Automatic identification of spike-wave events and non-convulsive seizures with a reduced set of electrodes.

Epileptiform activity in the brain, whether localized or generalized, constitutes an important category of abnormal electroencephalogram (EEG). Seizures are episodes of relatively brief disturbances of mental, motor or sensory activity caused by paroxysmal cerebral activity. They are not always accompanied by the characteristic convulsions that we commonly associate with the word epilepsy. In this case, they may be referred to as non-convulsive status epilepticus (NCSE) or as absence seizures (formerly called "petit mal" seizures). They often manifest themselves in scalp-recorded EEG as large-amplitude spike-wave "patterns" (or "events"), usually occurring in bursts. If left undetected and untreated, they can potentially cause significant brain and behavioral dysfunctions, interfere with information processing, or otherwise contribute to altered mental status. In this paper, we describe an algorithm to be implemented in a prototype BrainScope_ED instrument meant to alert to a detected seizure in an emergency department (ED) or other clinical setting. BrainScope_ED uses a reduced electrode set (8 instead of 19). The proposed signal processing algorithm is based on the detection of spike-wave events obtained from a wavelet analysis of the EEG signal, combined with an analysis of the complexity of the EEG using fractal dimension estimates. We show that this algorithm has excellent sensitivity and specificity. In particular, the fractal analysis is a key factor in the removal of falsely detected spike-wave events (false positives) that can be caused by voluntary or involuntary artifacts such as fast eyelid flutter.

Optimal denoising of brainstem Auditory Evoked Response (BAER) for automatic peak identification and brainstem assessment.

Brainstem auditory evoked responses (BAER) are transient signals embedded in the EEG recorded from scalp electrodes, when a subject is presented with a series of acoustic clicks. These signals typically have a signal-to-noise ratio (SNR) well below -10 dB. The extraction of BAER signals from the EEG for the purpose of automatically computing features of interest from the BAER waveform(s) is described in this paper. These features are: 1) Presence of an actual BAER response (at least peak I), 2) Presence of peak V, 3) Inter-peak latency I-V. We propose to combine a signal-adaptive denoising technique based on complex wavelets with a signal quality metric referred to as the FSP variance ratio for quantitative evaluation of signal quality in order to optimally denoise BAER signals and perform reliable waveform analysis.

Fast wavelet estimation of weak biosignals.

Wavelet-based signal processing has become commonplace in the signal processing community over the past decade and wavelet-based software tools and integrated circuits are now commercially available. One of the most important applications of wavelets is in removal of noise from signals, called denoising, accomplished by thresholding wavelet coefficients in order to separate signal from noise. Substantial work in this area was summarized by Donoho and colleagues at Stanford University, who developed a variety of algorithms for conventional denoising. However, conventional denoising fails for signals with low signal-to-noise ratio (SNR). Electrical signals acquired from the human body, called biosignals, commonly have below 0 dB SNR. Synchronous linear averaging of a large number of acquired data frames is universally used to increase the SNR of weak biosignals. A novel wavelet-based estimator is presented for fast estimation of such signals. The new estimation algorithm provides a faster rate of convergence to the underlying signal than linear averaging. The algorithm is implemented for processing of auditory brainstem response (ABR) and of auditory middle latency response (AMLR) signals. Experimental results with both simulated data and human subjects demonstrate that the novel wavelet estimator achieves superior performance to that of linear averaging.