Ahmet Mert

Detrended fluctuation thresholding for empirical mode decomposition based denoising

Signal decompositions such as wavelet and Gabor transforms have successfully been applied in denoising problems. Empirical mode decomposition (EMD) is a recently proposed method to analyze non-linear and non-stationary time series and may be used for noise elimination. Similar to other decomposition based denoising approaches, EMD based denoising requires a reliable threshold to determine which oscillations called intrinsic mode functions (IMFs) are noise components or noise free signal components. Here, we propose a metric based on detrended fluctuation analysis (DFA) to define a robust threshold. The scaling exponent of DFA is an indicator of statistical self-affinity. In our study, it is used to determine a threshold region to eliminate the noisy IMFs. The proposed DFA threshold and denoising by DFA-EMD are tested on different synthetic and real signals at various signal to noise ratios (SNR). The results are promising especially at 0 dB when signal is corrupted by white Gaussian noise (WGN). The proposed method outperforms soft and hard wavelet threshold method.

Breast Cancer Detection with Reduced Feature Set

This paper explores feature reduction properties of independent component analysis (ICA) on breast cancer decision support system. Wisconsin diagnostic breast cancer (WDBC) dataset is reduced to one-dimensional feature vector computing an independent component (IC). The original data with 30 features and reduced one feature (IC) are used to evaluate diagnostic accuracy of the classifiers such as k-nearest neighbor (k-NN), artificial neural network (ANN), radial basis function neural network (RBFNN), and support vector machine (SVM). The comparison of the proposed classification using the IC with original feature set is also tested on different validation (5/10-fold cross-validations) and partitioning (20%–40%) methods. These classifiers are evaluated how to effectively categorize tumors as benign and malignant in terms of specificity, sensitivity, accuracy, F-score, Youdens index, discriminant power, and the receiver operating characteristic (ROC) curve with its criterion values including area under curve (AUC) and 95% confidential interval (CI). This represents an improvement in diagnostic decision support system, while reducing computational complexity.

Random subspace method with class separability weighting

The random subspace method RSM is one of the ensemble learning algorithms widely used in pattern classification applications. RSM has the advantages of small error rate and improved noise insensitivity due to ensemble construction of the base-learners. However, randomness may cause a reduction of the final ensemble decision performance because of contributions of classifiers trained by subsets with low class separability. In this study, we present a new and improved version of the RSM by introducing a weighting factor into the combination phase. One of the class separability criteria, J3, is used as a weighting factor to improve the classification performance and eliminate the drawbacks of the standard RSM algorithm. The randomly selected subsets are quantified by computing their J3 measure to determine voting weights in the model combination phase, assigning lower voting weight to classifiers trained by subsets with poor class separability. Two models are presented including J3-weighted RSM and optimized J3 weighted RSM. In J3 weighted RSM, computed weighting values are directly multiplied by class assignment posteriors, whereas in optimized J3 weighted RSM, computed weighting values are optimized by a pattern search algorithm before multiplying by posteriors. Both models are shown to provide better error rates at lower subset dimensionality.

EOG denoising using Empirical Mode Decomposition and Detrended Fluctuation Analysis

In this study, a method is presented for the removal of electrooculogram (EOG) noise from electroencephalography (EEG) recordings by using recently proposed data driven approach called Empirical Mode Decomposition (EMD). The EMD represents the signal as a combination of Intrinsic Mode Functions (IMFs). It is an important problem to determine which IMFs belong to signal and noise in multi-component or noisy signals. Detrended Fluctuation Analysis (DFA) is a successful method to characterize non-stationary signals. In our approach, a threshold is determined from the DFA, and used to select the noise IMFs. Performance of the proposed method is demonstrated by means of computer simulations using noisy EEG signals.

ECG signal analysis based on variational mode decomposition and bandwidth property

In this paper, the bandwidth properties of the modes obtained using the variational mode decomposition (VMD) are analyzed to detect arrhythmia electrocardiogram (ECG) beats. The VMD is an enhanced version of the empirical mode decomposition (EMD) algorithm to analyze non-linear and nonstationary signals. It decomposes the signal into a set of bandlimited amplitude and frequency modulated oscillations called modes. ECG signals from MIT-BIH arrhythmia database are decomposed using the VMD, and the amplitude modulation bandwidth BAM, frequency modulation bandwidth BFM and total bandwidth B of the modes are deployed as feature vector. Heart beats such as normal (N), premature ventricular contraction (PVC), left bundle branch block (LBBB), right bundle branch block (RBBB), paced beat( PB) and atrial premature beat (APB) are classified using these features. Class discrimination capability of the VMD based features are indicated giving different instantaneous frequency (IF) and amplitude (IA) spectra. Finally, single classifiers such as k-nearest neighbor, artificial neural network and decision tree with their ensemble methods are used to evaluate the performance of the proposed method.

Epilepsy detection using Empirical Mode Decomposition and detrended Fluctuation Analysis

In this study, a new method is presented to analyze electroencephalography (EEG) signals by deploying recently proposed adaptive and data driven signal processing method called Empirical Mode Decomposition (EMD). The EMD algorithm represents a signal as a combination of Intrinsic Mode Functions (IMFs) which are extracted from the signal. It is possible to analyze each component of a multi-component signal by using the IMFs. Thus, detrended Fluctuation Analysis (DFA) which is suggested to characterize the auto-correlation properties of non-stationary signals. Frequency and time-frequency domain methods are successfully employed to analyze EEG signals during epileptic seizure. In this study, however, we present a time domain method to analyze and classify EEG signals by investigating the auto-correlation properties of their IMFs extracted by EMD. In the proposed method the IMF features are analyzed by using DFA to determine the epileptic EEG signals.

Analysis of EEG signals by emprical mode decomposition and mutual information

Empirical mode decomposition has been recently proposed to analyze non-stationary signals. It decomposes the signal into intrinsic mode functions (IMF) which are derived from the signal itself. However, it is still an unknown issue which IMF involves more information of the signal. In this study, single channel EEG signals from normal and epileptic recordings are analyzed. Hence, mutual information is computed between the autocorrelation function (ACF) of a reference and a given EEGs first IMF. The proposed method is applied to two different datasets to show its classification capability of normal and epileptic EEG signals.