Xuan Kong

An adaptive estimation of periodic signals using a Fourier linear combiner

Presents an adaptive algorithm for estimating from noisy observations, periodic signals of known period subject to transient disturbances. The estimator is based on the LMS algorithm and works by tracking the Fourier coefficients of the data. The estimator is analyzed for convergence, noise misadjustment and lag misadjustment for signals with both time invariant and time variant parameters. The analysis is greatly facilitated by a change of variable that results in a time invariant difference equation. At sufficiently small values of the LMS step size, the system is shown to exhibit decoupling with each Fourier component converging independently and uniformly. Detection of rapid transients in data with low signal to noise ratio can be improved by using larger step sizes for more prominent components of the estimated signal. An application of the Fourier estimator to estimation of brain evoked responses is included. >

Watermarking medical signals for telemedicine

The modern telecommunication infrastructure supports the possibility of delivering quality health care without the physical presence of medical experts. The integrity of biomedical signals being transmitted through the communication channels must be established before their utilization. This paper investigates three digital watermarking techniques for signal integrity verification in an electroencephalogram (EEG) monitoring application for brain injury detection. The techniques studied are the patchwork, least-significant bit and quantization watermarking methods. The three techniques are evaluated and compared in the following areas: sensitivity to noise contamination, robustness to EEG signal characteristic changes due to brain injury, and consistency under various communication channel models. The patchwork method performs best for noise contamination rejection among the three methods. The noise contamination detection rates of all three methods remain relatively stable across a wide range of EEG characteristics.

Context-based lossless and near-lossless compression of EEG signals

We study compression techniques for electroencephalograph (EEG) signals. A variety of lossless compression techniques, including compress, gzip, bzip, shorten, and several predictive coding methods, are investigated and compared. The methods range from simple dictionary based approaches to more sophisticated context modeling techniques. It is seen that compression ratios obtained by lossless compression are limited even with sophisticated context based bias cancellation and activity based conditional coding. Though lossy compression can yield significantly higher compression ratios while potentially preserving diagnostic accuracy, it is not usually employed due to legal concerns. Hence, we investigate a near lossless compression technique that gives quantitative bounds on the errors introduced during compression. It is observed that such a technique gives significantly higher compression ratios (up to 3-bit/sample saving with less than 1% error). Compression results are reported for EEGs recorded under various clinical conditions.

Adaptive estimation of latency changes in evoked potentials

Changes in latency of evoked potentials (EP) may indicate clinically and diagnostically important changes in the status of the nervous system. A low signal-to-noise ratio of the EP signal makes it difficult to estimate small, transient, time-varying changes in latency, or delays. Here, the authors present an adaptive algorithm that estimates small delay (latency change) values even when EP signal amplitudes are time-varying. When the delay is time invariant, the adaptive algorithm produces an unbiased estimate with delay estimation error less than half of the sampling interval. A lower estimation error variance is obtained when, in a pair of signals, the adaptive algorithm delays the signal with the higher SNR. The adaptive delay estimation algorithm was tested on intra-operative recordings of somatosensory EP, and analysis of those recordings reveals that the anesthetic etomidate produces a step change in the amplitude and latency of the EP signals.

Adaptive estimation of latency change in evoked potentials by direct least mean p-norm time-delay estimation

Evoked potentials (EP) have been widely used to quantify neurological system properties. Changes in EP latency may indicate impending neurological injury. Traditional EP analyses are developed under the condition that the background noise in EP analysis are Gaussian distributed. This paper proposes a latency change detection and estimation algorithm under /spl alpha/-stable noise condition, a generalization of Gaussian noise assumption. An analysis shows that the /spl alpha/-stable model fits the noises found in the impact acceleration experiment under study better than the Gaussian model. The robustness of the proposed algorithm is demonstrated through computer simulations and experimental data analysis under both Gaussian and /spl alpha/-stable noise environments.

Performance analysis of adaptive eigenanalysis algorithms

We present a rigorous analysis of several popular forms of short memory adaptive eigenanalysis algorithms using a stochastic averaging method. A first-order analysis shows that the algorithms do not have any equilibrium points despite published claims to the contrary. Through averaging analysis, we show that they hover around an appropriate eigenvector. A second-order analysis is also given without the Gaussian noise assumption, and our results greatly outperform an earlier approximation in the literature. The second-order analysis has been of much interest in the offline study but, in the dynamic adaptive case, is uncommon.

Quantification of injury-related EEG signal changes using distance measures

Novel indicators based on distance measures are developed and compared to quantify changes in electroencephalogram signal resulting from hypoxic-asphyxic injury. An injury index is derived based on the measures. The Itakura distance-based index is found to have the highest correlation with the long-term outcome as measured by the neurological deficit scores.

Nonlinear changes in brain's response in the event of injury as detected by adaptive coherence estimation of evoked potentials

Injury-related changes in evoked potentials are studied with the aid of the coherence function, which effectively measures the degree of linear association between a pair of signals recorded during normal and abnormal states of the brain. The performance of an adaptive algorithm for estimating coherence function is studied, and the effects of additive noise on the estimated coherence function is discussed. Further, a linearity index is formulated and, through analysis and simulations, the index is shown to respond in a predictable manner to increasing nonlinearity while maintaining the robustness to the observation noise. Somatosensory evoked potentials are shown to be sensitive to injury resulting from acute cerebral hypoxia. The authors analyze the somatosensory evoked potentials recorded from anesthetized cats during inhalation of 8-9% oxygen gas mixtures and during recovery with 100% oxygen. Analyses of the experimental data show a very sharp drop in the magnitude coherence estimates during hypoxic injury and a corresponding rapid decline in the linearity index at the very early stages of the hypoxic injury. Thus, injury may lead to nonlinearities in the electrical response of the brain, and such measurements analyzed by the adaptive coherence estimation method may be used for diagnostic purposes.<<ETX>>

Latency change estimation for evoked potentials via frequency selective adaptive phase spectrum analyzer

This paper addresses the problem of detecting and estimating latency changes in evoked potentials (EPs). EPs have been widely used to quantify neurological system properties. Transient and time-varying changes in latency may indicate impending neurological injury. Traditional time averaging and correlation methods for EP latency estimation are inefficient under low signal-to-noise ratio (SNR) and/or strong periodic interference conditions. This paper proposes an adaptive phase spectral time delay estimation method to detect and estimate the time-varying latency changes when both the SNR and the signal-to-interference ratio (SIR) are low. A theoretical analysis and computer simulation demonstrate that the proposed method can track the time-varying latency changes effectively and accurately when both the SNR and the SIR are as low as -5 dB. The method is also suitable for real time detection and estimation of the latency changes.

Quantification of injury-related EEG signal changes using Itakura distance measure

Accurate detection and characterization of changes in the EEG signal is crucial for clinical assessment of the neurological system condition. Several distance measures are tested and evaluated for their effectiveness of detecting injury-related changes in EEG. Itakura distance is found to be a very efficient means to characterize changes in EEG for both signaling injury and predicting recovery. The efficiency of the Itakura distance measure is further established through a comparison study of spectral distance measure and Kullback-Leibler information.