Mohammad Hassan Shariat

Activation detection of intracardiac electrogram during atrial fibrillation based on the variance equality test

Performance of the algorithms which process intracardiac electrograms (IEGMs) highly depends on the accuracy of estimating the times that electrical waves pass the area under the electrodes. Estimating these activation times (ATs) from IEGMs during atrial fibrillation (AF) is extremely challenging as electrical activities of atria are very complex, non-stationary, and irregular. In this paper, we propose a new activation detector which is based on the test of the equality of variance of two sets of data. At any time t, we consider two sets of IEGM data: 1) data in a bounded interval around t, 2) data in bounded intervals around the first interval. We show that the activation zone can be extracted by comparing the variance of these two sets, i.e., we introduce a new preprocessing approach and show that it can effectively highlight activation zones of IEGMs. Our simulation results on bipolar atrial IEGMs gathered during AF confirm the efficiency of the proposed preprocessing method.

Relay design for SNR maximization in MIMO communication systems

In this paper a multi-input multi-output (MIMO) communication system is considered in which a relay with multiple antennas is used to assist the data transmission from the source to destination. We assume that the relay applies a linear transform matrix on its received signal vector and retransmit the transformed vector to the destination node. We optimize the transform matrix by maximizing the average end-to-end SNR. The relay transfer matrix is related to the eigenvectors of correlation matrices of the two channels. For some special cases, we derive the optimum solution and the best unitary excitation matrix. We show that this solution is applicable in some cases where the instantaneous end-to-end SNR has to be maximized. Interestingly, the optimum transfer matrix is the dominant mode excitation, for the specific case where either the source-to-relay (SR) channel enticers or the relay-to-destination (RD) channel enticers are uncorrelated and have identical variances, i.e., in such a case all the relay power must be dissipated in the direction of dominant eigenvectors of SR and RD channels correlation matrices. Computer simulations are used to evaluate the effect of different relaying matrices on the end-to-end SNR.

Maximum likelihood cardiac conduction velocity estimation in the presence of ambiguities in the locations and activation times of the recording points

The cardiac conduction velocity vector (CCVV) provides valuable information that is used to distinguish/treat various heart diseases. The CCVV can be estimated using the intracardiac electrograms; i.e., the activation times (ATs) and locations of the recording catheters electrodes are used for the VV estimation. In practice, the ATs and locations of the electrodes are estimated, and if their estimation errors are not taken into consideration, they may significantly degrade the quality of the velocity estimation. Thus, in this paper, we assume that the locations and the ATs of the catheters electrodes are contaminated with white Gaussian noise with known variances, and we derive the maximum likelihood (ML) cardiac velocity estimator for the planar wavefront. We further calculate the ML estimate of the AT that the planar wavefront reaches at any given location. Our simulation results show the efficiency of the proposed ML estimators for the velocity and AT, and they also confirm the validity of the derived relations for the covariance of the error of the estimators.

Modified maximum time difference intracardiac conduction velocity estimation

The cardiac conduction velocity vector (VV) at a desired point in any chamber of the heart can be estimated by processing the local intracardiac electrograms, i.e., the activation times (ATs) and the locations of the recording catheters electrodes can be used for the VV estimation. In this paper, we modify the maximum time difference (MTD) method, which is a simple computational efficient cardiac VV estimation approach. In the MTD, first, the ATs of the electrograms are extracted, and the corresponding wavefronts are estimated. For each wavefront, the VV is estimated as the vector connecting the first to the last activated electrodes of the catheter divided by the time duration between the activation of those two electrodes. In the proposed modified MTD (MMTD) approach, which is slightly more computationally complex than the MTD, we divide the ATs of each considered wavefront into two groups, such that one group contains the electrodes with early activations, and the other contains those with late activations. We properly assign a time value and a location value to these groups and follow the same procedure as the MTD method to estimate the VV using the assigned values. Using synthetic data, we show that the proposed MMTD improves the quality of the cardiac conduction VV estimation of the MTD method, i.e., the MMTD is more robust to the AT estimation error and is able to determine the VV of the planar wavefronts more accurately.

Localization of the ectopic spiral electrical source using intracardiac electrograms during atrial fibrillation

Atrial fibrillation (AF) is a major global health issue as it is the most prevalent supraventricular arrhythmia. Multiple ectopic electrical sources in the atria are believed to sustain AF. Catheter-based ablation of these sources is considered an effective AF treatment. Based on the Hough transform (HT), we propose a general framework that processes the atrial intracardiac electrograms (IEGMs) to localize the tip of an ectopic source with a spiral wavefront shape. Using the locations of the catheters electrodes and the activation times of the IEGMs, we provide a method that can estimate the location of the tip of a spiral wavefront to be eliminated by ablation. By providing various examples, it is shown that the proposed method can accurately localize the tip of the spiral rotor.

Optimal training sequence for wireless MIMO channel estimation

The capability to increase the channel capacity of wireless systems without increasing the transmit power depends on the accuracy of channel estimation at the receiver. In this paper, the problem of designing training signals is investigated for correlated and uncorrelated multi-input multioutput (MIMO) channels. For this, the optimum channel estimation is briefly addressed in the sense of mean square error. Then the optimum training signal is designed for correlated and uncorrelated channels by maximizing the mutual information (carried by training signal) between the channel and the received sequence during the training. Simulations show that employing the optimal training provides improvement as compared with employing the orthogonal training. This improvement is more significant at low SNR and for correlated channels. However, this performance gap between the optimum training and orthogonal training is negligible in a high SNR scenario or for uncorrelated channels.

Cardiac conduction velocity estimation from sequential mapping assuming known Gaussian distribution for activation time estimation error

In this paper, we study the problem of the cardiac conduction velocity (CCV) estimation for the sequential intracardiac mapping. We assume that the intracardiac electrograms of several cardiac sites are sequentially recorded, their activation times (ATs) are extracted, and the corresponding wavefronts are specified. The locations of the mapping catheters electrodes and the ATs of the wavefronts are used here for the CCV estimation. We assume that the extracted ATs include some estimation errors, which we model with zero-mean white Gaussian noise values with known variances. Assuming stable planar wavefront propagation, we derive the maximum likelihood CCV estimator, when the synchronization times between various recording sites are unknown. We analytically evaluate the performance of the CCV estimator and provide its mean square estimation error. Our simulation results confirm the accuracy of the proposed method and the error analysis of the proposed CCV estimator.

Computationally efficient method for localizing the spiral rotor source using synthetic intracardiac electrograms during atrial fibrillation

Atrial fibrillation (AF), the most common sustained cardiac arrhythmia, is an extremely costly public health problem. Catheter-based ablation is a common minimally invasive procedure to treat AF. Contemporary mapping methods are highly dependent on the accuracy of anatomic localization of rotor sources within the atria. In this paper, using simulated atrial intracardiac electrograms (IEGMs) during AF, we propose a computationally efficient method for localizing the tip of the electrical rotor with an Archimedean/arithmetic spiral wavefront. The proposed method deploys the locations of electrodes of a catheter and their IEGMs activation times to estimate the unknown parameters of the spiral wavefront including its tip location. The proposed method is able to localize the spiral as soon as the wave hits three electrodes of the catheter. Our simulation results show that the method can efficiently localize the spiral wavefront that rotates either clockwise or counterclockwise.

Optimal Non-regenerative Linear MIMO Relay for Orthogonal Space Time Codes

We consider a multi-input multi-output (MIMO) communication system in which a linear non-regenerative MIMO relay assists the data transmission between a source and a destination node using orthogonal space time block codes (OSTBC). We derive the relay transformation matrix (RTM) by minimizing the symbol error rate (or maximizing the capacity) of the system under a constraint on the relays transmitted power. The RTM is derived in the presence of the direct source-destination path and for spatially-correlated additive noise vectors. We show that the proposed RTM can significantly enhance the performance of the system in terms of the capacity/symbol error rate. Interestingly, our simulation results, in various conditions, reveal that the proposed method provides close to the optimum data rate even for the system without OSTBC. This suboptimal approach is justified, as the complexity of computing the proposed RTM is significantly less than the computation complexity of the RTM which maximizes the non-regenerative capacity.

Texture classification in different illumination conditions via testing the covariance matrices and mean vectors

Texture classification is of utmost importance in the image processing. In this paper the problem of texture classification is considered based on testing the covariance matrices and mean vectors. This allows us to determine the class of different images without the necessity of the training data. The generalized likelihood ratio (GLR) test is derived in order to classify several images. To make the classification robust to illuminance changes, we assume that the means of different images in one group, could differ by a constant value. Consequently the proposed test is invariant to the constant difference in the means of observations in each group. Computer simulations also confirm the efficiency of the classifier in dealing with the images with different illumination conditions.