基于二次成像技术标志点获取和自动配准的长段周围神经功能束三维可视化关键技术研究  

KEY TECHNOLOGIES OF FUNCTIONAL FASCICULAR GROUPS THREE DIMENSIONAL VISUALIZATION OF LONG SEGMENTED PERIPHERAL NERVE BASED ON TWO-TIME IMAGING TECHNIQUE AND AUTOMATIC REGISTRATION

作　　者：张毅[1] 刘小林[2] 钟映春[3] 唐平[3] 周家铭[4] 戚剑[2] 何彩凤[5] 

机构地区：[1]中山大学附属第一医院整形修复外科,广州510080 [2]中山大学附属第一医院显微创伤骨科,广州510080 [3]广东工业大学自动化学院 [4]中山大学附属第六医院科研科 [5]中山大学附属第一医院骨科研究所,广州510080

出　　处：《中国修复重建外科杂志》2010年第3期325-331,共7页Chinese Journal of Reparative and Reconstructive Surgery

基　　金：广东省自然科学基金资助项目(9151008901000006);广东省医学科研基金资助项目(A2009173)~~

摘　　要：目的探讨四肢周围神经功能束三维可视化重建过程中的关键技术问题及解决方法,并验证其可行性。方法取自愿捐献新鲜成人尸体右上臂20cm尺神经标本。以4根成年女性头发作为标志线,连续横断面冰冻切片,厚15μm,层距0.5mm,共400张切片,行乙酰胆碱酯酶(acetylcholinesterase,AchE)染色;采用二次成像技术,以配置MSHTO MD90显微数码成像装置的Olympus Z61型体视显微镜于AchE染色前、后分别获取同一切片的2张全景图像,通过Photoshop图像处理软件图层叠加进行图像处理,获取含4个完整定位标志点的神经断面二维全景图像;以优化最小二乘支持向量机方法作为分类器自动识别处理图像,并利用空间变换方法实现自动配准,结合人工辅助获取功能束轮廓,通过Amira 4.1医学三维重建软件进行长段尺神经功能束三维重建,评估重建逆向还原效果。结果利用二次成像Photoshop图像处理软件处理同一张切片染色前、后的图像,图像轮廓吻合度良好,合成图像获得的标志点定位精确,操作简便快捷。优化最小二乘支持向量机方法识别精度高,误差率为8.250%;三维模型可直观观察神经内部不同性质功能束交叉融合的变化,其任意水平切割逆向还原基本吻合。结论基于二次成像技术和Photoshop图像处理软件图层处理技术的计算机自动配准可应用于长段周围神经的三维可视化重建,重建快捷,效果较好。Abstract： Objective To explore and solve the key technologies of the three dimensional （3D） visual ization reconstruction of functional fascicular groups inside long segmented peri pheral nerve. Methods A 20 cm ulnar nerve from upper arm of fresh adult dead body was embedded by OCT with four pieces of woman＇s hair which was used as locating material, then the samples were serially horizontally sl iced into 400 sl ices with 15 μm thickness and 0.5 mm interval. All sl ices were stained with acetylcholinesterase （AchE） histochemical staining. After that, the 2D panorama images of the same sl ice were obtained with Olympus stereomicroscope and MSHOT MD90 micro figure image device before and after AchE staining. Using the layer processing technique of Photoshop image processing software, the recomposition images including complete 4 location pots were obtained, based on which the algorithm of optimized least square support vector machine （Optimized LS-SVM） and space transformation method was used to fulfill automatic registration. Finally, with artificial assistant outline obtaining, the 3D visual ization reconstruction model of functional fascicular groups of 20 cm ulnar nerve was made using Amira 4.1, and the effects of reverse reduction and the suitabil ity of 3D reconstruction software were evaluated. Results The two-time imaging technique based on the layer process of Photoshop image processing software had the advantages： the image outline had high goodness of fit; the locating pots of merging image was accurate; and the whole procedure was simple and fast. The algorithm of Optimized LS-SVM had high degree of accuracy, and the error rate was only 8.250%. The 3D reconstruction could display the changes of the chiastopic fusion of different nerve functional fascicular groups directly. It could extract alone, merge and combine arbitrarily, and revolve at any angles. Furthermore, the reverse reduction on arbitrarily level dissection of the 3D model was very accurately. Conclusion Based on the t

关 键 词：周围神经 三维重建 功能束 关键技术 配准 

分 类 号：R651.3[医药卫生—外科学]