基于支持向量机的脊柱离体组织电阻抗识别  

Electrical impedance recognition of spine tissue in vitro based on support vector machine

作　　者：陈炳荣 施勇旺 李嘉浩 翟吉良[1] 刘梁[3] 刘文勇[4] 胡磊[5] 赵宇[1] CHEN Bingrong;SHI Yongwang;LI Jiahao;ZHAI Jiliang;LIU Liang;LIU Wenyong;HU Lei;ZHAO Yu(Department of Orthopaedic Surgery,Peking Union Medical College Hospital,Chinese Academy of Medical Sciences and Peking Union Medical College,Beijing 100730,China;Chinese Academy of Medical Sciences and Peking Union Medical College,Beijing 100730,China;China Astronaut Research and Training Center,Beijing 100094,China;Department of Biomedical Engineering,Beihang University,Beijing 100191,China;Department of Mechanical Engineering and Automation,Beihang University,Beijing 100191,China)

机构地区：[1]中国医学科学院北京协和医学院北京协和医院骨科,北京100730 [2]中国医学科学院北京协和医学院,北京100730 [3]中国航天员科研训练中心,北京100094 [4]北京航空航天大学生物与医学工程学院,北京100083 [5]北京航空航天大学机械工程及自动化学院,北京100083

出　　处：《山东大学学报（医学版）》2023年第3期57-62,114,共7页Journal of Shandong University:Health Sciences

基　　金：国家重点研发计划(2018YFB1307603);北京市自然科学基金-海淀原始创新联合基金(L192061)。

摘　　要：目的测量不同脊柱组织的电阻抗,基于支持向量机建立电阻抗数据的组织分类算法并验证算法的准确性,寻找不同组织电阻抗分类阈值。方法取离体脊柱组织,应用电化学分析仪采集10~100 kHz频率范围内皮质骨、松质骨、脊髓、肌肉、髓核的电阻抗。将两只猪采集的数据集分别作为训练集和测试集,应用主成分分析降维至二维数据,训练和验证基于支持向量机(SVM)建立的分类算法,应用集成学习的方法计算不同组织分类的电阻抗阈值。结果5种组织在10~100 kHz的测量频率内,电阻抗值差异有统计学意义(P<0.001)。应用主成分分析降维的数据集建立的支持向量机分类算法识别不同组织的准确率为100%。应用集成学习建立的多个分类器计算出了不同组织的电阻抗分类阈值。结论基于支持向量机可以实现脊柱术区组织电阻抗的准确识别,有望应用于临床协助医生提升组织识别准确率。Objective To measure impedances of different spinal tissues,to develop a tissue classification algorithm based on the impedance data using support vector machine(SVM),in order to verify the accuracy of the algorithm and to search for thresholds for impedance classification of different tissues.Methods Isolated spinal tissue was collected,and the electrical impedances of cortical bone,cancellous bone,spinal cord,muscle,and nucleus pulposus were collected using an electrochemical analyzer in the frequency range of 10-100 kHz.The data sets of two pigs were used as training set and test set,respectively.Principal component analysis(PCA)was used to reduce the dimensionality to two-dimensional data,and SVM based classification algorithm was trained and validated.Ensemble learning method was used to calculate the electrical impedance threshold for different tissue classifications.Results The electrical impedances of the five tissues were significantly different(P<0.001)within the measurement frequency of 10-100 kHz.The accuracy of PCA dimensionality reduction based SVM classification algorithm was 100%.Ensemble learning based multi-classifiers were used to compute impedance classification thresholds for different tissues.Conclusion Accurate identification of the electrical impedance of tissues in spinal surgery can be achieved based on support vector machines,which are expected to help surgeons improve the accuracy of tissue identification.

关 键 词：生物电阻抗 组织识别 支持向量机 主成分分析 集成学习 

分 类 号：R687.1[医药卫生—骨科学]