Performance of Empirical Mode Decomposition for Frequency Identification in SSVEP Based BCI.

Empirical mode decomposition based conventional correlation (EMDCC) method is proposed to identify the frequency components in steady state visual evoked potentials (SSVEP) in electroencephalogram(EEG).The main aim of the proposed EMDCC method is to recognise narrow band frequency components that are present in SSVEP. The study is evaluated on two datasets. The first one is a 40 target benchmark dataset obtained from 35 subjects and the second is a 4 class Inhouse dataset collected from 10 healthy participants. The mean detection accuracy of the conventional correlation method is 85.64 % for the benchmark dataset and it is improved to 93.79 % in the proposed method. The mean detection accuracy of the conventional correlation method is 67.5 % for the Inhouse dataset and it is increased to 82.5 % in the proposed method. The mean detection accuracy of the proposed EMDCC method is also compared to time-weighting canonical correlation analysis (TWCCA) for the benchmark dataset. The mean detection accuracy of TWCCA is 91.04 %. Hence the results show better detection accuracies in the proposed EMDCC method than the simple conventional correlation method and also the existing TWCCA method.

Reentry in cardiac ventricular epicardial tissue due to SCN5A L812Q gene mutation: a computational study.

Cardiovascular diseases are the major cause of sudden death. Brugada syndrome is an inherited rare disease, that leads to death due to ventricular fibrillation (VF). Brugada Syndrome is related to mutations in the genes that encode SCN5A, a subunit of sodium ion channel (NaV). This computational study investigates the mechanism of loss of function gene mutation (SCN5A L812Q) in sodium ion channel that leads to spiral wave and further develops into VF in an epicardial tissue with homozygous condition. Study was made on wild type, L812Q heterozygous mutated and homozygous mutated ventricular tissues. Ten Tusscher human ventricular cell model (TP06) was used for the simulation study. VF is developed when a spiral wave that causes ventricular arrhythmia breaks. This leads to the formation of multiple spiral waves that are activated on different regions of the ventricles called wave break. This is observed in the epicardial tissue with homozygous condition as the effect of SCN5A L812Q gene mutation. This indicates that VF occurs in the SCN5A L812Q gene mutated homozygous ventricular epicardial tissue that may further lead to Brugada syndrome.

Performance Evaluation of Compressed Deep CNN for Motor Imagery Classification using EEG.

Recently, deep learning and convolutional neural networks (CNNs) have reported several promising results in the classification of Motor Imagery (MI) using Electroencephalography (EEG). With the gaining popularity of CNN-based BCI, the challenges in deploying it in a real-world mobile and embedded device with limited computational and memory resources need to be explored. Towards this objective, we investigate the impact of the magnitude-based weight pruning technique to reduce the number of parameters of the pre-trained CNN-based classifier while maintaining its performance. We evaluated the proposed method on an open-source Korea University dataset which consists of 54 healthy subjects' EEG, recorded while performing right-and left-hand MI. Experimental results demonstrate that the subject-independent model can be maximumly pruned to 90% sparsity, with a compression ratio of 4.77× while retaining classification accuracy at 84.44% with minimal loss of 0.02% when compared to the baseline model's performance. Therefore, the proposed method can be used to design more compact deep CNN- based BCIs without compromising on their performance.

A Simulation Study on Electrical Activity of Ventricular Endocardial Tissue due to SCN5A L812Q Mutation.

Brugada Syndrome is a rare arrhythmia, hereditary in nature. It is caused due to mutation in genes that encodes sodium ion channels and it results sudden cardiac death in young adults. This paper aims to model a two dimensional SCN5A L812Q mutated endocardial tissue by modifying the model equations for sodium ion channel in the Ten Tusscher model for human ventricular tissue. Results show that the propagation of electrical activity in the mutated cells is slower when compared to the normal cells of the endocardial tissue. From this it is concluded that there is a large reduction of sodium current in the mutated region of the endocardial tissue. This leads to reduction in the total ionic current as well and further reduces the membrane potential. It also leads to the slower propagation of action potential in the mutated region when compared to the normal endocardial tissue.Clinical Relevance- This establishes the propagation of electrical activity in endocardial tissue for SCN5A L812Q gene mutation that results in arrhythmia called Brugada Syndrome.

Particle swarm optimization based fusion of ultrasound echographic and elastographic texture features for improved breast cancer detection.

Breast cancer remains the main cause of cancer deaths among women in the world. As per the statistics, it is the most common killer disease of the new era. Since 2008, breast cancer incidences have increased by more than 20%, while mortality has increased by 14%. The statistics of breast cancer incidences as per GLOBOCAN project for the years 2008 and 2012 show an increase from 22.2 to 27% globally. In India, breast cancer accounts for 25% to 31% of all cancers in women. Mammography and Sonography are the two common imaging techniques used for the diagnosis and detection of breast cancer. Since Mammography fails to spot many cancers in the dense breast tissue of young patients, Sonography is preferred as an adjunct to Mammography to identify, characterize and localize breast lesions. This work aims to improve the performance of breast cancer detection by fusing the texture features from ultrasound elastographic and echographic images through Particle Swarm Optimization. The mean classification accuracy of Optimum Path Forest Classifier is used as an objective function in PSO. Seven performance metrics were computed to study the performance of the proposed technique using GLCM, GLDM, LAWs and LBP texture features through Support Vector Machine classifier. LBP feature provides accuracy, sensitivity, specificity, precision, F1 score, Mathews Correlation Coefficient and Balanced Classification Rate as 96.2%, 94.4%, 97.4%, 96.2%, 95.29%, 0.921, 95.88% respectively. The obtained performance using LBP feature is better compared to the other three features. An improvement of 6.18% in accuracy and 11.19% in specificity were achieved when compared to those obtained with previous works.

Improving The Performance of Motor Imagery Based Brain-Computer Interface Using Phase Space Reconstruction.

In recent decades, motor imagery (MI) based brain-computer interface (BCI) is served as a control system or rehabilitation tool for motor disabled people. But it has limited applications because of its lower classification performance (classification accuracy, Cohen's kappa coefficient and etc.). The performance depends on the feature extraction techniques and extraction of relevant features from the brain is challenging task. The existing techniques have low classification performance and are computationally inefficient. This paper introduces phase space reconstruction (PSR) to detect various MI activities and improve the performance of the system. First, raw signals were decomposed into multiple frequency sub-bands using filter bank technique. Second, PSR was applied to each sub-band and dynamical behavior of the brain activities has been analyzed. The optimal parameters (time delay and embedding dimension) of PSR were calculated by average mutual information (AMI) and false nearest neighbors (FNN) methods. The time delay and embedding dimension extracted significant features related to MI activities. The significant features were fed into multi-class support vector machine (SVM) and performance of the classifier was evaluated. The performance of the system is based on classification accuracy (%CA) and Cohen's kappa coefficient (K). The proposed algorithm and classifier were tested on BCI competition-2005, MI dataset-III-a. The results show that the proposed technique increases the classification accuracy by 3.7% and achieved higher performance (%CA = 89.20% and K= 0.85).

Simulation study of the ionic mechanisms underlying Torsade de Pointes in a 2D cardiac tissue.

BACKGROUND: To understand the ionic mechanism behind the genesis of Torsade de Pointes (TdP) occurring with long QT syndrome 2 (LQTS2) in a remodelled transmural tissue. METHODS: The TP06 model is used to simulate the electrical activity of cells in a 2D transmural ventricular model. LQTS2 is realised by reducing the potassium current (IKr) to 0.5 in each cell. Each cell of the tissue is remodelled by increasing the conductance of calcium current (GCaL). The above two factors make the cells prone to early after depolarizations (EADs) development. The rise in GCaL that can develop a sustained TdP at normal pacing rate is determined from this study. A look at the calcium dynamics, sodium-calcium exchanger current (INaCa) and slow delayed rectifier potassium current (IKs) distribution maps of the tissue helps us in analysing the mechanism of TdP generation. RESULTS: A TdP type pattern at normal pacing rate is generated when GCaL is more than 3.5 times the control parameter. From the M-cell island, an adequate number of cells spontaneously release calcium from their sarcoplasmic reticulum leading to increased intracellular calcium and inward sodium current through the sodium calcium exchanger current (INaCa). These contribute to the development of EADs which create a depolarising wavefront that triggers TdP in the tissue. When GCaL is less than 3.5 times the control value, premature ventricular complexes (PVC) occur interspersed between normal beats. CONCLUSION: Normal pacing rates can induce a multi focal TdP when sufficient number of M-cells simultaneously undergo spontaneous calcium release (SCR) events.

Study of factors affecting the progression and termination of drug induced Torsade de pointes in two dimensional cardiac tissue.

INTRODUCTION: To study the conditions leading to the initiation and termination of drug induced Torsade de pointes (TdP) along with QT prolongation. METHODS: A 2D anisotropic transmural section of the ventricular myocardium is modeled using the TP06 equations and the cells are interconnected with gap junction conductances (GJC). The tissue is remodeled by reducing the repolarization reserve (by increasing calcium current (ICaL)) of all cells thus making them vulnerable to development of early after depolarizations (EADs). RESULTS: Clinical risk conditions like decreased potassium current (IKr), bradycardia, hypokalemia and short-long-short (SLS) triggering sequences are included in the tissue. A pseudo-electrocardiogram is created to realize the intensity of remodeling required in presence of risk factors to initiate TdP. On initiating TdP, the effect of increasing GJC and decreasing ICaL is shown to terminate a non-self-limiting TdP. CONCLUSION: Without the inclusion of underlying increase in ICaL along with risk factors, TdP cannot be initiated.

Optimal channels election for multi-Channels SVEP Detection and Classification in BCIS.

Many multi-channel techniques for Steady-State Visual-Evoked Potential (SSVEP) detection from EEG have shown significant improvement in the performance of Brain-Computer Interfaces (BCIs). Multi-channel methods, generally involve deriving a spatial filter to linearly combine the EEG channels so as to minimize the noise energy and enhance the SSVEP response. In this paper, three state of the art multi-channel techniques are studied and compared. The performance of the classifiers for varying number and combination of the EEG channels is studied to determine the optimal choice of channels that yield maximum classification accuracy. The correlation of different channel parameters with the net montage performance is also investigated. Results indicate that Minimum Energy Channel (MEC) based classifier yields the highest accuracy values using 6 channels for all the 3 subjects. Significance of non-occipital locations for signal acquisition has been observed. Further, results indicate that the choice of channels to be used in the montage is to be made keeping in mind their effectivesignal strength, co-channel noise correlation values and signal to noise ratios. This ensures that a particular montage has effectively assimilated the signal and noise components.

Automatic Severity Assessment of Dysarthria using State-Specific Vectors.

In this paper, a novel approach to assess the severity of the dysarthria using state-specific vector (SSV) of phone-cluster adaptive training (phone-CAT) acoustic modeling technique is proposed. The dominant component of the SSV represents the actual pronunciations of a speaker. Comparing the dominant component for unimpaired and each dysarthric speaker, a phone confusion matrix is formed. The diagonal elements of the matrix capture the number of correct pronunciations for each dysarthric speaker. As the degree of impairment increases, the number of phones correctly pronounced by the speaker decreases. Thus the trace of the confusion matrix can be used as objective cue to assess di?erent severity levels of dysarthria based on a threshold rule. Our proposed objective measure correlates with the standard Frenchay dysarthric assessment scores by 74 % on Nemours database. The measure also correlates with the intelligibility scores by 82 % on universal access dysarthric speech database.

Regional Variation of Gap Junction Conductance on Reentrant Arrhythmia.

Arrhythmia generating conditions like ventricular hypertrophy, myocardial infarction or ischemia modify the intercellular coupling by modifying the conductance of gap junctions in the normal electric propagation pathway of the heart. A discrete ventricular cell network of 100x100 cells interconnected using resistive gap junctions is simulated to study the effect of size, shape and position of inhomogeneity as well as the value of gap junction conductance of inhomogeneity on the occurrence of reentrant arrhythmia. In addition to lowering the conduction, a unidirectional block is also created using exactly timed stimulation inputs thus setting the ideal conditions for a reentrant activation to arise from the zone of varied gap junction conductance. The shape and endurance of generated reentrant waves is analyzed. The electrical activity of each cell is simulated using the Ten Tusscher –Panfilov 2006 model. Simulation results show that the positions as well as the size of the inhomogeneity play a major role in the creation of reentrant waves while the shape of the inhomogeneity does not have a significant effect. Also, reentrant waves occur at a certain level of decoupling. Too much or too little decoupling also doesn’t induce reentrant waves. The amplitude and duration of action potential is heavily dependent on the gap junction conductance.

Near-lossless multichannel EEG compression based on matrix and tensor decompositions.

A novel near-lossless compression algorithm for multichannel electroencephalogram (MC-EEG) is proposed based on matrix/tensor decomposition models. MC-EEG is represented in suitable multiway (multidimensional) forms to efficiently exploit temporal and spatial correlations simultaneously. Several matrix/tensor decomposition models are analyzed in view of efficient decorrelation of the multiway forms of MC-EEG. A compression algorithm is built based on the principle of "lossy plus residual coding," consisting of a matrix/tensor decomposition-based coder in the lossy layer followed by arithmetic coding in the residual layer. This approach guarantees a specifiable maximum absolute error between original and reconstructed signals. The compression algorithm is applied to three different scalp EEG datasets and an intracranial EEG dataset, each with different sampling rate and resolution. The proposed algorithm achieves attractive compression ratios compared to compressing individual channels separately. For similar compression ratios, the proposed algorithm achieves nearly fivefold lower average error compared to a similar wavelet-based volumetric MC-EEG compression algorithm.

Acoustic analysis and detection of hypernasality using a group delay function.

In this paper, we describe a group delay-based signal processing technique for the analysis and detection of hypernasal speech. Our preliminary acoustic analysis on nasalized vowels shows that, even though additional resonances are introduced at various frequency locations, the introduction of a new resonance in the low-frequency region (around 250 Hz) is found to be consistent. This observation is further confirmed by a perceptual analysis carried out on vowel sounds that are modified by introducing different nasal resonances, and an acoustic analysis on hypernasal speech. Based on this, for subsequent experiments the focus is given only to the low-frequency region. The additive property of the group delay function can be exploited to resolve two closely spaced formants. However, when the formants are very close with considerably wider bandwidths as in hypernasal speech, the group delay function also fails to resolve. To overcome this, we suggest a band-limited approach to estimate the locations of the formants. Using the band-limited group delay spectrum, we define a new acoustic measure for the detection of hypernasality. Experiments are carried out on the phonemes /a/, /i/, and /u/ uttered by 33 hypernasal speakers and 30 normal speakers. Using the group delay-based acoustic measure, the performance on a hypernasality detection task is found to be 100% for /a/, 88.78% for /i/ and 86.66% for /u/. The effectiveness of this acoustic measure is further cross-verified on a speech data collected in an entirely different recording environment.

HVS scheme for DICOM image compression: Design and comparative performance evaluation.

Advanced digital imaging technology in medical domain demands efficient and effective DICOM image compression for progressive image transmission and picture archival. Here a compression system, which incorporates sensitivities of HVS coded with SPIHT quantization, is discussed. The weighting factors derived from luminance CSF are used to transform the wavelet subband coefficients to reflect characteristics of HVS in best possible manner. Mannos et al. and Daly HVS models have been used and results are compared. To evaluate the performance, Eskicioglu chart metric is considered. Experiment is done on both Monochrome and Color Dicom images of MRI, CT, OT, and CR, natural and benchmark images. Reconstructed image through our technique showed improvement in visual quality and Eskicioglu chart metric at same compression ratios. Also the Daly HVS model based compression shows better performance perceptually and quantitatively when compared to Mannos et el. model. Further "bior4.4" wavelet filter provides better results than "db9" filter for this compression system. Results give strong evidence that under common boundary conditions; our technique achieves competitive visual quality, compression ratio and coding/decoding time, when compared with jpeg2000 (kakadu).

Quantitative evaluation of left ventricle performance from two dimensional echo images.

OBJECTIVES: We sought to quantify the left ventricle systolic dysfunction by a geometric index from two-dimensional (2D) echocardiography by implementing an automated fuzzy logic edge detection algorithm for the segmentation. BACKGROUND: The coronary injuries have repercussions on the left ventricle producing changes on wall contractility, the shape of the cavity, and as a whole changes on the ventricular function. METHODS: 2D echocardiogram and M-mode recordings were performed over the control group and those with the dysfunctions. From 2D recordings, individual frames were extracted for at least five cardiac cycles and then segmentation of left ventricle was done by automated fuzzy systems. In each frame, the volumes are measured and a geometric index, eccentricity ratio (ER), was derived. The endocardial fractional shortening (FS), midwall fractional shortening (mFS), and the relative wall thickness (RWT) were also measured in each case. RESULTS: Depressed value of endocardial FS (20.39 +/- 5.43 vs 34.28 +/- 9.36, P = 0.0046), mFS (33 +/- 8.3 vs 52.5 +/- 11.7, P = 0.0047), and the RWT (0.337 +/- 0.096 vs 0.525 +/- 0.119, P = 0.0002) was observed with dysfunction. ER measured at end-diastole (2.86 +/- 0.703 vs 4.14 +/- 0.38) and end-systole (3.14 +/- 0.79 vs 5.48 +/- 0.74) was found to be decreased in the dysfunction group and more significant at the end-systole (P = 0.00017 vs 6.6E-06). CONCLUSION: This work concludes that the regional and global left ventricle systolic dysfunction can be assessed by the ER measured at end-diastole and end-systole from 2D echocardiogram and may contribute to the high rate of cardiovascular disorders.

Brain tissue phantoms for optical near infrared imaging.

In this study, solid, stable, and cost-effective optical phantoms of scalp-skull, white matter and grey matter are developed by inverse method. To begin with, to obtain a range of optical parameters, absorption and reduced scattering coefficients (mu(a) and mu(s)', respectively), 20 homogeneous phantoms were made of paraffin wax by using optically contrast black and highly scattering white colouring materials in different combination. By comparing the measured reflectance values for each phantom got from the four channel reflectometer with that obtained from steady-state diffusion equation, the values of mu(a) and mu(s)' were determined. Next, phantoms which exhibit specific optical properties of scalp-skull, white and grey matter are developed iteratively by comparing actual reflectance measurements got by adjusting the colour concentration with the predicted reflectance values from the diffusion equation. This is done as follows: to obtain mu(a) of 0.04 mm(-1) for scalp-skull, 9.5 mg of black dye per 100 ml of wax added since more attenuation of light occurs in bone tissue. To obtain a mu(s)' 6.0 mm(-1) for white matter in brain tissue, 190 mg of white dye per 100 ml of wax was used to facilitate more scatter of light. The colour concentrations of phantoms were then adjusted to obtain the predetermined values of optical parameters for scalp-skull, grey and white matter.

A novel multiple frequency stimulation method for steady state VEP based brain computer interfaces.

The objective is to increase the number of selections in brain computer interfaces (BCI) by recording and analyzing the steady state visual evoked potential response to dual stimulation. A BCI translates the VEP signals into user commands. The frequency band from which stimulation frequency can be selected is limited for SSVEP. This paper discusses a method to increase the number of commands by using a suitable combination of frequencies for stimulation. A biopotential amplifier based on the driven right leg circuit (DRL) is used to record 60 s epochs of the SSVEP (O(z)-A(1)) on 15 subjects using simultaneous overlapped stimulation (6, 7, 12, 13 and 14 Hzs and corresponding half frequencies). The power spectrum of each recording is obtained by frequency domain averaging of 400 ms SSVEPs and the spectral peaks were normalized. The spectral peaks of the combination frequencies of stimulation are predominant compared to individual stimulating frequencies. This method increases the number of selections by using a limited number of stimulating frequencies in BCI. For example, six selections are possible by generating only three frequencies.

Detection of left ventricle systolic dysfunction from shape deformity.

Coronary artery disease producing ischemic cardiomyopathy is the most frequent cause of left ventricular systolic dysfunction. Non-ischemic cardiomyopathies can also produce systolic dysfunction; they may be inherited as genetic disorders or occur sporadically. These coronary injuries have repercussions on the left ventricle producing changes on wall contractility, the shape of the cavity and also changes on ventricular function. This study is focused on the 2D echocardiograms of the left ventricle. Apical two chamber and four chamber view recordings were performed on normal and systolic dysfunction subjects. Individual frames were extracted for at least five cardiac cycles. After pre-processing these images, segmentation of the left ventricle was performed by Fuzzy systems. Then the volumes were measured by single and biplane methods along with the perimeter, short axis length and long axis length in each frame, from which the two indices Sphericity Index (SI) and Normalized Eccentricity Index (NEI) was determined. It was found that the diastolic phase is short in the case of systolic dysfunction, and its volume variation is not uniform as in the normal case. Also, in the case of systolic dysfunction, the span of either the long or short axis length variation is less than 0.5 cm. This depicts that akinesis is in the corresponding direction; the value of SI is less than 2 for systolic dysfunction. A sharp peak is seen at each systole point in the NEI plot and also its variation is smooth in subjects having LVEF > 45%, which is not the case for dysfunction.

Neural network based approach for anomaly detection in the lungs region by electrical impedance tomography.

In this paper, we have shown a simple procedure to detect anomalies in the lungs region by electrical impedance tomography. The main aim of the present study is to investigate the possibility of anomaly detection by using neural networks. Radial basis function neural networks are used as classifiers to classify the anomaly as belonging to the anterior or posterior region of the left lung or the right lung. The neural networks are trained and tested with the simulated data obtained by solving the mathematical model equation governing current flow through the simulated thoracic region. The equation solution and model simulation are done with FEMLAB. The effect of adding a higher number of neurons to the hidden layer can be clearly seen by the reduction in classification error. The study shows that there is interaction between the size (radius) and conductivity of anomalies and for some combination of these two factors the classification error of neural networks will be very small.