自我運動追蹤術：應用於磁振造影血流速測定及高解析度心臟造影術

(1)

行政院國家科學委員會專題研究計畫成果報告

自我運動追蹤術：應用於磁振造影血流速測定及高解析度心臟造影術

研究成果報告(精簡版)

計畫類別：個別型

計畫編號： NSC 95-2314-B-011-001-

執行期間： 95 年 08 月 01 日至 96 年 09 月 30 日執行單位：國立臺灣科技大學電機工程系

計畫主持人：黃騰毅共同主持人：劉益瑞

計畫參與人員：碩士班研究生-兼任助理：王基崇、唐育尉、王禹舜、陳俊文

報告附件：國外研究心得報告

處理方式：本計畫涉及專利或其他智慧財產權，1 年後可公開查詢

中華民國 96 年 12 月 25 日

(2)

行政院國家科學委員會專題研究計畫成果報告

自我運動追蹤術：應用於磁振造影血流速測定及高解析度心臟造影術計畫編號： NSC-95-2314-B-011-001

執行期限：95 年 8 月 1 日至 96 年 9 月 30 日主持人：黃騰毅台灣科技大學電機工程系一、中文摘要

我們的研究目的在利用全新的投影式編碼法以及螺旋槳式編碼法，應用在運動的追蹤上。

利用磁振造影術所收取到的 k 空間資料，運算得知因心跳所造成的血流速波動或是因呼吸所引起的器官位移。使用這項資訊，將相同運動狀況的資料重組成一張影像，便有機會得到一張無運動假影的高解析度影像。運用這項技術，將可省去臨床上準備心電圖或呼吸帶的麻煩，更可利用自由呼吸的造影術，使用更多的取像時間來獲取高解析度高信雜比的影像。應用在動物模型的磁振影像相關研究也能夠降低假影，提高實驗的可靠度。相關研究已在一年半的進行中，有了顯著的成果，包含投影式編碼法的完成以及螺旋槳式編碼法初步結果。相關研究已刊登在國際磁振造影學會年會，更預備撰寫投稿至國際期刊。

關鍵詞：運動追蹤，心肌 Introduction

For clinical cardiac MR imaging, patients are generally required to hold their breath during the scan to minimize respiratory motion-related artifacts. However, some patients cannot hold their breath due to illness or limited breath capacity. To solve this problem, several methods have been developed including gating from the signal of respiratory belt or navigator echoes. Recently, Larson et al. reported a novel self-gating technique that continuously extracts the respiratory trace from low-resolution images acquired by radial scanning with each heart beat [1]. Similarly to radial scanning, the PROPELLER encoding method [2] can reconstruct a low-resolution image from every blade due to k-space center oversampling. In this study, we investigated the feasibility of extracting respiratory-trace from the blades acquired by PROPELLER encoding for the application of free-breathing high resolution cardiac imaging.

Material and Methods

A PROPELLER-encoded ECG-triggered Turbo-Field Echo (TFE) sequence with T2 magnetization preparation (T2-prep) [3] was implemented on a 3T whole-body MR system (Achieva 3T, Philips) for T2-weighted myocardial imaging. Three subjects participated in this

(3)

study. For each subject, a set of one hundred PROPELLER blades were collected using the following parameters: FOV: 340mm, TR/TE:4ms/2.3ms, flip angle: 10 degrees, TET2-prep:50ms, end-systolic ECG gating, PROPELLER blade size: 30x256, PROPELLER rotation step: 7 degrees. The phase-encoding order for each blade was set to centric-reordering to maximize T2 contrast from the magnetization preparation.

The collected blades were processed with Matlab® (Mathworks, Natick, MA, USA) using a self-gating PROPELLER reconstruction algorithm. In order to extract the respiratory trace, the low resolution blades were first transformed to image domain and then smoothed using a median filter. A ROI was selected over the myocardium and all blade images were cropped using the same ROI. The correlation analysis was applied between the cropped blades and one selected blade. The series of correlation coefficients (CC) for the full set of 100 blades was used to identify the respiratory phases. Finally, the low resolution blade images with CC values higher than a preset threshold were combined and the PROPELLER reconstruction algorithm was applied to reconstruct a high resolution image.

Results

Fig.1 shows the respiratory trace obtained from cross correlation of the PROPELLER blades (see Fig.2) collected in one of the subjects. Notice that the respiratory duration detected from this trace was approximately 5 to 7 heart beats, which roughly matched the respiratory period observed on the subject outside the scanner. Fig.3 shows the reconstructed T2-weighted myocardial images with four different CC thresholds (a:0.0 , b:0.5, c:0.9, d:0.95) and the number of blades with CC values larger than each of the four threshold values (a:100, b:73, c:47,d:27).

The images show blurring instead of ghosting due to the PROPELLER k-space trajectory. Fig.3.a, corresponding to a non respiratory-gated image, shows most blurring. The other images (Fig.3.b-d) appear progressively sharper with increasing CC threshold values.

Discussion and Conclusions

In this study, PROPELLER encoding was implemented with a T2-prep TFE sequence to acquire high resolution cardiac images without breath-holding. Using the k-space center over-sampling property of PROPELLER encoding, the heart position can be identified in each low-resolution blade acquired in a heart beat. Thus, a high resolution image can be reconstructed from the blades selected with almost the same heart position. The scan efficiency depends on the CC threshold.

From our results, the reconstructed image showed less blurring with half of the blades selected for PROPELLER reconstruction. An even sharper reconstructed image can be obtained with less scan efficiency (ex. Fig.3(d) :27%). Further enhancement of the scan efficiency may be achieved by image domain registration if the myocardium movement can be assumed “in-plane” [2]. In our

(4)

study, free-breathing high resolution T2-weighted myocardium imaging was achieved with the proposed self-gating method. The method can be applied to myocardial BOLD investigations for which signal averaging is desired to enhance detection of small T2-related signal changes during vasodilatory stimulation. Using our method, the myocardial blades collected at rest and during vasodilatory challenge can be retrospectively selected, yielding registered myocardial images for BOLD signal analysis. In clinical application of myocardial viability imaging with delayed enhancement, patients have to hold their breath repeatedly with each high resolution slice acquisition. The respiratory self-gating method could greatly facilitate such routine exam for the patient. By comparison to the streak artifacts characteristic of radial trajectories, the phase-encoding of each PROPELLER blade is essentially a Cartesian trajectory, yielding low resolution blade images without aliasing artifacts. Furthermore, most imaging method applicable to Cartesian encoding can be applied to each blade (e.g. partial Fourier and SENSE). However, contrary to the directionally uniform sampling of radial scans, the PROPELLER blades are prone to achievable precision from the low spatial resolution in the phase-encoding direction, thus limiting the precision of the detected respiratory trace. Nonetheless, our study demonstrated that the respiratory phases could be clearly identified from the estimated traces and the reconstructed images did not show prominent blurring. The PROPELLER encoding has thus the potential to offer a robust method for free-breathing cardiac imaging.

Reference

[1] Larson AC et al, MRM (2005) 53(1):159. [2] Pipe JG ,MRM(1999) 42(5):963. [3] Huang TY et al., 13^th ISMRM (2005), p.521.

Fig.1 Respiratory trace extracted from cross- correlation of blade images cropped

0 10 20 30 40 50 60 70 80 90 100

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Blade number

Correlation coeficient

(5)

over the heart.

Fig.2. Series of low resolution blade images acquired during successive heart beats. Notice that the heart movement can be clearly identified and the respiratory trace can be extracted.

Fig.3. Series of propeller images reconstructed with different CC thresholds (a:0.0 , b:0.5, c:0.9, d:0.95) and corresponding number of blades with cc values higher than the specific threshold (a:100, b:73, c:47,d:27).

(6)

赴國外研究心得報告

計畫編號 NSC 95-2314-B-011-001

計畫名稱自我運動追蹤術：應用於磁振造影血流速測定及高解析度心臟造影術

出國人員姓名

服務機關及職稱黃騰毅

出國時間地點 96/7/6 ~96/7/28

國外研究機構哈佛醫學院/麻省綜合醫院磁共振研究中心

工作記要：

本次研究計畫為國際合作計畫，合作對象為 Martinos’ Center in Harvard Medical School/Massachusetts General Hospital 。而技術研發部份在台灣已經大部份完成，本人於學期結束後，於 7/26 日啟程前往波士頓。在經過一個週末的時差調整後，本人開始於該中心進行為期三個星期的研究。計畫中所提出的自我運動追蹤術，在過去一年的執行中，已完成部份成果。本人的研究生王基崇與本人研究伙伴 Dr.Brigitte Poncelet 並已將研究成果發表於今年在德國柏林舉行之國際磁振造影學會年會。

但由於 Martinos’ Center 的 MRI 系統版本已經更新了許多，所以本人在台灣所撰寫的脈衝序列控制程式並不能直接套用在該中心的系統上。在該中心的相關研究伙伴協助之下，本人將脈衝序列設計軟體將原先的”VA25”提升

至 ”VB13”，經過了兩個星期的努力，本人成功地將合作所需要的脈衝序列控制程式更新至最新的版本，並且在 Martinos’ Center 的 MRI scanner 上測試無誤。

(7)

轉交給研究伙伴 Dr.Brigitte Poncelet 以進行接下去的臨床測試以及動物實驗，而此脈衝序列被證實可於動物實驗時取得高品質影像。

在波士頓期間，除了忙於進行程式修改的工作之外，本人與哈佛醫學院學者 Dr. Kenneth Kwong 以及同在該中心的台大醫工所林發暄教授，也同時若干研究合作計畫，包含多平面激發之面回訊影像、身體擴散權重影像、逆梯度之面迴訊影像修正、高時間解析度之功能性磁振造影等。

所開發之自我運動追蹤之血氧對比磁振造影序列

影像顯示本實驗能夠正確的在

動物實驗(Porcine)上運作，並能偵測血管擴張

(8)

工作時程說明：

7/6~7/9

搭機前往 Boston，經由 LA 轉機。抵達後，安頓自己並調整時差。

7/9~7/15

前往 Martinos 中心，開始進行研究計畫。更新脈衝序列軟體 IDEA 版本。

7/16~7/22

將台灣所開發之脈衝序列，更新至最新版本，並進行實機測試。

7/23~7/26

將成果轉交給研究伙伴 Dr.Brigitte Poncelet，以進行後續實驗。

該研究計畫的初步成果發表如下：

1. (SCI) Huang TY*, Liu YJ, Stemmer A., Brigitte P. (2007) “Free-breathing T2-prepared transient-state TrueFISP: application to myocardial BOLD imaging.” Magnetic Resonance in Medicine 57:960-966 (IF=3.468)

2. Wang CC, Huang TY “Self-gated PROPELLER cine cardiac imaging:

Simultaneously tracking the cardiac pulsation and the respiratory motion” , International Society for Magnetic Resonance in Medicine, Berlin, Germany, 2007

3. B. P. Poncelet-Belliveau, T-Y. Huang, and D. Sosnovik “Myocardial T2 身體擴散張量影像之纖維追蹤術

(9)

imaging - Comparison of Free-Breathing T2-prepared Transient-State TrueFISP to Breath-Hold T2-prepared Segmented True FISP at 1.5T and 3T” International Society for Magnetic Resonance in Medicine, Berlin, Germany, 2007

4. Huang TY, Liu YJ, Juan CY, Chen CY, Poncelet BP, Kwong KK, “Respiratory self-gating with PROPELLER encoding: Application to free-breathing cardiac imaging,” in International Society of Magnetic Resonance in Medicine, 14th Annual Meeting, Seattle, U.S.A., 2006 (Oral presentation )

5. Peng HH, Huang TY, Tseng WY, Chung HW ”Radial Flow-Gated Phase Contrast Imaging in Carotid Artery“in International Society of Magnetic Resonance in Medicine, 14th Annual Meeting, Seattle, U.S.A., 2006