最近幾年便攜式超音波越來越普及，這些便攜式超音波若能用於居家照護，有機會讓用戶可以在家監測自己的健康狀況，而居家照護中的一個項目是由動脈硬化引起的心血管疾病。動脈硬度是心血管疾病的一個指標，已經通過大量研究得到證實。脈搏波速度（PWV）是動脈硬化程度的一個指標，可以預測心血管疾病。在此研究中，我們提出新的頸動脈脈波相關參數，並驗證其用來判別受試者心血管的健康狀況的可行性。在此研究中我們會使用的頸動脈脈波相關參數有基於頸動脈直徑擴張波形的收縮足、收縮期峰值和重搏切跡估計之局部PWV、計算上述三種PWV的調整R平方和基於收縮足與重博切跡的PWV差異 (PWV_sf-PWV_dn)。估計上述參數時所需動脈壁位移的估計是對31個患者包括9個健康病例，14個心血管病例和8個頸動脈斑塊病例的超音波射頻信號使用1 D複數互相關的斑點追蹤和相位零交叉方法計算取得。取得上述參數後，在這項研究中，我們使用LIBSVM藉由這7個頸動脈脈搏波關聯參數對31例患者進行心血管狀況正常及異常的分類，分類（正常和異常標記）檢驗準確率為70.97％而AUC（ROC曲線下面積）為 0.70，比單用收縮足PWV來進行分類更佳。此外，我們還嘗試將第一次分類的21個異常標記細分為心血管和頸動脈斑塊標籤，分類（其他心血管疾病和具頸動脈斑塊標記）的測試準確度分別為71.43％和AUC為0.60，大於0.5。

In recent years, portable ultrasound has become more and more popular. The portable ultrasound may have the chances to be used for home care so that users can monitor their health status at home. One of the potential portable-ultrasound applications in home care is to identify cardiovascular health status related to arterial stiffness. Arterial stiffness as an indicator of cardiovascular diseases has verified by a lot of studies. The local pulse wave velocity (PWV) is an index of the measurement of arterial stiffness and has been shown to has the potential in classification of cardiovascular health status. In this thesis, we propose several parameters associated with carotid arterial pulse wave and try to verify the feasibility of identification of cardiovascular health status using the proposed parameters. These parameters are local PWVs estimated via systolic foot, systolic peak, and dicrotic notch of the carotid-diameter distension waveforms, the difference of PWV (PWV_sf-PWV_dn) from systolic foot and dicrotic noth, and adjusted R-squared derived from the estimation procedure of the three PWVs. The technique used for the estimation of the arterial wall displacement required for PWV calculation is the 1 D complex cross correlation technique over ultrasound RF data of the patient along with phase zero crossing tuning. The used ultrasound RF data are acquired from 31 patients, including 9 healthy cases, 8 carotid plaque cases, and 14 cases with other cardiovascular diseases. With the above seven parameters being estimated, the support vector machine tool – LIBSVM is used for the classification of the cardiovascular health status of the 31 patients. For the normal and abnormal classification, the test accuracy is 70.97% and the AUC (area under the ROC) is 0.70. Both are higher than those classified only with systolic-foot PWV. Furthermore, we also try to sub-classify the cases labeled as abnormal in the first classification into carotid plaque and other cardiovascular disease labels. The accuracy of the sub-classification is 71.43% and the AUC is 0.6, which is larger than 0.5.

摘要 I
Abstract II
Table of Contents IV
List of Figure VI
List of Table IX
Chapter 1. Introduction 1
1.1 Portable ultrasound 1
1.2 Pulse wave velocity 3
1.2.1 Global pulse wave velocity 4
1.2.2 Local pulse wave velocity 6
1.3 Motivation 8
1.4 Thesis organization 10
Chapter 2. Materials and Methods 11
2.1 Data acquisition 11
2.2 Parameters associated with carotid arterial 13
2.2.1 Pulse wave velocity 13
2.2.1.1 Displacement waveforms of the anterior and posterior wall 13
2.2.1.2 Distension waveform 18
2.2.1.3 Pulse wave velocity calculation 20
2.2.2 Adjusted R squared 23
2.2.3 Difference of PWV (PWVsf-PWVdn) 26
2.3 Classifier 26
2.3.1 The parameters of SVM model 27
2.3.2 Grid search with K-fold cross-validation 28
2.3.3 Input data of classifier 31
Chapter 3. Results and Discussions 33
3.1 The data of 7 features in different cases 33
3.2 Result of the classification 34
3.2.1 The classification for dataset of PWV 34
3.2.2 The classification for dataset of 7 features 36
3.2.3 The classification for dataset of 6 features 38
3.2.4 The dichotomy of classification for dataset of 7 features 40
3.3 Discussion of classification 42
Chapter 4. Conclusions and Future Work 44
4.1 Conclusions 44
4.2 Future work 44
References 45

[1]https://www.butterflynetwork.com
[2]https://www.ahajournals.org/doi/full/10.1161/HYPERTENSIONAHA.118.10839
[3]Huang C, Su Y, Zhang H, Qian LX, Luo J, “Comparison of different pulse waveforms for local pulse wave velocity measurement in healthy and hypertensive common carotid arteries in vivo,” Ultrasound in Med. & Biol, pp. 1–13, May, 2016.
[4]Huang C, Su Y, Zhang H, Qian LX, Luo J, “Pulse wave velocity measurement in healthy and diseased carotid arteries in vivo,” in: Proc. IEEE ULTSYM, 2015.
[5]Luo J, Li RX, Konofagou EE, “Pulse wave imaging of the human carotid artery: an in vivo feasibility study,” IEEE Trans Ultrason Ferroelectr Freq Control, vol. 59(1), pp. 174-81, Jan, 2012.
[6]G Zahnd, D Galbrun, S Qorchi, K Saito, A Sérusclat, P Moulin, et al, “Pattern analysis of the kinematics in ultrasound videos of the common carotid artery – application to cardiovascular risk evaluation,” in: Proceedings of International Forum on Medical Imaging in Asia (IFMIA), 2017.
[7]Li RX, Luo J, Balaram SK, Chaudhry FA, Lantis JC, Shahmirzadi D, Konofagou EE, “In vivo application of pulse wave Imaging for arterial stiffness measurement under normal and pathological conditions,” 33rd Annual International IEEE EMBS Conference, Boston, MA, pp. 567–570, 2011.
[8]Huang C, Ren T, Luo J, “Effects of parameters on the accuracy and precision of ultrasound-based local pulse wave velocity measurement: a simulation study,” IEEE Trans Ultrason Ferroelectr Freq Control, vol. 61(12), pp. 2001–2018, Dec, 2014.
[9]Luo J, Li RX, Konofagou EE, “Pulse wave imaging of the human carotid artery: An in vivo feasibility study,” IEEE Trans Ultrasonics Ferroelectr Freq Control, vol. 59(1), pp. 174–181, Jan, 2012.
[10]Vappou J, Luo J, Konofagou EE, “Regional measurement of arterial stiffness using pulse wave imaging: phantom validation and preliminary clinical results,” in: Proc. IEEE Ultrason. Symp, 2010.
[11]Chen XC, Zohdy MJ, Emelianov SY, O’Donnell M, “Lateral speckle tracking using synthetic lateral phase,” IEEE trans UFFC, vol. 51, pp. 540–550, May, 2004.
[12]Luo J, Konofagou EE, “A fast normalized cross correlation calculation method formotion estimation,” IEEE Trans Ultrason Ferroelectr Freq Control, vol. 57, pp. 1347-57, Jun, 2010.
[13]Hermeling E, Reesink KD, Reneman RS, Hoeks AP, “Confluence of incident and reflected waves interferes with systolic foot detection of the carotid artery diameter waveform,” J Hypertens, vol. 26(12), pp. 2374–2380, Dec, 2008.
[14]Hermeling E, Reesink KD, Kornmann LM, Reneman RS, Hoeks AP, “The dicrotic notch as alternative time-reference point to measure local pulse wave velocity in the carotid artery by means of ultrasonography,” J Hypertens, vol. 27(10), pp. 2028–2035, Oct, 2009.
[15]Luo J, Ying K, P. He, and J. Bai, “Properties of savitzky-golay digital differentiators,” Digit. Signal Process., vol. 15, no. 2, pp. 122– 136, Mar. 2005.
[16]https://www.csie.ntu.edu.tw/~cjlin/libsvm/