唐文超，杨华元，刘堂义，高　明，徐　刚

（上海中医药大学针灸推拿学院，上海201203）

摘要：目的：通过自组织特征映射神经网络（SOM）分析由德国Simi Motion 3D三维运动图像解析系统导出的教师提插手法参数，获取手法分类及特征。方法： 研究选择30名针灸教师，取一侧“曲池穴”作为施术穴位，记录提插平补平泻、补法、泻法的运动视频。视频经由Simi Motion 3D三维运动图像解析系统分析，得出拇指指尖跟踪标记点位原始运动参数，通过SOM分析教师手法分类参数。结果： 针刺手法参数呈非正态分布，离散度较大；平补平泻与补法可分为4类，泻法可分为5类，各分类的特征集中程度一般。结论：针刺手法参数总体上表现为非正态分布的多样性特征，为典型“人为控制曲线”，表现出神经网络分类集中度较低的特点，其分类主要根据曲线外形及周期长短进行，且与手法难以程度相关。该技术可应用于各类手法的量化分析与技术传承研究，并能为针灸规范化与标准化提供参考。

关键词：针刺手法；运动视频；神经网络；分类研究

Research on Quantization and Classification of Acupuncture Manipulation "Lifting-Thrusting"  Based on Motion Video and Self-Organizing Feature Map Neural Network

TANG Wenchao, YANG Huayuan*, LIU Tangyi, GAO Ming, Xu Gang

(School of Acupuncture-Moxibustion and Tuina, Shanghai University of Traditional Chinese Medicine，Shanghai 201203，China)

Abstract: Objective:  The parameters of acupuncture manipulation "lifting-thrusting" derived from the Simi Motion 3D motion image analysis system was analyzed analysis by self-organizing feature mapping neural network (SOM) in order to obtain the classification and their characteristics.Methods: A total of 30 teachers majoring in acupuncture were selected for this study and the Quchi (LI-11) point on one side was selected as the operation point. Motion videos were record for 3 subtypes of "lifting-thrusting". The videos were analyzed using the Simi Motion 3D software with 2D coordinates and the original movement parameters of thumb tip tracking mark point are obtained to be analyzed by SOM.  Results:The parameters of acupuncture manipulation were non-normal distribution with large dispersion. The mild reinforcing-attenuating" and "reinforcing" methods were divided into 4 subtypes respectively, "attenuating" was divided into 5 subtypes. The characteristics of each classification were general. Conclusion：The parameters of acupuncture manipulation are diversity with non-normal distribution, which is a typical "artificial control curve" with low concentration of SOM classification. The classification is mainly based on the shape and period of the curve and related to difficulty degree of manipulation. This technique can be applied to quantitative analysis of various acupuncture manipulations and its inherited research, also provide reference for the normalization and standardization of acupuncture.

Key words: Acupuncture manipulation，Motion video， Neural network，Classification study