探討切片厚度重建法估算心臟電腦斷層血管攝影左心室容積之準確性

The accuracy of estimated left ventricular volume for Cardiac Computed Tomography Angiography by reconstruction of slice thickness

楊宗烜(Tsung-Hsuan Yang)

指導教授 ： 丁慧枝

共同指導教授 ： 陳泰賓(Tai-Been Chen)

義守大學/電機資訊學院/資訊工程學系/碩士(2018年)

摘要

本研究目的是利用心臟電腦斷層血管攝影（Cardiac Computed Tomography Angiography, Cardiac CTA）影像利用五種幾何公式包含雙平面長軸法（Biplane area-length technique, BALT）、單平面長軸法（Single-plane area-length technique, SPALT）、辛普森法（Modified simpson technique, MST）、圓柱法（Cylineder method, CM）、迴歸估算法（Teichholz method, TM）找出合理的重組條件，提升估算左心室容積準確性。
本研究利用心臟CTA的收縮期至舒張期（10組範例影像），將影像旋轉後重組，新的重組條件為0.35, 0.7, 1.0, 3.0, 5.0 mm與未旋轉(0)，利用五種幾何公式計算左心室容積大小，並與手繪金標準評估其準確性，將所得到之容積利用散佈圖來做線性迴歸分析。另外將3D列印假體的影像，重組五個不同的切面厚度，條件為0.35, 0.7, 1.0, 3.0, 5.0 mm與未旋轉(0)，利用五種幾何公式計算左心室容積大小，並與3D假體真實容積佐證公式的準確性。
實驗結果得知五種幾何方法在未重組與五個重組條件中，BALT與GOT在切面厚度0.35誤差最小，值為1.37，BALT與GOT迴歸係數趨近於1，Beta=1.000，R2=0.999。舒張期結果BALT與GOT在切面厚度0.7所得誤差最小，值為1.30，BALT與GOT迴歸係數趨近於1，Beta=0.971，R2=0.994。收縮期結果BALT與GOT在切面厚度0.35所得誤差最小，值為0.57，BALT與GOT迴歸係數趨近於1，Beta=1.018，R2=0.999。
本研究利用五種幾何方法，針對不同重組條件，估算左心室容積，以BALT方法在切片厚度0.35時較合理並且準確，並利用3D列印假體評估公式的誤差，是有必要的。且利用重組後旋轉細切是可行的，以BALT方法在切片厚度0.35時較合理並且準確。對於數據有差異性，未來可以增加數據量，提高可信度。圈選感興趣範圍時，未來若使用HU值或是灰階度來當作參考，可能可以降低因受測者不同的主觀性。

關鍵字

心臟電腦斷層血管攝影 ； 左心室容積 ； 雙平面長軸法

並列摘要

The purpose of this study was to use the Cardiac Computed Tomography Angiography (Cardiac CTA) image to use five geometric formulas including the Biplane area-length technique (BALT) and the Single-plane area method (SPALT), Modified Simpson technique (MST), Cylinder method (CM), and Teichholz method (TM) were used to find reasonable reconstruction conditions to improve the accuracy of estimated left ventricular volume.
This study used the systolic to diastolic phase of cardiac CTA (10 group of sample images). After the image is rotated, it is reorganized. The new reconstruction conditions were 0.35, 0.7, 1.0, 3.0, 5.0 mm and non-rotating (0), using five geometric formulas to calculation of the left ventricle volume, and evaluate the accuracy with the hand-painted gold standard, the resulting volume was used for linear regression analysis using scatter plots.
In addition, the images of 3D printed phantom use five different section thicknesses were reconstructed. The conditions were 0.35, 0.7, 1.0, 3.0, 5.0 mm and unrotated (0), and using five geometric formulas to calculation of the left ventricle volume, use 3D printed phantom true volume corroborates the accuracy of the formula.
The experimental results show that the five geometric methods are in the unreconstructed and five reconstructed conditions. BALT and GOT had the smallest error at a cut surface thickness of 0.35, with a value of 1.37. BALT and GOT regression coefficients approached 1, Beta=1.000, and R2=0.999. The diastolic phase BALT and GOT had the smallest error at a slice thickness of 0.7, with a value of 1.30. BALT and GOT regression coefficients approached 1, Beta=0.971, and R2=0.994. The systolic outcome BALT and GOT had the smallest error at a cut thickness of 0.35, with a value of 0.57. The BALT and GOT regression coefficients approached 1, Beta=1.018, and R2=0.999.
In this study, five geometric methods were used to estimate the left ventricular volume for different recombination conditions. It is necessary to use the BALT method to be reasonable and accurate at a slice thickness of 0.35, and it is necessary to use the 3D printed phantom to evaluate the error of the formula. It is feasible to use the fine cut after recombination, and the BALT method is more reasonable and accurate when the slice thickness is 0.35. For data differences, the amount of data can be increased and credibility can be improved in the future. When circled the ROI, if the HU value or gray scale is used as a reference in the future, it may be possible to reduce the subjectivity of the subject.

並列關鍵字

Cardiac computed tomography ； LV volume ； Biplane area-length technique

參考文獻

[1] Song I, Ko SM, Yi JG, Chee HK, Kim JS. Differences in Aortic Valve and Left Ventricular Parameters Related to the Severity of Myocardial Fibrosis in Patients with Severe Aortic Valve Stenosis. PLoS One. 2017; 12(1):e0170939.

[2] Aaron CP, Hoffman EA, Lima JAC, Kawut SM, Bertoni AG, Vogel-Claussen J, Habibi M, Hueper K, Jacobs DR Jr, Kalhan R, Michos ED, Post WS, Prince MR, Smith BM, Ambale-Venkatesh B, Liu CY, Zemrak F, Watson KE, Budoff M, Bluemke DA, Barr RG. Pulmonary vascular volume, impaired left ventricular filling and dyspnea: The MESA Lung Study. PLoS One. 2017;12(4):e0176180.

[3] Aviram G, Rozenbaum Z, Ziv-Baran T, Berliner S, Topilsky Y, Fleischmann D, Sung YK, Zamanian RT, Guo HH. Identification of Pulmonary Hypertension Caused by Left-Sided Heart Disease (World Health Organization Group 2) Based on Cardiac Chamber Volumes Derived From Chest CT Imaging. Chest. 2017; 152(4):792-799.

[4] Zhang H, Ma Y, Song Z, Lv J, Yang Y. Predictive value of insufficient contrast medium filling in pulmonary veins in patients with acute pulmonary embolism. Medicine (Baltimore). 2017; 96(37):e7926.

[5] Singh RM, Singh BM, Mehta JL. Role of cardiac CTA in estimating left ventricular volumes and ejection fraction. World Journal of Radiology. 2014; 6(9):669-76.