A study on feature selection using multi-domain feature extraction for automated k-complex detection

Abstract

Background: K-complex detection plays a significant role in the field of sleep research. However, manual annotation for electroencephalography (EEG) recordings by visual inspection from experts is time-consuming and subjective. Therefore, there is a necessity to implement automatic detection methods based on classical machine learning algorithms. However, due to the complexity of EEG signal, current feature extraction methods always produce low relevance to k-complex detection, which leads to a great performance loss for the detection. Hence, finding compact yet effective integrated feature vectors becomes a crucially core task in k-complex detection.

Method: In this paper, we first extract multi-domain features based on time, spectral analysis, and chaotic theory. Those features are extracted from a 0.5-s EEG segment, which is obtained using the sliding window technique. As a result, a vector containing twenty-two features is obtained to represent each segment. Next, we explore several feature selection methods and compare their performance in detecting k-complex. Based on the analysis of the selected features, we identify compact features which are fewer than twenty-two features and deemed as relevant and proceeded to the next step. Additionally, three classical classifiers are employed to evaluate the performance of the feature selection models.

Results: The results demonstrate that combining different features significantly improved the k-complex detection performance. The best performance is achieved by applying the feature selection method, which results in an accuracy of 93.03%$\pm$7.34, sensitivity of 93.81%$\pm$5.62%, and specificity 94.13$\pm$5.81, respectively, using a smaller number of the combined feature sets.

Conclusion: The proposed method in this study can serve as an efficient tool for the automatic detection of k-complex, which is useful for neurologists or doctors in the diagnosis of sleep research.

Keywords: detection; electroencephalography (EEG); feature selection; k-complex; multi-domain features.

Figure 1

The k-complex detection flowchart of the proposed methods.

Figure 2

An filtered EEG signal [(A) is EEG signals with k-complex, and (B) represents EEG signals without k-complex].

Figure 3

The scheme of ReliefF. The dimension of feature subsets is reduced based feature selected of ReliefF, and the selected features are used for further analysis.

Figure 4

The scheme of CFS. The dimension of feature subsets is reduced based feature selected of CFS.

Figure 5

The scheme of SFS. The dimension of feature subsets is reduced based feature selected of SFS.

Figure 6

The scheme of INTERACT method. The dimension of feature subsets is reduced based feature selected of INTERACT.

Figure 7

Correlation coefficient and significance test for the different domain features with features of k-complex and non-k-complex (if the value of p below than 0.005, it is marked with **, and if the p-value is below than 0.05, it is marked with *. (A) is for spectral feature, and (B) is for chaotic features).

Figure 8

Comparison of J1 value between different features and subjects for k-complex detection.

Figure 9

The accuracy of multi-domain features based on decision trees model for each features. Each box represents the 25–75th percentiles, and central line is the median value, the tiny vertical lines extend to the most extreme data not considering as outliers, which are plotted individually.

Figure 10

Comparison of feature selection methods and without feature selection method using three detection algorithm (A) is for LDA algorithm, (B) is for LSVM algorithm, and (C) is for DT algorithm. No selected means that without any feature selection process.