【摘要】：隨著醫(yī)學(xué)診斷和治療模式的改變,使得醫(yī)療活動(dòng)的模式也發(fā)生了巨大轉(zhuǎn)變。主要表現(xiàn)在計(jì)算機(jī)輔助檢測(cè)的快速發(fā)展上。肺部呼吸系統(tǒng)疾病是一類重要的非傳染性流行疾病。計(jì)算機(jī)輔助檢測(cè)在肺部疾病的檢測(cè)和診斷中的應(yīng)用得了充分的肯定。肺部氣管的三維樹狀結(jié)構(gòu)分析方法的研究是當(dāng)前計(jì)算機(jī)輔助檢測(cè)系統(tǒng)的一個(gè)重要分支。 本課題的目的是提出了新的三維肺部氣管樹狀結(jié)構(gòu)分析算法,實(shí)現(xiàn)了氣管骨架的自動(dòng)提取,進(jìn)而提取出氣管中心線,正式生成樹形結(jié)構(gòu),最后獲取氣管2D橫截面。本課題的CT影像數(shù)據(jù)均來自遼寧沈陽(yáng)中國(guó)醫(yī)科大學(xué)附屬盛京醫(yī)院。在本課題中首先對(duì)CT數(shù)據(jù)提取二值化氣管的骨架,提出拓?fù)浼?xì)化算法,滿足細(xì)化算法要求的四個(gè)條件,通過簡(jiǎn)單點(diǎn)判斷,刪除邊界點(diǎn)直到只剩下單體素的骨架。對(duì)于細(xì)化后多余的分支,通過剪枝算法,獲得剪枝后的骨架。為了生成正式樹形結(jié)構(gòu),對(duì)特征點(diǎn)進(jìn)行識(shí)別,即骨架上的分叉點(diǎn)、線上點(diǎn)、線端點(diǎn)和根點(diǎn)。采用二叉樹拓?fù)浣Y(jié)構(gòu)生成三維肺部氣管樹形結(jié)構(gòu),對(duì)不同級(jí)別氣管進(jìn)行分級(jí)別存儲(chǔ)。最后平滑骨架,得到氣管道橫截面法向量,從而通過坐標(biāo)變換獲取2D橫斷面。 本課題對(duì)18組數(shù)據(jù)進(jìn)行了實(shí)驗(yàn)分析,提取出了其肺部氣管骨架,并且通過剪枝算法設(shè)置不同閡值得到的氣管骨架可以滿足不同臨床要求。生成了正式樹形結(jié)構(gòu),并且實(shí)現(xiàn)分級(jí)存儲(chǔ)信息,標(biāo)記了各級(jí)氣管信息,最后獲得了各級(jí)支氣管的2D橫斷面數(shù)據(jù)。本課題提出的三維肺部氣管樹狀結(jié)構(gòu)分析算法,為定量分析打下了基礎(chǔ),同時(shí)可滿足臨床需求,對(duì)氣管道疾病的診斷提供了幫助。
[Abstract]:With the change of the mode of medical diagnosis and treatment, the mode of medical activity has changed greatly. The main performance is the rapid development of computer aided detection. Pulmonary respiratory disease is an important non-communicable epidemic disease. The application of computer aided detection in the detection and diagnosis of pulmonary diseases has been fully affirmed. The research of three-dimensional tree structure analysis method of lung trachea is an important branch of computer aided detection system. The purpose of this paper is to propose a new three-dimensional trachea tree structure analysis algorithm, which can automatically extract the trachea skeleton, then extract the trachea center line, formally generate the tree structure, and finally obtain the trachea 2D cross section. The CT image data of this subject are all from Shengjing Hospital affiliated to Shenyang Medical University of Liaoning Province. In this paper, the skeleton of binary trachea is extracted from CT data, and a topology thinning algorithm is proposed, which meets the four requirements of thinning algorithm. Through simple point judgment, the boundary point is deleted until only the skeleton of single element is left. For the redundant branches after thinning, the skeleton after pruning is obtained by pruning algorithm. In order to generate the formal tree structure, the feature points are identified, that is, the bifurcation points, the line endpoints and the root points on the skeleton. Three dimensional lung trachea tree structure was generated by binary tree topology, and different grades of trachea were stored. Finally, the normal cross section vector of gas pipeline is obtained by smoothing the skeleton, and the 2D cross section is obtained by coordinate transformation. In this paper, 18 groups of data were analyzed experimentally, and the tracheal skeleton of lung was extracted, and the tracheal skeleton with different threshold was set by pruning algorithm to meet different clinical requirements. The formal tree structure was generated, and the information was stored at different levels, and the tracheal information was labeled. Finally, 2D cross-sectional data of all bronchi were obtained. The three-dimensional trachea tree structure analysis algorithm proposed in this paper lays a foundation for quantitative analysis and can meet the clinical needs and provide help for the diagnosis of gas pipeline diseases.
【學(xué)位授予單位】：東北大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2013
【分類號(hào)】：R816.41;TP391.72

【參考文獻(xiàn)】

相關(guān)期刊論文 前5條

1 歐宗瑛,秦緒佳,紀(jì)鳳欣,吳良武,張勇,侯建華;多葉光柵適形放射治療系統(tǒng)圖像圖形處理軟件研究與開發(fā)[J];大連理工大學(xué)學(xué)報(bào);2001年06期

2 符祥;郭寶龍;;三次均勻B樣條插值曲線和曲面的矩陣形式[J];光電子.激光;2007年08期

3 滕奇志;康瑕;唐棠;何小海;;基于升序復(fù)核的并行三維圖像骨架化算法[J];光學(xué)精密工程;2009年10期

4 段佐剛,譚小蘋,裴覺民;肺氣管樹分形結(jié)構(gòu)的數(shù)學(xué)描述[J];四川大學(xué)學(xué)報(bào)(工程科學(xué)版);2003年04期

5 黃坤武;唐杰;武港山;;針對(duì)面片的Reeb圖骨架抽取算法[J];系統(tǒng)仿真學(xué)報(bào);2006年S1期

，

本文編號(hào)：2291443