基于形态学滤波器与RF模型的棘波检测仿真实验

Simulation Experiment for Spike Detection Based on Morphological Filter and RF Model

针对癫痫棘波检测方法中棘波表现形式多样化、人工识别工作量大等问题,提出采用基于形态学滤波器和随机森林(Random Forest, RF)模型的癫痫棘波检测算法以Matlab软件为基础开发平台,利用形态学滤波技术对单通道双极导联脑电信号进行棘波检测,根据阈值判断法消除假阳性棘波,获得候选棘波,然后提取候选棘波的波形特征,训练RF分类模型,实现癫痫棘波的自动检测。实验结果表明,该实验所用的癫痫棘波检测方法在测试数据集上取得了较好的分类效果。该仿真实验可用于数字信号处理及Matlab仿真实验教学,增强学生应用编程仿真工具解决实际问题的能力。

In view of the problems of diversified forms of spike waves in epileptic spike detection methods and heavy manual identification workload,anautomatic epileptic spike detection methodbased on morphological filter and random forest(RF)model is proposed.This proposed algorithm is developed on the platform of Matlab software.Itadoptsmorphological filtering technology to perform spike detection of bipolar channel EEG signals,eliminatesfalse positive spikes with the threshold judgment method to obtain candidate spikes.And then spike waveform features are extracted from candidate spikes to train RF classification model,and automatic detection of epileptic spikesis realized.Experimental results show that the epileptic spike detection method used in the experiment achieves good results on the testing dataset.The simulation experiment can be used in the digital signal processing and Matlab simulation experiment teaching,which improves students’ability to use programming simulation tools to solve practical problems.