基于脑电信号分析建立飞行员慢性腰痛的分类模型研究

A classification model of chronic low back pain in pilots based on EEG signal analysis

目的结合飞行员慢性腰痛(chronic low back pain,CLBP)患者脑电信号和临床评估指标,建立慢性腰痛分类模型。方法研究纳入飞行员CLBP30例(研究组)、健康志愿者30人(对照组);64导脑电图机收集所有受试者脑电信号;通过测量受试者的外周神经电流感觉阈值(current perception threshold,CPT)收集外周电生理指标以评估其外周神经损伤程度;后者为测定受试者的主观疼痛程度和情感心理状态,主要参考健康调查简表(SF-36)的躯体疼痛(body pain,BP)和精神健康(mental health,MH)评分;应用SPSS 22.0软件,脑电信号预处理、功率谱分析使用MatLab软件,对不同类指标结合精确性分析,建立分类模型。结果2组BP、MH和CPT值组间比较差异有统计学意义(P<0.05)。2组仅有左前额β波、γ波差异有统计学意义(P<0.05)。外周电生理指标与异常脑波相关性分析显示,研究组BP值与左前额γ波之间存在相关性,MH值与左前额β波之间存在相关性;外周电生理指标使用SVM作为分类器,以γ波、CPT_5Hz作为特征,最高精确度为82.21%。结论通过脑电信号建立的分类模型可客观评估飞行员是否处于“慢性腰痛状态”,在临床诊断与治疗上具有一定的应用价值。

Objective To establish a classification model for CLBP(chronic low back pain) patients based on analysis of EEG signals of pilots with chronic pain caused by lumbar disc herniation,those of healthy volunteers and clinical evaluation indicators in order to supplement the clinical differential diagnosis of CLBP.Methods Thirty pilots with CLBP and 30 normal subjects were included in the study.The EEG signals of each subject were recorded by 64 channel electroencephalograph.Peripheral electrophysiological indexes were collected by measuring the subjects’ current perception threshold(CPT) to evaluate the degree of peripheral nerve injury.The subjective pain degree and emotional and psychological state of the subjects were measured based on the two scores of "body pain"(BP) and "mental health"(MH) in SF-36.Using SPSS 22.0 software,EEG signal preprocessing and power spectrum analysis with Matlab software,a classification model was established by combining the accuracy analysis of different indexes in the same category.Results There was significant difference in BP,MH and CPT values between CLBP patients and healthy subjects(P<0.05).Only β and γwaves in left prefrontal lobes were significantly different(P<0.05).The correlation analysis of peripheral electrophysiological indexes and abnormal brain waves showed that there was correlation between BP values and left frontal γ waves in the study group.MH values were correlated with left frontal β waves.SVM was used as a classifier for peripheral electrophysiological indexes.With γ waves and CPT_5 Hz as the feature,the highest accuracy was 82.21%.Conclusion With this classification model,we can find out whether pilots are in "chronic low back pain state",so this approach is applicable to clinical diagnosis and treatment.