-
摘要: 研究大脑对不同气味的识别能力在嗅觉功能障碍评估和诊断等方面具有重要意义。本文提出将小波能量矩（WEM）作为嗅觉诱发脑电图（EEG）信号特征并用于气味分类。首先，通过试验采集 13 种气味的嗅觉诱发 EEG 数据；其次，从嗅觉诱发 EEG 数据中提取 WEM 作为信号特征，并将功率谱密度（PSD）、近似熵、样本熵及小波熵作为对比特征；最后，利用 k 近邻（k-NN）、支持向量机（SVM）、随机森林（RF）和决策树分类器识别不同的气味。结果表明，使用以上 4 种分类器，WEM 特征分类准确率均高于其它特征，其中 k-NN 分类器与 WEM 特征结合的分类准确率最高（91.07%）。本文进一步对不同 EEG 信号的频带进行了探究，发现大多数基于 γ 频带的分类准确率优于全频带及其他频带，其中 γ 频带 WEM 特征结合 k-NN 分类器的分类准确率最高（93.89%）。本文的研究结果一方面可为嗅觉功能评价提供新的客观依据，另一方面，也可为嗅觉诱发情绪的研究提供新的思路。Abstract: Studying the ability of the brain to recognize different odors is of great significance in the assessment and diagnosis of olfactory dysfunction. The wavelet energy moment (WEM) was proposed as a feature of olfactory electroencephalogram (EEG) signal and used for odor classification. Firstly, the olfactory evoked EEG data of 13 odors were collected by an experiment. Secondly, the WEM was extracted from olfactory evoked EEG data as the signal feature, and the power spectrum density (PSD), approximate entropy, sample entropy and wavelet entropy were used as the contrast features. Finally, k-nearest neighbor (k-NN), support vector machine (SVM), random forest (RF) and decision tree classifier were used to identify different odors. The results showed that using the above four classifiers, the classification accuracy of WEM feature was higher than other features, and the k-NN classifier combined with WEM feature had the highest classification accuracy (91.07%). This paper further explored the characteristics of different EEG frequency bands, and found that most of the classification accuracy based on the features of γ band was better than that of the full band and other bands, among which the WEM feature of the γ band combined with the k-NN classifier had the highest classification accuracy (93.89 %). The research results of this paper could provide a new objective basis for the evaluation of olfactory function. On the other hand, it could also provide new ideas for the study of olfactory-induced emotions.
-
Key words:
- olfactory /
- electroencephalogram /
- wavelet energy moment /
- pattern recognition
-
图 2 13 种气味分别进行 35 次测试的嗅觉诱发 EEG 信号试验流程图
Figure 2. Flowchart of olfactory EEG experiment: 35 tests for each of thirteen scents
图 3 采样频率为 128 Hz 的 EEG 信号 4 层小波分解
Figure 3. 4-layer wavelet decomposition of EEG signals with a sampling rate of 128 Hz
图 4 采用不同特征和不同分类器对 13 种气味的分类结果
Figure 4. Recognition results of 13 scents with different features and classifiers
表 1 不同频带的分类准确率（%）
Table 1. Classification accuracy of different frequency bands (%)
分类器 特征 全频带 θ α β γ k-NN WEM 91.07 36.17 68.83 66.69 93.89 PSD 83.76 33.28 36.15 58.55 89.07 近似熵 69.74 29.95 50.94 60.21 68.65 样本熵 65.16 26.78 48.15 57.54 66.08 小波熵 55.47 55.79 47.50 56.73 57.63 SVM WEM 84.98 21.91 77.85 77.48 93.52 PSD 79.02 20.51 25.04 64.43 92.61 近似熵 67.51 22.66 49.27 57.86 70.72 样本熵 64.71 24.34 46.67 57.76 68.90 小波熵 56.41 52.72 37.06 58.83 53.19 决策树 WEM 84.43 30.62 59.42 59.71 86.90 PSD 79.83 27.47 30.54 53.95 81.85 近似熵 57.17 24.38 42.79 47.44 56.07 样本熵 55.79 23.32 40.41 45.60 54.82 小波熵 47.94 46.22 36.43 42.95 43.28 RF WEM 88.38 35.31 65.76 66.43 91.06 PSD 85.83 31.60 35.51 62.41 86.55 近似熵 70.54 28.60 50.33 58.20 64.14 样本熵 67.29 28.24 47.80 55.67 61.97 小波熵 53.65 51.27 41.72 49.82 48.54 
下载: 导出CSV
-
[1] 廖凯. 嗅觉刺激对学习和记忆的影响及其机制的研究. 武汉: 华中科技大学, 2011. [2] Ehrlichman H, Bastone L. Olfaction and emotion//Serby M J, Chobor K L. Science of olfaction. New York: Springer, 1992: 410-438. [3] Skorić M K, Adamec I, Jerbić A B, et al. Electroencephalographic response to different odors in healthy individuals: a promising tool for objective assessment of olfactory disorders. Clinical EEG and Neuroscience, 2014, 46(4): 370-376. [4] Chen M C, Fang S H, Fang L. The effects of aromatherapy in relieving symptoms related to job stress among nurses. International Journal of Nursing Practice, 2015, 21(1): 87-93. doi: 10.1111/ijn.12229 [5] Nie D, Wang X W, Shi L C, et al. EEG-based emotion recognition during watching movies//International IEEE/EMBS Conference on Neural Engineering, Cancun: IEEE, 2011: 667-670. [6] Jatupaiboon N, Pan-Ngum S, Israsena P. Real-time EEG-based happiness detection system, The Scientific World Journal, 2013: 618-649. [7] Siuly S, Li Y. Discriminating the brain activities for brain–computer interface applications through the optimal allocation-based approach. Neural Computing and Applications, 2015, 26(4): 799-811. doi: 10.1007/s00521-014-1753-3 [8] Billot P E, Andrieu P, Biondi A, et al. Cerebral bases of emotion regulation toward odours: a first approach. Behavioural Brain Research, 2017, 317: 37-45. doi: 10.1016/j.bbr.2016.09.027 [9] Millot J L, Laurent L, Casini L. The influence of odors on time perception. Frontiers in Psychology, 2016, 7: 181. [10] Whitcroft K L, Aziz M, Croy I, et al. Short inter-stimulus intervals can be used for olfactory electroencephalography in patients of varying olfactory function. Neuroscience, 2017, 363: 26-33. doi: 10.1016/j.neuroscience.2017.08.046 [11] Joussain P, Thevenet M, Rouby C, et al. Effect of aging on hedonic appreciation of pleasant and unpleasant odors. PLoS One, 2013, 8(4): e61376. doi: 10.1371/journal.pone.0061376 [12] Frasnelli J, Christiane W, Thomas H. The influence of stimulus duration on odor perception. International Journal of Psychophysiology, 2006, 62(1): 24-29. doi: 10.1016/j.ijpsycho.2005.11.006 [13] Cherninskii A A, Zima I G, Makarchouk N Y, et al. Modifications of EEG related to directed perception and analysis of olfactory information in humans. Neurophysiology, 2009, 41(1): 63-70. doi: 10.1007/s11062-009-9078-z [14] Iacoviello D, Petracca A, Spezialetti M, et al. A classification algorithm for electroencephalography signals by self-Induced emotional stimuli. IEEE Trans Cybern, 2015, 46(12): 3171-3180. [15] Kroupi E, Vesin J M, Ebrahimi T. Subject-independent odor pleasantness classification using brain and peripheral signals. IEEE Transactions on Affective Computing, 2016, 7(4): 422-434. doi: 10.1109/TAFFC.2015.2496310 [16] Yavuz E, Aydemir O. Olfaction recognition by EEG analysis using wavelet transform features//International Symposium on Innovations in Intelligent Systems and Applications (INISTA), Sinaia: IEEE, 2016: 1-4. [17] Aydemir Ö. Classification of electroencephalography signals recorded during smelling of valerian and rosewater odors//International Conference on Telecommunications and Signal Processing (TSP), Sinaia: IEEE, 2016: 392-395. [18] Aydemir Ö. Olfactory recognition based on EEG Gamma-band Activity. Neural Comput, 2017, 29(6): 1667-1680. doi: 10.1162/NECO_a_00966 [19] 邵晨曦, 卢继军, 周颢. 基于小波变换的脑电图癫痫波形检测. 生物医学工程学杂志, 2002, 19(2): 259-263. doi: 10.3321/j.issn:1001-5515.2002.02.020 [20] 徐宝国, 宋爱国, 王爱民. 基于小波包能量的脑电信号特征提取方法. 东南大学学报:自然科学版, 2010, 40(6): 1203-1206. [21] 周卫东, 尹立新, 王照军. 脑电在线分析系统的研究及小波变换的应用. 生物医学工程学杂志, 2001, 18(2): 169-172. doi: 10.3321/j.issn:1001-5515.2001.02.001 [22] Yang H, Yu L. Feature extraction of wood-hole defects using wavelet-based ultrasonic testing. Journal of Forestry Research, 2017, 28(2): 395-402. doi: 10.1007/s11676-016-0297-z [23] Namazi H, Akrami A, Nazeri S, et al. Analysis of the influence of complexity and entropy of odorant on fractal dynamics and entropy of EEG signal, BioMed Research International, 2016: 1-5. [24] Pincus S M. Approximate entropy as a measure of system complexity. Proceedings of the National Academy of Sciences, 1991, 88(6): 2297-2301. doi: 10.1073/pnas.88.6.2297 [25] Richman J S, Moorman J R. Physiological time-series analysis, using approximate entropy and sample entropy. Am J Physiol Heart Circ Physiol, 2000, 278(6): H2039-H2049. doi: 10.1152/ajpheart.2000.278.6.H2039 [26] 李昕, 蔡二娟, 田彦秀, 等. 一种改进脑电特征提取算法及其在情感识别中的应用. 生物医学工程学杂志, 2017, 34(4): 510-517, 528. -
下载:
点击查看大图
图(4) / 表(1)
计量
- 文章访问数: 289
- HTML全文浏览量: 243
- PDF下载量: 0
- 被引次数: 0
京公网安备 11010502036328号 京ICP备17033152号