S. Dandapat

Wavelet energy based diagnostic distortion measure for ECG

Abstract In this paper, a novel Wavelet Energy based diagnostic distortion (WEDD) measure is proposed to assess the reconstructed signal quality for ECG compression algorithms. WEDD is evaluated from the Wavelet coefficients of the original and the reconstructed ECG signals. For each ECG segment, a Wavelet energy weight vector is computed via five-level biorthogonal discrete Wavelet transform (DWT). WEDD is defined as the sum of Wavelet energy weighted percentage root mean square difference of each subband. The effectiveness of this measure is validated by linear (linear polynomial and cubic polynomial) and nonlinear (logistic) regression analysis between the computed WEDD values and the mean opinion score (MOS) given by cardiologists. WEDD provides a better prediction accuracy and exhibits a statistically better monotonic relationship with the MOS ratings than Wavelet based weighted percentage root mean square difference (PRD) measure (WWPRD), PRD and other objective measures. Standard correlation coefficient and Spearman rank-order correlation coefficient (SROCC) between the WEDD/MOS ratings is 0.969 and 0.9624, respectively.

ECG signal denoising using higher order statistics in Wavelet subbands

Abstract In this work, we propose a novel denoising method based on evaluation of higher-order statistics at different Wavelet bands for an electrocardiogram (ECG) signal. Higher-order statistics at different Wavelet bands provides significant information about the statistical nature of the data in time and frequency. The fourth order cumulant, Kurtosis , and the Energy Contribution Efficiency (ECE) of signal in a Wavelet subband are combined to assess the noise content in the signal. Accordingly, four denoising factors are proposed. Performance of the denoising factors is evaluated and compared with the soft thresholding method. The filtered signal quality is assessed using Percentage Root Mean Square Difference (PRD), Wavelet Weighted Percentage Root Mean Square Difference (WWPRD), and Wavelet Energy-based Diagnostic Distortion (WEDD) measures. It is observed that the proposed denoising scheme not only filters the signal effectively but also helps retain the diagnostic information.

Wavelet threshold based ECG compression using USZZQ and Huffman coding of DSM

Abstract In this paper, a new Wavelet threshold based ECG signal compression technique using uniform scalar zero zone quantizer (USZZQ) and Huffman coding on differencing significance map (DSM) is proposed. Wavelet coefficients are selected based on the energy packing efficiency of each sub-band. Significant Wavelet coefficients are quantized with uniform scalar zero zone quantizer. Significance map is created to store the indices of the significant coefficients. This map is encoded efficiently with less number of bits by applying Huffman coding on the differences between indices in the significance map. ECG records from the MIT-BIH arrhythmia database are selected as test data. For the record 117, the proposed technique achieves a compression ratio of 18.7:1 with lower percentage root mean square difference (PRD) compared to other threshold based methods. The proposed technique is tested for MIT-BIH arrhythmia record 119 and a compression ratio of 21.81:1 is achieved with a PRD value of 3.716% which is much lower compared to the reported PRD value of 5.0 and 5.5% of set partitioning in hierarchical tress (SPIHT) and analysis by synthesis ECG compressor (ASEC), respectively. The noise eliminating capability of the proposed technique is also demonstrated in this work. The proposed technique achieves the required compression ratio with less reconstruction error for GSM-based cellular telemedicine system.

Sinusoidal model-based analysis and classification of stressed speech

In this paper, a sinusoidal model has been proposed for characterization and classification of different stress classes (emotions) in a speech signal. Frequency, amplitude and phase features of the sinusoidal model are analyzed and used as input features to a stressed speech recognition system. The performances of sinusoidal model features are evaluated for recognition of different stress classes with a vector-quantization classifier and a hidden Markov model classifier. To find the effectiveness of these features for recognition of different emotions in different languages, speech signals are recorded and tested in two languages, Telugu (an Indian language) and English. Average stressed speech index values are proposed for comparing differences between stress classes in a speech signal. Results show that sinusoidal model features are successful in characterizing different stress classes in a speech signal. Sinusoidal features perform better compared to the linear prediction and cepstral features in recognizing the emotions in a speech signal.

Multichannel ECG Data Compression Based on Multiscale Principal Component Analysis

In this paper, multiscale principal component analysis (MSPCA) is proposed for multichannel electrocardiogram (MECG) data compression. In wavelet domain, principal components analysis (PCA) of multiscale multivariate matrices of multichannel signals helps reduce dimension and remove redundant information present in signals. The selection of principal components (PCs) is based on average fractional energy contribution of eigenvalue in a data matrix. Multichannel compression is implemented using uniform quantizer and entropy coding of PCA coefficients. The compressed signal quality is evaluated quantitatively using percentage root mean square difference (PRD), and wavelet energy-based diagnostic distortion (WEDD) measures. Using dataset from CSE multilead measurement library, multichannel compression ratio of 5.98:1 is found with PRD value 2.09% and the lowest WEDD value of 4.19%. Based on, gold standard subjective quality measure, the lowest mean opinion score error value of 5.56% is found.

Multiscale Energy and Eigenspace Approach to Detection and Localization of Myocardial Infarction

In this paper, a novel technique on a multiscale energy and eigenspace (MEES) approach is proposed for the detection and localization of myocardial infarction (MI) from multilead electrocardiogram (ECG). Wavelet decomposition of multilead ECG signals grossly segments the clinical components at different subbands. In MI, pathological characteristics such as hypercute T-wave, inversion of T-wave, changes in ST elevation, or pathological Q-wave are seen in ECG signals. This pathological information alters the covariance structures of multiscale multivariate matrices at different scales and the corresponding eigenvalues. The clinically relevant components can be captured by eigenvalues. In this study, multiscale wavelet energies and eigenvalues of multiscale covariance matrices are used as diagnostic features. Support vector machines (SVMs) with both linear and radial basis function (RBF) kernel and K-nearest neighbor are used as classifiers. Datasets, which include healthy control, and various types of MI, such as anterior, anteriolateral, anterioseptal, inferior, inferiolateral, and inferioposterio-lateral, from the PTB diagnostic ECG database are used for evaluation. The results show that the proposed technique can successfully detect the MI pathologies. The MEES approach also helps localize different types of MIs. For MI detection, the accuracy, the sensitivity, and the specificity values are 96%, 93%, and 99% respectively. The localization accuracy is 99.58%, using a multiclass SVM classifier with RBF kernel.

Wavelet-based electrocardiogram signal compression methods and their performances: A prospective review

Abstract Cardiovascular disease (CVD) is one of the most widespread health problems with unpredictable and life-threatening consequences. The electrocardiogram (ECG) is commonly recorded for computer-aided CVD diagnosis, human emotion recognition and person authentication systems. For effective detection and diagnosis of cardiac diseases, the ECG signals are continuously recorded, processed, stored, and transmitted via wire/wireless communication networks. But long-term continuous cardiac monitoring system generates huge volume of ECG data daily. Therefore, a reliable and efficient ECG signal compression method is highly demanded to meet the real-time constraints including limited channel capacity, memory and battery-power of remote cardiac monitoring, ECG record management and telecardiology systems. In such scenarios, the main objective of the ECG signal compression is to reduce the data rate for effective transmission and/or storage purposes without significantly distorting the clinical features of different kinds of PQRST morphologies contained in the recorded ECG signal. Numerous ECG compression methods have been proposed by exploiting the intra-beat correlation, inter-beat correlation and intra-channel correlation of the ECG signals. This paper presents a prospective review of wavelet-based ECG compression methods and their performances based upon findings obtained from various experiments conducted using both clean and noisy ECG signals. This paper briefly describes different kinds of compression techniques used in the one-dimensional wavelet-based ECG compression methods. Then, the performance of each of the wavelet-based compression methods is tested and validated using the standard MIT-BIH arrhythmia databases and performance metrics. The pros and cons of different wavelet-based compression methods are demonstrated based upon the experimental results. Finally, various practical issues involved in the validation procedures, reconstructed signal quality assessment, and performance comparisons are highlighted by considering the future research studies based on the recent powerful digital signal processing techniques and computing platform.

ECG Signal Compression using Discrete Sinc Interpolation

This paper presents a novel ECG data compression algorithm based on discrete sinc interpolation (DSI) technique. The compression and decompression of ECG data is achieved using discrete sinc interpolation (DSI), which is realized by an efficient discrete Fourier transform (DFT). The proposed algorithm is evaluated using MIT-BIH arrhythmia database (sampled at 360 Hz with 11 bits resolution). The performance of the proposed DSI based algorithm is compared with the performance of the widely used ECG data compression algorithms such as AZTEC, FAN, Hilton and Djohan algorithms. It is observed that higher compression ratio (CR) is achieved with a relatively lower percentage RMS difference (PRD) by DSI algorithm. The diagnostic distortion is measured in terms of average absolute error (AAE), which is lower in case of the DSI algorithm compared to the AZTEC and FAN algorithm

Detection of Shockable Ventricular Arrhythmia using Variational Mode Decomposition

Ventricular tachycardia (VT) and ventricular fibrillation (VF) are shockable ventricular cardiac ailments. Detection of VT/VF is one of the important step in both automated external defibrillator (AED) and implantable cardioverter defibrillator (ICD) therapy. In this paper, we propose a new method for detection and classification of shockable ventricular arrhythmia (VT/VF) and non-shockable ventricular arrhythmia (normal sinus rhythm, ventricular bigeminy, ventricular ectopic beats, and ventricular escape rhythm) episodes from Electrocardiogram (ECG) signal. The variational mode decomposition (VMD) is used to decompose the ECG signal into number of modes or sub-signals. The energy, the renyi entropy and the permutation entropy of first three modes are evaluated and these values are used as diagnostic features. The mutual information based feature scoring is employed to select optimal set of diagnostic features. The performance of the diagnostic features is evaluated using random forest (RF) classifier. Experimental results reveal that, the feature subset derived from mutual information based scoring and the RF classifier produces accuracy, sensitivity and specificity values of 97.23 %, 96.54 %, and 97.97 %, respectively. The proposed method is compared with some of the existing techniques for detection of shockable ventricular arrhythmia episodes from ECG.

Wavelet weighted blood vessel distortion measure for retinal images

Abstract In this paper, a novel wavelet transform based blood vessel distortion measure (WBVDM) is proposed to assess the image quality of blood vessels in the processed retinal images. The wavelet analysis of retinal image shows that different wavelet subbands carry different information about the blood vessels. The WBVDM is defined as the sum of wavelet weighted root of the normalized mean square error of subbands expressed in percentage. The proposed WBVDM is compared with other wavelet based distortion measures such as wavelet mean square error(WMSE), Relative WMSE(Rel WMSE) and root of the normalized WMSE(RNWMSE). The results show that WBVDM performs better in capturing the blood vessel distortion. For distortion in clinically nonsignificant regions, the proposed WBVDM shows a low value of 1.1676 compared to a large mean square error value of 7.9909. The evaluation of correlation using Pearson linear correlation coefficient (PLCC) and Spearman rank order correlation coefficient (SROCC) shows a higher value for the correlation between WBVDM and subjective score. The experimental observations show that WBVDM is able to capture the distortion in blood vessels more effectively and responds weakly to the distortion inherent in the other retinal features.