跳到主要內容

臺灣博碩士論文加值系統

(44.221.73.157) 您好!臺灣時間:2024/06/20 10:48
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

: 
twitterline
研究生:陳栢勛
研究生(外文):Chen, Po-Syun
論文名稱:利用電腦叢集實現兩路相位旋轉式斑點追蹤和最小平方法正則化之四維主應變影像
論文名稱(外文):Cluster Computing Implementation of Ultrasound 4D Principal Stretch Imaging Using Two-Pass Phase-Rotated Speckle Tracking and Least-Squares
指導教授:鄭耿璽
指導教授(外文):Jeng, Geng-Shi
口試委員:黃執中沈哲州羅孟宗李夢麟鄭耿璽
口試委員(外文):Huang, Chih-ChungShen, Che-ChouLo, Men-TzungLi, Meng-LinJeng, Geng-Shi
口試日期:2023-01-03
學位類別:碩士
校院名稱:國立陽明交通大學
系所名稱:電子研究所
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2023
畢業學年度:111
語文別:中文
論文頁數:103
中文關鍵詞:心血管疾病超音波彈性影像斑點追蹤PatchMatch最小平方法台衫一號主軸定理
外文關鍵詞:Cardiovascular diseaseUltrasoundElastic imagingSpeckle trackingPatchMatchLeast squaresTaiwania1Principal theorem
相關次數:
  • 被引用被引用:1
  • 點閱點閱:110
  • 評分評分:
  • 下載下載:4
  • 收藏至我的研究室書目清單書目收藏:0
心血管疾病如心肌梗塞、心室肥大以及心臟衰竭等高居全球十大死因之一,如能發展更為定量之心臟疾病診斷用影像將具有臨床價值。目前臨床上主要採用的影像模式為心臟超音波及核磁共振影像,超音波的臨床應用包含M-mode計算射血率、組織都卜勒影像以及應變影像,其中應變影像採用斑點追蹤法定量心肌應變量。這些模式僅提供一或二維形變估計,無法準確定量複雜的心肌三維運動,也因此許多研究著重於三維心肌形變估計,主要待解問題為如何減少龐大的計算量,以及低幀率限制斑點追蹤之位移估計準確度。此外,臨床應變影像尚有兩大問題,一為只估計正向應變量,忽略了剪切應變量之影響; 二為估計精準度與定義之心臟座標軸有關,每個影像點之心臟座標定義不同,估計結果取決於影像擷取角度與座標軸轉換。為了解決上述這些臨床問題,本論文主要目的有二: 一為加速三維超音波斑點追蹤法之運算時間,二為改善三維應變影像估計準確度。為了加速位移估計,我們採用電腦叢集搭配多執行續運算,實作加速先前所提出以隨機搜索演算法PatchMatch為主之斑點追蹤法。在提升應變影像精準度方面,我們提出傾斜濾波器用於斑點追蹤,能夠解決因為形變過大造成斑點去相關問題。為了完整定量九個應變張量,我們也提出基於最小平方法與心肌不可壓縮性之應變估計法。最後,為了解決心臟座標軸定義問題,我們開發出獨立於座標軸之三維主應變量與軸影像,特別是主應變軸的方向直接能夠定量主要形變方向。所有方法都在三維模擬心臟影像與實際動物影像進行驗證,以及將演算法實現於國家高速網路中心台衫一號,結果顯示本研究整合之電腦叢集運算(使用兩種架構: 900個核心搭配192 GB記憶體,以及560個核心搭配384 GB記憶體)能夠大幅減少等待時間。相較於傳統斑點追蹤法,所提出之傾斜濾波器能夠改善將近10倍的位移估計精準度,此外,在九個應變量的計算上,相較於傳統空間梯度法,本研究所提出之方法可降低1.7倍的估計誤差,因此能夠增加主應變量與軸之估計準確度。最後,藉由模擬、動物正常與心肌異常之結果比對,本研究所提出之主應變軸影像能夠藉由軸向量之方向,成功辨別局部心肌異常之區域,提供一個全新且較為直觀之定量應變影像。
Cardiovascular diseases such as myocardial infarction, ventricular hypertrophy and heart failure are the leading cause of death worldwide. Developing quantitative imaging techniques for diagnostics of heart functions is of clinical value. Currently echocardiography and tagged cardiac magnetic resonance imaging have been routinely used. Imaging modes associated with echocardiography include measurements of ejection fraction in M-mode, tissue Doppler imaging, and myocardial strain (rate) imaging, where strain imaging tracks ultrasound speckles inside the myocardium to quantify deformations. These modes only allow one (1D) or two-dimensional motion estimation, and thus are unable to completely quantify 3D deformations. Extending 2D to 3D strain imaging has been extensively studied. Efforts have been made to reduce computational complexity, and address speckle decorrelation due to low imaging volume rate. In addition, there are still two major problems associated with clinical strain imaging. One is that only normal strains are evaluated without considering shear strains. The other is that the cardiac coordinate uniquely adopted in echocardiography is view dependent, i.e., each pixel inside the myocardium has its own coordinate definition. Estimation accuracy may be affected by different views and coordinate transformation. To address these problems, the purposes of this thesis are (1) to speed up 3D speckle tracking, and (2) to improve the performance of 3D strain imaging. First, we implemented our previously proposed speckle tracking method incorporating randomized searching called PatchMatch by using computer clusters with multiple executions. Second, to improve 3D displacement estimation, we further proposed a phase-rotated correlation filter (called tilt filter). Third, 9 strain tensors are fully estimated using the proposed least-squares (LS) optimization approach constrained by myocardial incompressibility. Finally, we developed a new, coordinate-independent 3D strain imaging based on the principal stretch and axis. Specifically, the principal axis corresponding to the maximum principal stretch indicates the primary deformation direction. All proposed methods were verified by using 3D cardiac simulated and animal data. The core algorithms were implemented on Taiwania1 provided by National Center for High-Performance Computing, where two different configurations of 900 cores with 192 GB memory size and 540 cores with 384 GB memory size were employed. It is shown that our proposed 3D speckle tracking significantly reduces execution time compared to conventional speckle tracking. Besides, tilt filtering improves the estimation accuracy by a factor of almost 10. The proposed LS-based estimation of 9 strain tensors outperforms conventional spatial derivatives by a factor of 1.7. The proposed principal axis enables discrimination of normal and ischemia regions in vivo, providing a new, intuitive, quantitative cardiac strain imaging.
摘要 i
Abstract iii
致謝 v
目錄 vi
表目錄 ix
圖目錄 x
第一章 緒論 1
1.1 心臟疾病 1
1.2 現有三維應變影像 7
1.3 動機 8
1.4 研究目標 12
1.5 論文貢獻 14
1.6 論文架構 14
第二章 現存方法 16
2.1 三維相位敏感式斑點追蹤介紹 16
2.1.1 Normalized Correlation Coefficient (NCC) 17
2.1.2 相關性濾波器 18
2.1.3 基頻信號與位移估計 19
2.2 中位數濾波器 20
2.3 應變計算 20
2.3.1 拉格朗日應變及尤拉應變 21
2.3.2 應變張量(strain tensor)計算 24
2.4 座標系統 26
第三章 方法與材料 28
3.1 測試資料介紹 28
3.1.1 模擬資料 28
3.1.2 成年犬活體資料 32
3.2 二路隨機搜尋斑點追蹤法 35
3.2.1 二路搜尋搭配相關性濾波器相位敏感斑點追蹤 35
3.2.1.1 PatchMatch演算法 35
3.2.1.2 對齊NCC map 36
3.2.2 二路搜尋搭配傾斜濾波器相位敏感斑點追蹤 37
3.2.2.1 傾斜濾波器 41
3.3 基於最小平方法應變之影像估計 44
3.3.1 使用最小平方法估計應變 44
3.3.1.1 9-LS最小平方法 44
3.3.1.2 6-LS最小平方法 48
3.4 主應變影像 52
第四章 結果 55
4.1 位移估計結果 55
4.1.1 模擬資料位移估計結果 55
4.1.2 活體資料位移估計結果 60
4.2 應變估計結果 63
4.2.1 模擬資料應變估計結果 63
4.2.2 活體資料應變估計結果 69
4.3 主應變影像結果 72
4.3.1 模擬資料主應變影像 72
4.3.2 活體資料主應變影像 79
第五章 電腦叢集運算實現以及運算效能分析 81
5.1 運算平台以及選用和時間分析 83
第六章 討論 87
6.1 位移估計結果討論 87
6.2 應變估計結果討論 90
6.3 主應變影像結果討論 92
6.4 效能分析結果討論 94
第七章 結論與未來展望 95
7.1 結論 95
7.2 未來展望 96
參考資料 97
[1] 衛生福利部統計處, 110年死因統計結果分析(含附表).pdf, 2022. Retrieved from https://www.mohw.gov.tw/cp-16-70314-1.html
[2] E. A. Zerhouni, D. M. Parish, W. J. Rogers, A. Yang, and E. P. Shapiro, “Human heart: tagging with MR imaging--a method for noninvasive assessment of myocardial motion,” Radiology, vol. 169, no. 1, pp. 59-63, 1988.
[3] M. Lu, S. Wang, A. Sirajuddin, A. E. Arai, and S. Zhao, “Dynamic stress computed tomography myocardial perfusion for detecting myocardial ischemia: A systematic review and meta-analysis,” International journal of cardiology, vol. 258, pp. 325-331, 2018.
[4] M. S. Amzulescu, M. De Craene, H. Langet, A. Pasquet, D. Vancraeynest, A.-C. Pouleur, J.-L. Vanoverschelde, and B. Gerber, “Myocardial strain imaging: review of general principles, validation, and sources of discrepancies,” European Heart Journal-Cardiovascular Imaging, vol. 20, no. 6, pp. 605-619, 2019.
[5] J. F. Pombo, B. L. Troy, and R. O. RUSSELL JR, “Left ventricular volumes and ejection fraction by echocardiography,” Circulation, vol. 43, no. 4, pp. 480-490, 1971.
[6] A. Kosaraju, A. Goyal, Y. Grigorova, and A. N. Makaryus, “Left ventricular ejection fraction,” 2017. Retrieved from
https://europepmc.org/article/NBK/nbk459131
[7] C. L. Cunha, E. R. Giuliani, V. Fuster, J. B. Seward, R. O. Brandenburg, and D. C. McGoon, “Preoperative M-mode echocardiography as a predictor of surgical results in chronic aortic insufficiency,” The Journal of Thoracic and Cardiovascular Surgery, vol. 79, no. 2, pp. 256-265, 1980.
[8] R. O. L. V. STRAIN, “Head-to-head comparison of global longitudinal strain measurements among nine different vendors,” J Am Soc Echocardiogr, vol. 28, pp. 1171-81, 2015.
[9] M. A. Konstam, and F. M. Abboud, “Ejection fraction: misunderstood and overrated (changing the paradigm in categorizing heart failure),” Circulation, vol. 135, no. 8, pp. 717-719, 2017.
[10] L. Herbots, F. Maes, J. D'hooge, P. Claus, S. Dymarkowski, P. Mertens, L. Mortelmans, B. Bijnens, J. Bogaert, and F. Rademakers, “Quantifying myocardial deformation throughout the cardiac cycle: a comparison of ultrasound strain rate, grey-scale M-mode and magnetic resonance imaging,” Ultrasound in medicine & biology, vol. 30, no. 5, pp. 591-598, 2004.
[11] C. Jia, R. Olafsson, S.-W. Huang, T. J. Kolias, K. Kim, J. M. Rubin, H. Xie, and M. O'Donnell, “Comparison of 2-D speckle tracking and tissue Doppler imaging in an isolated rabbit heart model,” IEEE transactions on ultrasonics, ferroelectrics, and frequency control, vol. 57, no. 11, pp. 2491-2502, 2010.
[12] L. Kapusta, J. M. Thijssen, M. H. Cuypers, P. G. Peer, and O. Daniëls, “Assessment of myocardial velocities in healthy children using tissue Doppler imaging,” Ultrasound in medicine & biology, vol. 26, no. 2, pp. 229-237, 2000.
[13] J.-U. Voigt, G. Pedrizzetti, P. Lysyansky, T. H. Marwick, H. Houle, R. Baumann, S. Pedri, Y. Ito, Y. Abe, and S. Metz, “Definitions for a common standard for 2D speckle tracking echocardiography: consensus document of the EACVI/ASE/Industry Task Force to standardize deformation imaging,” European Heart Journal-Cardiovascular Imaging, vol. 16, no. 1, pp. 1-11, 2015.
[14] S. Queirós, J. L. Vilaça, P. Morais, J. C. Fonseca, J. D’hooge, and D. Barbosa, "Fast left ventricle tracking in CMR images using localized anatomical affine optical flow." pp. 30-36, 2015.
[15] Nesbitt, Gillian C., Sunil Mankad, and Jae K. Oh. "Strain imaging in echocardiography: methods and clinical applications." The international journal of cardiovascular imaging 25.1 (2009): p9-22.
[16] M. Alessandrini, B. Heyde, S. Queirós, S. Cygan, M. Zontak, O. Somphone, O. Bernard, M. Sermesant, H. Delingette, and D. Barbosa, “Detailed evaluation of five 3D speckle tracking algorithms using synthetic echocardiographic recordings,” IEEE transactions on medical imaging, vol. 35, no. 8, pp. 1915-1926, 2016.
[17] R. Jasaityte, B. Heyde, and J. D’hooge, “Current state of three-dimensional myocardial strain estimation using echocardiography,” Journal of the American Society of Echocardiography, vol. 26, no. 1, pp. 15-28, 2013.
[18] A. Elen et al., "Three-Dimensional Cardiac Strain Estimation Using Spatio–Temporal Elastic Registration of Ultrasound Images: A Feasibility Study," in IEEE Transactions on Medical Imaging, vol. 27, no. 11, pp. 1580-1591, Nov. 2008.
[19] R. Mukherjee, C. Sprouse, A. Pinheiro, T. Abraham, and P. Burlina, “Computing myocardial motion in 4-dimensional echocardiography,” Ultrasound in medicine & biology, vol. 38, no. 7, pp. 1284-1297, 2012.
[20] M. A. Lubinski, S. Y. Emelianov, and M. O'Donnell, “Speckle tracking methods for ultrasonic elasticity imaging using short-time correlation,” IEEE transactions on ultrasonics, ferroelectrics, and frequency control, vol. 46, no. 1, pp. 82-96, 1999.
[21] Q. Duan, E. D. Angelini, S. L. Herz, C. M. Ingrassia, K. D. Costa, J. W. Holmes, S. Homma, and A. F. Laine, “Region-based endocardium tracking on real-time three-dimensional ultrasound,” Ultrasound in medicine & biology, vol. 35, no. 2, pp. 256-265, 2009.
[22] G.-S. Jeng, M. Zontak, N. Parajuli, A. Lu, K. Ta, A. J. Sinusas, J. S. Duncan, and M. O’Donnell, “Efficient two-pass 3-D speckle tracking for ultrasound imaging,” IEEE Access, vol. 6, pp. 17415-17428, 2018.
[23] C. Ma, X. Wang, and T. Varghese, “Segmental analysis of cardiac short-Axis views using Lagrangian radial and circumferential strain,” Ultrasonic imaging, vol. 38, no. 6, pp. 363-383, 2016.
[24] Y. Petrank, L. Huang, and M. O'Donnell, “Reduced peak-hopping artifacts in ultrasonic strain estimation using the Viterbi algorithm,” IEEE transactions on ultrasonics, ferroelectrics, and frequency control, vol. 56, no. 7, pp. 1359-1367, 2009.
[25] 國家高速網路計算中心, “Taiwania 1- 中文版使用者操作手冊_v20220412,”2022. Retrieved from
https://iservice.nchc.org.tw/nchc_service/nchc_service_qa.php?target=20
[26] Fei Zheng and E. Ebbini, "A two dimensional sub-sample estimator based on zero phase crossing in ultrasound," 2014 IEEE International Ultrasonics Symposium, 2014, pp. 1081-1084.
[27] Geiman, B., et al. "A comparison of algorithms for tracking sub-pixel speckle motion." 1997 IEEE Ultrasonics Symposium Proceedings. An International Symposium (Cat. No. 97CH36118). Vol. 2. IEEE, 1997 .
[28] A. Støylen. "Basic concepts. Motion and deformation" Retrieved from https://folk.ntnu.no/stoylen/strainrate/Basic_concepts.html
[29] C. Jia, R. Olafsson, K. Kim, T. J. Kolias, J. M. Rubin, W. F. Weitzel, R. S. Witte, S.-W. Huang, M. S. Richards, and C. X. Deng, “Two-dimensional strain imaging of controlled rabbit hearts,” Ultrasound in medicine & biology, vol. 35, no. 9, pp. 1488-1501, 2009.
[30] L. E. Malvern, Introduction to the Mechanics of a Continuous Medium, 1969.
[31] M. Alessandrini, M. De Craene, O. Bernard, S. Giffard-Roisin, P. Allain, I. Waechter-Stehle, J. Weese, E. Saloux, H. Delingette, and M. Sermesant, “A pipeline for the generation of realistic 3D synthetic echocardiographic sequences: Methodology and open-access database,” IEEE transactions on medical imaging, vol. 34, no. 7, pp. 1436-1451, 2015.
[32] N. Parajuli, A. Lu, K. Ta, J. Stendahl, N. Boutagy, I. Alkhalil, M. Eberle, G.-S. Jeng, M. Zontak, and M. O’Donnell, “Flow network tracking for spatiotemporal and periodic point matching: Applied to cardiac motion analysis,” Medical image analysis, vol. 55, pp. 116-135, 2019.
[33] A. Østvik, I. M. Salte, E. Smistad, T. M. Nguyen, D. Melichova, H. Brunvand, K. Haugaa, T. Edvardsen, B. Grenne, and L. Lovstakken, “Myocardial function imaging in echocardiography using deep learning,” IEEE Transactions on Medical Imaging, vol. 40, no. 5, pp. 1340-1351, 2021.
[34] A. Lu, S. S. Ahn, K. Ta, N. Parajuli, J. C. Stendahl, Z. Liu, N. E. Boutagy, G.-S. Jeng, L. H. Staib, and M. O’Donnell, “Learning-based regularization for cardiac strain analysis via domain adaptation,” IEEE transactions on medical imaging, vol. 40, no. 9, pp. 2233-2245, 2021.
[35] C. Barnes, E. Shechtman, A. Finkelstein, and D. B. Goldman, “PatchMatch: A randomized correspondence algorithm for structural image editing,” ACM Trans. Graph., vol. 28, no. 3, pp. 24, 2009.
[36] E. Y. Wong, M. O'Donnell, C. B. Compas, B. A. Lin, A. J. Sinusas and J. S. Duncan, "Phase rotation in correlation coefficient filtering and multi-pass methods for 3-D speckle tracking in 4-D echocardiography," 2014 IEEE International Ultrasonics Symposium, 2014, pp. 1-4,
[37] L. Huang, Y.-K. Petrank, S.-W. Huang, C. Jia, and M. O'Donnell, “Phase rotation methods in filtering correlation coefficients for ultrasound speckle tracking,” IEEE transactions on ultrasonics, ferroelectrics, and frequency control, vol. 56, no. 7, pp. 1368-1382, 2009.
[38] C. Kasai, K. Namekawa, A. Koyano, and R. Omoto, “Real-time two-dimensional blood flow imaging using an autocorrelation technique,” IEEE Transactions on sonics and ultrasonics, vol. 32, no. 3, pp. 458-464, 1985.
[39] F. Kallel, and J. Ophir, “A least-squares strain estimator for elastography,” Ultrasonic imaging, vol. 19, no. 3, pp. 195-208, 1997.
[40] A. Bistoquet, J. Oshinski, and O. Škrinjar, “Myocardial deformation recovery from cine MRI using a nearly incompressible biventricular model,” Medical image analysis, vol. 12, no. 1, pp. 69-85, 2008.
[41] Teed, Z., Deng, J. (2020). RAFT: Recurrent All-Pairs Field Transforms for Optical Flow. In: Vedaldi, A., Bischof, H., Brox, T., Frahm, JM. (eds) Computer Vision . ECCV 2020: Computer Vision – ECCV 2020 pp 402–419, vol 12347, 2020.
[42] R. E. Kalman, “A new approach to linear filtering and prediction problems,” 1960.
[43] D. Syed et al., "Investigation on Optimizing Cost Function to Penalize Underestimation of Load Demand through Deep Learning Modeling," IECON 2021 – 47th Annual Conference of the IEEE Industrial Electronics Society, 2021, pp. 1-7.
連結至畢業學校之論文網頁點我開啟連結
註: 此連結為研究生畢業學校所提供,不一定有電子全文可供下載,若連結有誤,請點選上方之〝勘誤回報〞功能,我們會盡快修正,謝謝!
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top
無相關期刊