指紋磁振造影中評估訊號收取次數與參數圖譜精密度的方法

Abstract

指紋磁振造影是一項同時定量多種磁振造影參數的新方法。其利用一系列隨機的射頻脈衝序列，使不同組織產生不同的信號，再利用模式辨認方法，將量測到的信號與事先模擬的信號字典進行比對，進而得到組織的參數。 在指紋磁振造影中，信號量測的次數影響最終的信號長度。信號長度越長，信號字典越大，實驗所需的時間與電腦計算的複雜度都會隨之增加。然而，當信號量測次數減少，參數圖譜的精確度也會隨之改變。為了設計出理想的脈衝序列，我們需要一種評估精確度的方法。 在本篇論文中，我們提出一項參數映射變異指標，可以用以反應參數圖譜的精確度。此外，這個指標可以在掃描前預測參數圖譜精確度隨不同信號收取次數的變化，進而提供實驗前脈衝序列設計參考。

MR Fingerprinting (MRF) is a novel technique to quantify multiple MR parameters simultaneously. A train of pseudorandomized radiofrequency (RF) excitations are used to generate unique signal evolution for different tissues, followed by matching the measured signals to a pre-established dictionary. The signal acquisition number in MRF is related to the signal length. Longer signals and larger dictionaries increase the scan time and computational complexity. However, as the signal acquisition reduces, the mapping precision also changes. Therefore, for designing an efficient MRF sequence, a method to evaluate the precision change is necessary. In this thesis, we propose a mapping variation index to reflect the mapping precision in MRF. Besides, this index can predict the precision change under different signal acquisition numbers before MRF scans and provide a reference for sequence designers to modify the signal acquisition number.