微循環影像非線性對位血流測量

Nonlinearity Image Registration of Microcirculation Image for Blood Flow Measurement

陳冠州(Kuan-Chou Chen)

指導教授 ： 林康平

中原大學/電機資訊學院/電機工程研究所/碩士(2017年)

https://doi.org/10.6840/cycu201700690

摘要

微循環與腫瘤組織有密不可分的關係，而透過觀察組織中微血管的血流速做為參考資料。本論文著重於內視鏡胃癌影像，由於活體蠕動與醫師操作造成的旋轉、放大、位移等型變且有內視鏡造成的光學失真，因此本論文提出非線性的對位方法。文中模擬了內視鏡影像的失真，並對失真後的影像進行對位，失真的影像有方格影像、模擬血流影像與真實蛙蹼血流影像。
常見的影像對位方式有兩種，相關係數法與傅立葉轉換法，但兩者處理內視鏡影像時，前者因為特徵的計算並不容易，後者因為相位變換過大處理不佳。本論文的方法目標是達到處理每張影像空間移動量大，且能應對廣角鏡造成的非線性失真的準確的對位方法，使用Thin-plate Spline(TPS)為核心結合A *(A-Star)搜尋法解決影像對位上的問題。
對位後的方格影像的面積變化確定旋轉、放大、位移對本論文方法的影響程度，並使用光流法估測對位後的模擬血流與蛙蹼血流影像的移動向量進驗證，其模擬血流的誤差在1.94 pixel/frame，蛙蹼血流的誤差在1.43 pixel/frame，從上述結果可以證實本方法確實能夠準確地針對內視鏡影像對位，在微循環影像的研究將可提供此方法作為影像對位的工具。

關鍵字

微循環 ； 血流速 ； 影像對位 ； 影像處理

並列摘要

Tumor tissue and microcirculation blood flow information indicated strong relationship. By observing the micro-vascular blood flow in the tissues and organs revel an useful reference for comparison of normal or abnormal parts. This these focuses on the study of video images of micro-vascular of gastric cancer recorded by endoscopy. Since of the irregular and unstable rotating, scaling and moving images caused by the movement of the body distortion, the operation of the physician, and the nonlinear optical distortion of endoscopy, there is required a method for solving these problems. In this these, the distortion of the endoscopic image is simulated. The proposed method has presented results for the distorted and gridded images to corrected and aligned target images. The simulated blood flow image and real frog web image have also been evaluated in this study.
There are two common methods of image alignment, correlation coefficient method and Fourier transform method. However, the two methods cannot deal with endoscopic images od nonlinear distortion. The former connot affort the heavey feature mattching problem. The latter cannot proper deal with the poor phase shift problem. The aim of this thesis is to deal with the accurate alignment method for correcting the distortion on each sequencial image and to deal with the nonlinear distortion caused by the wide-angle lens. Using Thin-plate Spline (TPS) as the core technique and combining the A -star search method, the thesis has solve the problem of image alignment among sequencial images.
To study the accuracy of the proposed method in this thesis, the image distortion of rotation, scaling and moving problems has been evaluated by the change of the area of the gridded image. The optical flow method is used to estimate the motion direction of the simulated blood flow and the frog web flow image. The error of the simulated blood flow is 1.94 pixel / frame, and the error of the blood flow is 1.43 pixel / frame. From the above results, it can be confirmed that the method can accurately target the endoscopic image, and the microcirculation image. This method can be provided as a tool for image alignment of endoscopy video images.

並列關鍵字

Microcirculation ； Blood flow ； Image registration ； Image processing

參考文獻

[23]. 林汶正，顯微造影技術在裸鼠微循環血管之血流速分析研究，中原大學電機工程學系碩士論文，2011。

[49]. 林東儒，動態顯微影像估測為循環血流速，中原大學電機工程學系碩士論文，2012。

[19]. Green, William DK. "Spline-based deformable models." SPIE's 1995 International Symposium on Optical Science, Engineering, and Instrumentation. International Society for Optics and Photonics, 1995.

[2]. Wollersheim, H., Reyenga, J., & Thien, T. H. (1988). Laser doppler velocimetry of fingertips during heat provocation in normals and in patients with raynaud's phenomenon. Scand J Clin Lab Invest, 48(1), 91-95. doi:10.3109/00365518809085399

[3]. Cesarone, M. R., Laurora, G., & Belcaro, G. V. (1992). Microcirculation in systemic hypertension. Angiology, 43(11), 899-903. doi:10.1177/000331979204301104