【摘要】 为对肌电图波形特征作多维分析和定量判断,我们采用图形识别法首次对60例氟骨症患者肌电波形的五种特征数值进行了采集,将它们表示为多维矢量。按Fisher准则计算出线性鉴别函数并对新采集的肌电信号作分类计算,提出了三个分类鉴别方程式,此外,还计算了正确分类概率的估值。运用图形识别法判断为肌电异常者来自正常组的概率仅为0.8％。我们还用此法对氟宁的疗效作了定量的评价。更多还原

【Abstract】 Usually, we recognize and judge an electromyogram signals by observating with eyes their waveform characteristies. In an analysis process by human brain the characteriste values of waveform were less obtained and a quantitative analysis can not be made. A pattern recorgnition method may satisfy the mentioned above requests. Five values of waveform in each of 60 cases of bone-flurosis have been gethered and indicated as polydimensional factors. Their line-discriminant functions have been calculated by Fisher rule and a new obtained EMG’s signal was classified and the probablity of procisc classification was counted as well. The probablity judged as anomalistic EMG which might be from normal one was 0.8% with the pat tern recognition method. We have also evaluated in quantity the therapeutic effects of Fu-ning by this method.更多还原