【作者】 曲德伟；

【导师】 李幼琼；

【作者基本信息】 吉林大学 ， 人体解剖与组织胚胎学， 2004， 硕士

【摘要】 在医学领域中，由于人体组织器官构造的复杂性和形态的多样性以及疾病的病灶位置与形态的不可预知性，因此没有相当专业知识和实践经验，很难在头脑中，通过这种二维图像信息进行整合的基础上构想出组织器官的立体形态以及这些组织器官与周围毗邻组织或器官之间的相互关系，并且由于个人思维方式的不同，则必然产生一定程度的偏差、甚至是错误。因此仅仅停留在二维图像上是很难满足日益发展的临床诊断、解释临床症状以及保证所制定医疗方案的正确性和客观性的要求。本课题的根本特点在于最终将完成人体整体三维重建工作，进而完成中国人体数字化研究，建立中国数字化人体。建成后可以进行多个器官、多种组织重叠或单独显示，而且同时具有方法简便、易于推广、实用性强等特点，在国内同类型研究中居于领先水平。本课题立题的目的在于建立一套正常人颅骨的三维立体图像与显示系统并可以不同角度观察和测量各结构的各种界限值，以便量化颅骨诸结构。材料与方法：1.断面图像标本的制备：取正常男性尸体一具，经检查判定，拟用的尸体头颅正常、无畸形，颅内无器质性病变。经福尔马林溶液固定处理一年，将离断后的头颅置于有机玻璃容器内浸泡，经弱<WP=29>酸反复脱钙后，以四根红线与固定盒四边平行放置以建立外参照系，并且于切片标本旁放置刻度尺，以表示切片标本的原始形态学数据，以其作为定位点间的相互距离和作为三维立体重建过程中的实际标准。然后进行冰冻处理。以眶耳平面为基准平面，取间距为1mm，开始断层切片，完成切片后共获得质量满意切片240片，无损耗。2.硬件系统：本课题所采用的真彩扫描仪为MICROTEK Scan Maker C6真彩扫描仪，高清晰36bit色彩位数，最大光学分辨率600×12OODPI；数码照相机为DC–260相机，1536×1024像数；计算机为目前国内主流机型配置，PENTIUM4，内存256MB，硬盘80GB，DELL纯平显示器。3.图像数字转换（A/D转换）：将端面标本按顺序逐片置于扫描仪平台上，将断面的原始图像转换成数字图像存入计算机硬盘； 4.图像坐标校正（对位）：三维立体重建过程中，各层面间的坐标位置的校正至关重要，它直接影响立体重建后诸结构空间位置的准确性。因此我们在计算机系统内装有程序对位软件，几何对位系统，对位精度为0.25mm。DC260变焦数位相机直接输入的图像信息，将由计算机系统内的软件自动完成对位。5.二维图像的处理：将以输入计算机硬盘内的二维数字图像（0—256个灰度级，24位色彩）调出，与原标本逐一对照，利用计算<WP=30>机所带的软件包对原始图像进行处理，其中包括灰度校正，图像边缘锐化等，经上述处理后，显示在计算机屏幕上的图像更加清晰，对所需要的感兴趣区的结构更加容易识别。6.对拟立体重建结构边缘的提取：在经过处理和位置校正后的二维图像上，对拟立体重建的结构（感性趣区）进行人工勾画提取边界和图像分割。将拟重建结构按各自命名采取逐片人工提取勾画该结构的边界，保存三维重建所需要的图像信息。7.三维重建的计算方法：对拟建结构的边界进行双参数样条插值，以提高重建后的显示效果。采用双参数样条插值计算，具有光滑效果好，连续性好，边界拟合好等特点。通过对重建结构边缘轮廓线凸包计算后，相邻断面结构边缘轮廓线采用“盖瓦片”的方法连接。然后，经过R，G，B彩色赋值、光照计算和消隐计算，即可以实现重建结构的三维立体显示。如若需要实现旋转或剖割，输入有关参数后则可进行坐标变换计算，再按新的坐标位置进行显示。结果：通过电子计算机，对240层人头部簿层连续解剖断面进行A/D转换、二维图像处理、图像分割和对位重建，成功的在计算机二维屏幕，对颅骨诸结构进行了三维重建及立体显示。显示的图像平滑、形态结构逼真、清楚、层次分明，可显示出所建结构各个不同侧面的凹凸形态。重建的所有结构即可以单独显示，整体显示，<WP=31>也可以任意搭配显示，既可以原色显示，也可以实现任意色彩显示。所有重建的诸结构均可以在三维空间位置上绕任意轴旋转任意角度，也可以在鼠标的控制下任意连续旋转。重建结构可以任意拆分、组合及任意剖割显示，因此重建结构即便于单一形态结构的观察，又便于显示某一结构在整体中的位置关系，以及显示各个结构在不同角度的形态、空间位置及其相互毗邻关系。可以对所有结构在任意方向上的经线和角度实时测量，为三维重建结构的定量研究提供了必要的手段。重建后的诸结构显示速度飞快，在动态显示过程中，全彩显示从一个角度到另一个角度的变换过程仅需要0.5秒就能完成，虽然还不达到小型机的处理速度，但是完全能达到三维立体显示的要求，与国内外同类型的研究相比，充分显示出了系统整体配置简单，易操作，经济实用等特点，因此便于在国内的普及推广和利用。更多还原

【Abstract】 Because the organs of our bodies are complex and different and the location and situation of disease is unpredictable that it is very hard for us to image both three-dimensional structure of the organ and the relationship the organ with others through information we got from two-dimensional images if without enough experience and knowledge. And different way of thinking leads to different understanding. So information we got from two-dimensional images could not satisfy the request of development of clinical medical sciences. The character of this study is it is a part of work belonging to building up a digital body of human being and this body could show several oranges overlapping or signally at a conventional and practical way. The study is advanced in this field in our country. The purpose of this study is to build up three-dimensional image of normal human cranium and make it could be seen for different viewpoint and could measure the threshold of every structures. Materials and Methods1. Preparation of cross-section specimens: the head of a normal male body treated with 10%formalin for a year and put it in a Lucite container. After decalcification by soft acid several times, we Create exterior reference standard by parallel placement of four read line with edges of fixation box, and put mark rule on section preparation to show primitive morphological data of section preparation and use it for practice standard of three-dimension rebuilding Coordinate. Then make freezing treatment. The standard plane is orbit-ear plane. The spacing is 1mm to make section specimens. After accomplishing we got 240 satisfaction section，no wastage.2. Hardware Systems: Scanner is MICROTEK Scan Maker C6 true color, 36bits, 600×12OODPI；Digital camera we used is DC–260; 1536×1024 picture elements；Computer is PENTIUM4，CPU–700MHz，RAM 256MB，Hard disk <WP=33>80GB。3. A/D changing: put the cross-section specimen onto panel of scanner and use software of windows to scan, then change the primary image into digital image and save in hard disk.4. Correct the coordinate of image: In the process of three-dimensional rebuilding, the correction of relationship of different layer is most important. It affects the accuracy of location of all structure in three-dimension after rebuilding. We install correction software before operation. And accuracy of correction is 0.25mm. DC260 digital camera input the information of image directly into computer and the correction software do correct work auto.5. Treatment for two-dimensional image：open the file of two-dimensional images saved in hard disk of computer（0-256 gray scale resolution，24 bits color）and compared with origin specimen respectively. We treat origin specimen by software sack, which include correction of gray scale, sharpen of edge of image, etc. After treatment list above, the image on screen could be more clear and more easily to recognize。6. Obtain the edge of preparing rebuilding structure: after treatment for two-dimensional image，we draw the edge of preparing rebuilding structure artificial one bye one according its denomination and divide the image and save the information.7. Method of calculation of three-dimension rebuilding：The method is to plug belt transect in the edge of structure preparing to rebuild to improve the apparent effect of rebuilding。The features of this method are smooth and glossy , continuous, border converging better。After calculation for three-dimension rebuilding，we <WP=34>connect border outline of break section close together by “cover tile” method。Then display three-dimension rebuilding after R，G，B color endowment、light calculation and recede and hidden calculation. If revolution and dissection is necessary ,we should put the relative parameter to change coordinate then the image would be showed on the screen according to position of new coordinate。Result：we use computer technique to do A/D changing、treatment for two-dimensional image、dividing image and correction for 240 continuous layers of head of human bei更多还原