利用心臟斷層影像偵測左心室心包膜外脂肪分佈

Detect of epicardial adipose tissue in Left ventricle using cardiac tomographic images

王瑞揚(Rui-Yang Wang)

指導教授 ： 徐良育

中原大學/工學院/生物醫學工程研究所/碩士(2021年)

https://doi.org/10.6840/cycu202100941

摘要

本研究目的為開發心臟左心室二維影像心外膜脂肪挑選與標記系統，將心內膜、心外膜進行圈選界定，以心外膜為基準進行脂肪挑選，並將脂肪部分進行變色標記，以及脂肪部分在二維影像中的體積評估。
將DICOM二維影像重建成三維影像後，從三維影像中確定左心室影像張數範圍後，利用二維區域成長法進行心內膜圈選，進而取得心內膜資訊，因心外膜型態與心內膜相似度高，將圈選出的心內膜依照設定倍率進行心內膜放大，並以此定位心外膜，最後利用獲得的心外膜資訊為基準執行脂肪挑選，挑選HU值範圍為 -50 至 -200 間，並將挑選的部分進行變色標記與紀錄，同時對脂肪體積計算評估。
透過本研究，使用者可以明顯得知左心室心包膜外脂肪組織於二維心臟電腦斷層影像的位置，並經由系統進行體積評估計算，每一位受試者的十時序脂肪挑選結果的標準差均介於0.5至2.2 之間，且每一位受試者的十時序平均脂肪挑選時間皆在2分鐘內，結果表明，本研究開發的左心室心包膜外脂肪標記系統具有較高的穩定性與可靠性，且成功縮短各時序脂肪挑選所需時間。

關鍵字

心臟電腦斷層影像 ； 心包膜外脂肪組織 ； 二維區域成長法

並列摘要

The aim of this research is to develop a system for detecting epicardial adipose tissue (EAT) in the left ventricle using cardiac tomographic images. By locating the endocardium and epicardium, the EAT can be identified and be colored via epicardium, as well as assessing its volume in 2D images.
After the 3D images were reconstructed using 2D images, and the distribution range of 2D images of the left ventricle was confirmed among 3D images, the information about the endocardium was acquired by locating the endocardium through the 2D region growing algorithm. Since the epicardium and endocardium share high contours similarity, the region of the epicardium can be approximated by enlarging the region of the endocardium with a predefined scale factor. Lastly, the EATs were located based on the HU value of the epicardium, in which the selected range was between -50 and -200, and the selected regions were marked and recorded while estimating its volume at the same time.
Through this research, users can distinctly get the locations of the left ventricle EATs in the 2D computed tomography images of cardiac, in which its volume was also be estimated by the proposed system. As a result, the standard deviation of selected EATs in ten-sequence from each subject lies between 0.5 and 2.2, and the average selecting times of EATs in ten-sequence for each subject were all between 2 minutes. the results show that the proposed left ventricle EATs marking system behave higher stability and reliability, and successfully shorten the needed time to selecting EATs in each sequence.

並列關鍵字

cardiac computed tomography ； epicardial adipose tissue ； 2D region growing algorithm

參考文獻

[1]. L.National Heart, and Blood Institute.(2019) How the Heart Works. Available: https://www.nhlbi.nih.gov/health-topics/how-heart-works

[2]. 蘇虹今, "利用十時序電腦斷層掃描心臟影像建立虛擬動態立體心臟," 中原大學生物醫學工程研究所學位論文, pp. 1-60, 2019.

[3]. R. L. Burden and J. D. Faires, "Numerical analysis (7th)," Prindle Weber and Schmidt, Boston, 2001.

[4]. 楊少康, "建構複數心結構動態模擬工具評估分析左心房左心室之運動功能," 中原大學生物醫學工程研究所學位論文, pp. 1-136, 2013.

[5]. M. Couselo‐Seijas, M. Rodríguez‐Mañero, J. R. González‐Juanatey, and S. Eiras, "Updates on epicardial adipose tissue mechanisms on atrial fibrillation," Obesity Reviews, p. e13277, 2021.