论文标题:近红外光谱分析技术鉴别周围神经运动束和感觉束的实验研究
The Preliminary Research of Identifying Motor and Sensory Fiber by Near-infrared Spectroscopy
论文作者
论文导师 曹晓建,论文学位 博士,论文专业 骨外科学
论文单位 南京医科大学,点击次数 96,论文页数 109页File Size7483K
2007-05-08论文网 http://www.lw23.com/lunwen_625744227/
Near-infrared spectroscopy;; Cluster analysis;; SIMCA;; Spinal nerve root;; Facial nerve;; Cutaneous nerve
目的：初步探讨近红外光谱分析法在鉴别周围神经运动束和感觉束的应用的可行性，以建立一种快速鉴别周围神经功能束性质的方法。 方法：将Beagle犬分次麻醉后处死，立即从后背正中将椎管打开，取从硬膜外到脊神经根管入口这一段走行的L_1-S_2脊神经的前根和后根，；将马尾切取；分别经颞部、外耳道前、下颌下切开皮肤、皮下组织，充分显露腮腺及其上、前、后缘，显露、解剖、切取面神经的颞支、颧支、颊支、下颌缘支；从胸所乳突肌后缘切开皮肤，显露、解剖从该肌后缘中份穿出的颈丛皮神经，切取皮神经。将标本分类装入标本瓶中，迅速放在液氮里保存备用；取用时在室温下自然解冻，解冻后保存在湿盒里，防止标本脱水干燥。应用MPA傅立叶变换近红外光谱仪采集光谱。取备好的脊神经根，每一根切取长2mm，使用仪器自带的光纤探头，置于光纤探头顶部或透射单元测量。每个样品每次采样为连续扫描64次，每一次约二十秒得光谱图，重复3次，求平均光谱。取Beagle犬的脊神经126根前根和126根后根(前根为运动神经，后根为感觉神经)，利用近红外漫反射方式进行快速采样，应用SIMCA(soft independent modeling of class analogy)法将前根和后根鉴别分类；分别取2条犬的12根脊神经的前根及14根脊神经的后根行漫透射和漫反射方式近红外光谱采样，运用模式识别法中的聚类分析法进行定性分类鉴别，比较两种采样方式；取4条犬的马尾，分离、切取马尾中未分类的40根前根和后根，用近红外漫反射方式进行快速采集光谱，运用模式识别法中的聚类分析法进行定性分类鉴别，这相同的40根脊神经的另一部分，进行切片，用Fuminori酶组织化学染色法染色并和进红外光谱法比较；取4条犬的20根面神经和20根颈丛皮神经，近红外漫反射采集光谱，运用聚类分析法分类鉴别。 结果：单纯从脊神经(或马尾)的前根和后根、面神经和颈丛皮神经的近红外光谱原图、一阶或二阶导数光谱图观察，不能直观鉴别分类前根和后根、面神经和颈丛皮神经；经聚类分析或SIMCA可以将脊神经的前根和后根归为两类，鉴别准确率分别是83.7％和84.6％；经聚类分析，面神经和颈丛皮神经归为两类，鉴别准确率为90.0％；马尾的前根和后根的鉴别准确率为87.5％。在时间上，Fuminori酶组织化学染色法染色2小时后鉴别前根和后根的准确率为100％，近红外光谱分析法鉴别感觉神经束和运动束只需约2分钟。 结论：近红外光谱分析技术有可能成为一种快速、原位、准确鉴别周围神经功能束性质的方法。
Objective The aim of this study was to determine the feasibility of usingnear-infrared spectroscopy (NIRS) to identify motor and sensory fascicles,providing a rapid method for identifying motor and sensory fiber. Methods 12 adult Beagle dogs were anesthetized generally and werekilled. From all 12 of the dogs, canalis vertebralis were opened, and bothanterior and posterior roots were harvested from L_1-S_2. 126 anteriorroots and 126 posterior roots were collected, cauda equines, 20 facialnerves and 20 cutaneous nerves of neck were harvested from the Beagledogs. These nerves were kept in the liquid nitrogen. When to use, thenerves deforest under the room temperature, and then were kept in moistcase to prevent the specimen drying and dehydrating. NIR spectra weremeasured using a MPA Fourier transform near infrared (FT-NIR)spectrometer. Each sample was put directly onto the top of the optic fiberprobe or in the external transmission of the FT-NIR. Each spectrum represented the average of 64 scans at a resolution of 8cm~(-1). Duplicates ofeach sample were scanned three times. The average spectrum of eachsample was used for further analysis. Second derivative spectra were usedto reduce baseline variations observed in original diffuse reflectancespectra as well as to enhance spectral features. To develop a calibrationmodel, soft independent modeling of class analogy (SIMCA) was used toclassify each class according to its analogy to the training samples.Cluster analysis of the pattern recognition was adopted. FuminoriEnzyme-Staining method was compared with NIRS. Results The results showed that the anterior and posterior roots havequite similar first-derivative and second-derivative spectra of thenear-infrared spectroscopy and unable to be directly distinguished. Therewere no clear differences from the spectra between the anterior andposterior roots of samples. The facial nerves and the cutaneous nerves ofneck were also. Different spinal nerve roots of the Beagles could bedirectly distinguished by clustering analysis or SIMCA. The facial nervesand the cutaneous nerves of neck were could be directly distinguished byclustering analysis. Original spectral data were transformed to secondderivate spectral data. SIMCA was employed to identify the anterior andposterior roots based upon differences in their spectral features. SIMCAModels correctly classified 79.2-100.0% for the anterior and posteriorroots of spinal nerves, and mean 83.7%. Clustering analysis correctly identified 87.5% for the anterior and posterior roots of the cauda equines,and 90.0% for the facial nerves and the cutaneous nerves of neck, and84.6% for the anterior and posterior roots of spinal nerves. The exact rateof 2 hour staining result of Fuminori Enzyme-Staining Techniques was100%, NIRS Techniques only needs 2 minutes to identify the property ofnerve fiber and have more superiority in clinical application. Conclusion This study demonstrated that NIR spectroscopy incombination with multivariate data analysis methods(SIMCA)/clusteringanalysis could be used to classify the sensory and motor nerves. Thisresult suggests that NIR spectroscopy may provide a rapid, correct,non-destructive, low-cost means to quickly differentiate motor andsensory fascicles in mixed nerves.
|
