正神經纖維方向與追蹤。 材料與方法 在3T的磁振造影掃描器內收集一位健康受試者的Q球造影資料。在驗證的程序中,我們先經由一百次重複掃瞄,觀察異常的訊號的特性,並使用心電圖限制延遲激發來消除異常的訊號。然後經由連續重複兩次的252方向Q球造影來得到參考資料,修補損壞的擴散權重影像訊號的方法是採用對稱補償法和鄰點內差法。我們把修正前的缺損資料和修正後的更正資料和參考資料相比較,觀察擴散權重影像和方位分佈函數的改變。 在模擬程序中,強度降低兩倍、數目增加兩倍和隨機產生的缺損訊號等用來測試鄰點補償法,並求最佳的閥值。在修正的程序中,先使用軟體和中位數濾波器在真實空間來對位和平滑原始影像。然後將受損影像和其相鄰的六個擴散方向的影像相比,接著將受損影像中低於閥值低訊號的畫素,用其Q空間上相鄰六個點訊號的距離倒數加權平均來修正。最後比較使用鄰點內差法前後,方位分佈函數和纖維追蹤結果與指標的變化。 結果 研究顯示異常訊號是強度減弱型,心電圖限制延遲激發不能消除所有的異常訊號,所以不能作為252方向Q球造影參考資料。用鄰點內差法來修復擴散權重影像並大幅降低方位分佈函數誤差,在三個程序中都是可行的,它也些微增加纖維追蹤的指標。 結論 在這個研究中,我們驗證缺損影像造成方位分佈函數和纖維追蹤的誤差,是可以用鄰點內差法來修復且不增加掃瞄時間。因此我們推定所提出的鄰點內差法是處理Q球造影資料的合適工具。 Objective Q-ball imaging (QBI) was a diffusion imaging technique capable of resolving intra-voxel fiber crossings. However, corrupted images were often found to occur in diffusion-weighted image (DWI) by Echo Planar Imaging (EPI), downgrading the accuracy of fiber orientations and tracking. Therefore, the purpose of this study was to propose a suitable post-processing method to restore the corrupted images, and correct the fiber orientations and tracking. Materials and Methods QBI data were collected from a healthy subject at a 3T MR scanner. In validation procedure, we observed the characteristics of outlier signals by 100 repeated scans and used ECG gating trigger to delay delete all corrupted signals. Then, we got the reference data by 2 repeated scans in 252 QBI. Afterwards, the algorithms using neighboring interpolation (NI) and symmetrical compensation (SC) methods were applied to remedy the corrupted signals in DWIs. Finally, DWIs and the orientation distribution function (ODF) were compared with those of the reference data before and after correction. In simulation procedure, neighboring interpolation was tested by simulated double drop, twice and randomly corrupted signals to find optimal thresholds. In correction procedure, a software and a median filter were used to register and smooth the images in real space. Then the corrupted images were compared by with their six neighboring direction images. Afterwards, the low signal pixels below the threshold in corrupted images were corrected by the distance weighted average of their six neighboring in Q-space. Finally, the orientation distribution function (ODF), fiber tracking results and indices were compared before and after neighboring interpolation correction. Results The study showed the outliers were decreasing signals and ECG gating trigger delay couldn’t delete all corrupted signals, so it couldn’t be a reference data for 252 QBI. The neighboring interpolation correction was feasible to restore corrupted DWIs and reduce the ODF errors greatly in three procedures. It also improved the fiber tracking indices slightly. Conclusions In this study, we demonstrated that ODFs and fiber tracking in QBI were altered by corrupted images, which were and recovered by the neighboring interpolation without increasing scan time. Therefore, we concluded that the proposed neighboring interpolation method was a suitable adjunct for QBI data processing. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/43009 |
全文授權: | 有償授權 |
顯示於系所單位: | 電機工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-100-1.pdf 目前未授權公開取用 | 2.49 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。