基于脑电信号的疼痛强度识别方法研究
|
摘要
目的:通过对疼痛患者的脑电信号进行特征提取和特征选择,实现对疼痛等级的量化评估。方法:对临床采集的脑电信号进行离散小波变换得到近似和细节系数,根据每层分解系数计算子带能量占比、系数统计特征、样本熵和锁相值,组成特征向量。利用随机森林进行特征选择和疼痛预测。结果:实现对疼痛等级的三分类,平均分类准确率为91.7%,其中无痛和重痛的分类准确率达100%。结论:本研究方法可以有效地对脑电信号进行特征提取和选择,以较高的准确率实现疼痛强度的识别,为临床疼痛的客观评估奠定基础。
Objective To perform feature extraction and feature selection for electroencephalogram(EEG) signals collected from patients with postherpetic neuralgia for quantitatively evaluating the level of pain. Methods Discrete wavelet transform was employed to decompose clinically collected EEG signals to obtain approximate and detail coefficients. The feature vectors were composed of sub-band energy ratio, coefficient statistics, sample entropy and phase-locked value which were calculated based on the decomposition coefficients of each level. Random forest was used for feature selection and pain intensity recognition.Results The proposed method realized the 3 classifications of pain levels, with an average classification accuracy of 91.7%.Moreover, the accuracy of the classification between no-pain and high-pain reached 100%. Conclusion The proposed method can be used to effectively extract and select features from EEG signals, and realize pain intensity recognition with a high accuracy,which lays a foundation for the objective evaluation of clinical pain.
Objective To perform feature extraction and feature selection for electroencephalogram(EEG) signals collected from patients with postherpetic neuralgia for quantitatively evaluating the level of pain. Methods Discrete wavelet transform was employed to decompose clinically collected EEG signals to obtain approximate and detail coefficients. The feature vectors were composed of sub-band energy ratio, coefficient statistics, sample entropy and phase-locked value which were calculated based on the decomposition coefficients of each level. Random forest was used for feature selection and pain intensity recognition.Results The proposed method realized the 3 classifications of pain levels, with an average classification accuracy of 91.7%.Moreover, the accuracy of the classification between no-pain and high-pain reached 100%. Conclusion The proposed method can be used to effectively extract and select features from EEG signals, and realize pain intensity recognition with a high accuracy,which lays a foundation for the objective evaluation of clinical pain.
引文
[1] KUO P C, CHEN Y T, CHEN Y S, et al. Decoding the perception of endogenous pain from resting-state MEG[J]. Neuroimage, 2016, 144(Pt A):1-11.
[2] OCAK H. Automatic detection of epileptic seizures in EEG using discrete wavelet transform and approximate entropy[J]. Expert Syst Appl, 2009, 36(2):2027-2036.
[3] GRUSS S, TREISTER R, WERNER P, et al. Pain intensity recognition rates via biopotential feature patterns with support vector machines[J].PLoS One, 2015, 10(10):e0140330.
[4] VRIES M D, WILDERSMITH O H, JONGSMA M L, et al. Altered resting state EEG in chronic pancreatitis patients:toward a marker for chronic pain[J]. J Pain Res, 2013, 25(6):815-824.
[5] JENSEN M P, SHERLIN L H, GERTZ K J, et al. Brain EEG activity correlates of chronic pain in persons with spinal cord injury:clinical implications[J]. Spinal Cord, 2013, 51(1):55-58.
[6] GRAVERSEN C, OLESEN S S, OLESEN A E, et al. The analgesic effect of pregabalin in patients with chronic pain is reflected by changes in pharmaco-EEG spectral indices[J]. Br J Clin Pharmacol,2012, 73(3):363-372.
[7] HADJILEONTIADIS L. Eeg-based tonic cold pain characterization using wavelet higher order spectral features[J]. IEEE Trans Biomed Eng, 2015, 62(8):1981-1991.
[8] ALAZRAI R, MOMANI M, KHUDAIR H A, et al. EEG-based tonic cold pain recognition system using wavelet transform[J]. Neural Comput Appl, 2017, 4:1-14.
[9] VIJAYAKUMAR V, CASE M, SHIRINPOUR S, et al. Quantifying and characterizing tonic thermal pain across subjects from EEG data using random forest models[J]. IEEE Trans Biomedical Eng, 2017, 64(12):2988-2996.
[10]GRAVERSEN C, FROKJAER J B, BROCK C, et al. Support vector regression correlates single-sweep evoked brain potentials to gastrointestinal symptoms in diabetes mellitus patients[C]//Engineering in Medicine and Biology Society. IEEE, 2012:5242-5245.
[11]VATANKHAH M, ASADPOUR V, FAZEL-REZAI R. Perceptual pain classification using ANFIS adapted RBF kernel support vector machine for therapeutic usage[J]. Appl Soft Comput, 2013, 13(5):2537-2546.
[12]BOULESTEIX A L, JANITZA S, KRUPPA J. Overview of random forest methodology and practical guidance with emphasis on computational biology and bioinformatics[J]. Wires Data Min Knowl,2012, 2(6):493-507.
[13]PAJA W, WRZESIEN M. Melanoma important features selection using random forest approach[C]//The International Conference on Human System Interaction. 2013:415-418.
[14]ORHAN U, HEKIM M, OZER M. EEG signals classification using the k-means clustering and a multilayer perceptron neural network model[J]. Expert Syst Appl, 2011, 38(10):13475-13481.
[15]DAUBECHIES I. Ten lectures on wavelets[M]. Philadelphia:Society for Industrial and Applied Mathematics, 1992:1-357.
[16]SARNTHEIN J, STERN J, AUFENBERG C, et al. Increased EEG power and slowed dominant frequency in patients with neurogenic pain[J]. Brain, 2006, 129(Pt 1):55-64.
[17]AMIN H U, MALIK A S, AHMAD R F, et al. Feature extraction and classification for EEG signals using wavelet transform and machine learning techniques[J]. Australas Phys Eng Sci Med, 2015, 38(1):1-11.
[18]CALDERO-BARDAJI P, LONGFEI X, JASCHKE S, et al. Detection of steering direction using EEG recordings based on sample entropy and time-frequency analysis[J]. Conf Proc IEEE Eng Med Biol Soc,2016, 38:833-836.
[19]JIANG G J, FAN S Z, ABBOD M F, et al. Sample entropy analysis of EEG signals via artificial neural networks to model patients'consciousness level based on anesthesiologists experience[J]. Biomed Res Int, 2015(3):343478.
[20]ZHANG J, WANG N, KUANG H, et al. An improved method to calculate phase locking value based on Hilbert-Huang transform and its application[J]. Neural Comput Appl, 2014, 24(1):125-132.
[21]GHARSALLI S, EMILE B, LAURENT H, et al. Random forest-based feature selection for emotion recognition[C]//International Conference on Image Processing Theory, TOOLS and Applications. IEEE, 2016:268-272.
[22]姚登举,杨静,詹晓娟.基于随机森林的特征选择算法[J].吉林大学学报(工学版), 2014, 44(1):137-144.YAO D J, YANG J, ZHAN X J. Feature selection algorithm based on random forest[J]. Journal of Jilin University(Engineering and Technology Edition), 2014, 44(1):137-144.
[23]WONG T T. Performance evaluation of classification algorithms by k-fold and leave-one-out cross validation[J]. Pattern Recognit, 2015,48(9):2839-2846.
[2] OCAK H. Automatic detection of epileptic seizures in EEG using discrete wavelet transform and approximate entropy[J]. Expert Syst Appl, 2009, 36(2):2027-2036.
[3] GRUSS S, TREISTER R, WERNER P, et al. Pain intensity recognition rates via biopotential feature patterns with support vector machines[J].PLoS One, 2015, 10(10):e0140330.
[4] VRIES M D, WILDERSMITH O H, JONGSMA M L, et al. Altered resting state EEG in chronic pancreatitis patients:toward a marker for chronic pain[J]. J Pain Res, 2013, 25(6):815-824.
[5] JENSEN M P, SHERLIN L H, GERTZ K J, et al. Brain EEG activity correlates of chronic pain in persons with spinal cord injury:clinical implications[J]. Spinal Cord, 2013, 51(1):55-58.
[6] GRAVERSEN C, OLESEN S S, OLESEN A E, et al. The analgesic effect of pregabalin in patients with chronic pain is reflected by changes in pharmaco-EEG spectral indices[J]. Br J Clin Pharmacol,2012, 73(3):363-372.
[7] HADJILEONTIADIS L. Eeg-based tonic cold pain characterization using wavelet higher order spectral features[J]. IEEE Trans Biomed Eng, 2015, 62(8):1981-1991.
[8] ALAZRAI R, MOMANI M, KHUDAIR H A, et al. EEG-based tonic cold pain recognition system using wavelet transform[J]. Neural Comput Appl, 2017, 4:1-14.
[9] VIJAYAKUMAR V, CASE M, SHIRINPOUR S, et al. Quantifying and characterizing tonic thermal pain across subjects from EEG data using random forest models[J]. IEEE Trans Biomedical Eng, 2017, 64(12):2988-2996.
[10]GRAVERSEN C, FROKJAER J B, BROCK C, et al. Support vector regression correlates single-sweep evoked brain potentials to gastrointestinal symptoms in diabetes mellitus patients[C]//Engineering in Medicine and Biology Society. IEEE, 2012:5242-5245.
[11]VATANKHAH M, ASADPOUR V, FAZEL-REZAI R. Perceptual pain classification using ANFIS adapted RBF kernel support vector machine for therapeutic usage[J]. Appl Soft Comput, 2013, 13(5):2537-2546.
[12]BOULESTEIX A L, JANITZA S, KRUPPA J. Overview of random forest methodology and practical guidance with emphasis on computational biology and bioinformatics[J]. Wires Data Min Knowl,2012, 2(6):493-507.
[13]PAJA W, WRZESIEN M. Melanoma important features selection using random forest approach[C]//The International Conference on Human System Interaction. 2013:415-418.
[14]ORHAN U, HEKIM M, OZER M. EEG signals classification using the k-means clustering and a multilayer perceptron neural network model[J]. Expert Syst Appl, 2011, 38(10):13475-13481.
[15]DAUBECHIES I. Ten lectures on wavelets[M]. Philadelphia:Society for Industrial and Applied Mathematics, 1992:1-357.
[16]SARNTHEIN J, STERN J, AUFENBERG C, et al. Increased EEG power and slowed dominant frequency in patients with neurogenic pain[J]. Brain, 2006, 129(Pt 1):55-64.
[17]AMIN H U, MALIK A S, AHMAD R F, et al. Feature extraction and classification for EEG signals using wavelet transform and machine learning techniques[J]. Australas Phys Eng Sci Med, 2015, 38(1):1-11.
[18]CALDERO-BARDAJI P, LONGFEI X, JASCHKE S, et al. Detection of steering direction using EEG recordings based on sample entropy and time-frequency analysis[J]. Conf Proc IEEE Eng Med Biol Soc,2016, 38:833-836.
[19]JIANG G J, FAN S Z, ABBOD M F, et al. Sample entropy analysis of EEG signals via artificial neural networks to model patients'consciousness level based on anesthesiologists experience[J]. Biomed Res Int, 2015(3):343478.
[20]ZHANG J, WANG N, KUANG H, et al. An improved method to calculate phase locking value based on Hilbert-Huang transform and its application[J]. Neural Comput Appl, 2014, 24(1):125-132.
[21]GHARSALLI S, EMILE B, LAURENT H, et al. Random forest-based feature selection for emotion recognition[C]//International Conference on Image Processing Theory, TOOLS and Applications. IEEE, 2016:268-272.
[22]姚登举,杨静,詹晓娟.基于随机森林的特征选择算法[J].吉林大学学报(工学版), 2014, 44(1):137-144.YAO D J, YANG J, ZHAN X J. Feature selection algorithm based on random forest[J]. Journal of Jilin University(Engineering and Technology Edition), 2014, 44(1):137-144.
[23]WONG T T. Performance evaluation of classification algorithms by k-fold and leave-one-out cross validation[J]. Pattern Recognit, 2015,48(9):2839-2846.