首页> 外国专利> METHOD FOR DETECTING FOCAL EPILEPTIFORM DISCHARGES IN A PROLONGED EEG RECORDING

METHOD FOR DETECTING FOCAL EPILEPTIFORM DISCHARGES IN A PROLONGED EEG RECORDING

机译:检测延长脑电图录制局灶性癫痫大均放电的方法

摘要

The invention relates to medicine and computing. Claimed is a computer-implemented method for automatically detecting focal epileptiform discharges in a prolonged EEG recording, comprising a preparatory step that involves: preprocessing at least one EEG signal on a computer to obtain input data, wherein at least one signal segment having a length of 1 second is isolated in said at least one EEG signal; marking the at least one segment isolated in the preceding step in accordance with an existing expert markup taken from a database; training a neural network by simultaneously feeding to the input thereof at least one pair of 1 second EEG signal segments that pertains to symmetrical channels, and a label corresponding to the first of the two symmetrical channels; carrying out cross-validation to assess classification quality; and selecting a probability cutoff threshold for predicted focal discharges. The claimed method further comprises a working step that involves: preprocessing at least one EEG signal on a computer, wherein at least one signal segment having a length of 1 second is isolated in said at least one EEG signal; launching the neural network trained in the preceding step, and feeding to the input thereof at least one pair of 1 second EEG signal segments that pertains to symmetrical channels; obtaining at the output an assessment of the probability that the first of the two symmetrical signal segments belongs to the class of epileptiform signals; applying a cutoff to the obtained probabilities in accordance with the predetermined threshold, and generating a list of predicted focal epileptiform discharges; carrying out a localization search for the source of the discharges by automatically applying the predicted focal discharges to a map taking into account the standard placement of electrodes, and performing interpolation to determine the distribution of focal epileptiform discharges. The technical result consists in reducing the time taken to detect focal epileptiform discharges.
机译:本发明涉及医学和计算。要求保护的计算机实现的方法,用于在延长的EEG记录中自动检测焦点癫痫型放电,包括涉及的准备步骤:预处理计算机上的至少一个EEG信号以获得输入数据,其中至少一个具有长度的信号段在所述至少一个EEG信号中被隔离1秒;根据从数据库中取出的现有专家标记,标记前一步中孤立的至少一个段;通过同时馈送到其输入来训练神经网络,至少一对涉及对称通道的1个第二EEG信号段,以及对应于两个对称通道中的第一的标签;进行交叉验证以评估分类质量;并选择预测焦点放电的概率截止阈值。要求保护的方法还包括涉及的工作步骤:预处理计算机上的至少一个EEG信号,其中至少一个长度为1秒的信号段在所述至少一个EEG信号中被隔离;启动在前面的步骤中训练的神经网络,并馈送到其输入,至少一对涉及对称通道的1个第二EEG信号段;获得输出的评估是,两个对称信号段中的第一个属于癫痫型信号的概率;根据预定阈值对所获得的概率应用截止,并产生预测的焦点癫痫大小排出的列表;通过自动将预测的焦点放电自动应用于地图来考虑电极的标准放置,并执行插值以确定焦点癫痫件放电的分布来对放电来源进行本地化搜索。技术结果包括减少检测局灶性癫痫株排出所需的时间。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号