Scalable and energy efficient seizure detection based on direct use of compressively-sensed EEG data on an ultra low power multi-core architecture.

Extracting information from dense multi-channel neural sensors for accurate diagnosis of brain disorders necessitates computationally expensive and advanced signal processing approaches to analyze the massive volume of recorded data. Compressive Sensing (CS) is an efficient method for reducing the computational complexity and power consumption in the resource-constrained multi-site neural systems. However, reconstructing the signal from compressed measurements is computationally intensive, making it unsuitable for real-time applications such as seizure detection. In this paper, a seizure detection algorithm is proposed to overcome these limitations by circumventing the reconstruction phase and directly processing the compressively sampled EEG signals. The Lomb-Scargle Periodogram (LSP) is used to extract the spectral energy features of the compressed data. Performance of the seizure detector using non-linear support vector machine (SVM) classifier, tested on 24 patients of the CHB-MIT data-set for compression ratios (CR) of 1-64x, is 96-93%, 92-87%, 0.95-0.91, and <1 s for sensitivity, accuracy, the area under the curve, and latency, respectively. A power-efficient classification method based on the utilization of dual linear SVM classifiers is proposed. The proposed classification method based on the dual linear SVM classification achieved better classification performance compared to commonly used classifiers, such as K-nearest neighbor, random forest, artificial neural network, and linear SVM, while consuming low power in comparison to non-linear SVM kernels. The hardware-optimized implementation of this algorithm is proposed on a low-power multi-core SoC for near-sensor data analytics: Mr. Wolf. Optimized implementation of this algorithm on Mr. Wolf platform leads to detecting a seizure with an energy budget of 18.4 µJ and 3.9 µJ for a compression ratio of 24x using non-linear SVM classifier and the dual linear SVM based classification method, respectively.

Simulation of effects of the electrode structure and material in the density measuring system of the peripheral nerve based on micro-electrical impedance tomography.

The electrode structure in micro-electrical impedance tomography (MEIT) highly influences the measurement sensitivity and therefore the reconstructed image quality. Hence, optimizing the electrode structure leads to the improvement of image quality in the reconstruction procedure. Although there have been many investigations on electrical impedance tomography (EIT) electrodes, there is no comprehensive study on their influence on images of the peripheral nerve. In this paper, we present a simulation method to study the effects of the electrode structure in the density measurement system of the peripheral nerve based on MEIT. The influence of the electrode structure such as dimensions, material and the number of electrodes and also the recognition feature of different radii of fascicle and different locations of fascicles has been studied. Data were reconstructed from the real and imaginary parts of complex conductivity data, respectively. It has been shown that the material of the electrodes had no effect on the reconstructed images, while the dimensions of the electrodes significantly affected the image sensitivity and thus the image quality. An increase in the number of electrodes increased the amount of data and information content. However, as the number of electrodes increased due to the given perimeter of the peripheral nerve, the area of the electrodes was reduced. This reduction affects the reconstructed image quality. The real and imaginary parts of the data were separately reconstructed for each case. Although, in real EIT systems, the reconstructed images using the real part of the signal have a better signal-to-noise ratio (SNR), this study proved that for a density measuring system of the peripheral nerve, the reconstructed images using the imaginary part of the signal had better quality. This simulation study proposes the effects of the electrode size and material and obtained spatial resolution that was high enough to reconstruct fascicles in a peripheral nerve.

Assessment on selectivity of multi-contact cuff electrode for recording peripheral nerve signals using Fitzhugh-Nagumo model of nerve excitation.

BACKGROUND: Nerve cuff electrodes provide a safe technique for recording nerve signals. Defining a more realized modeling to investigate the selectivity of a cuff electrode in recording from peripheral nervous system is an interesting field of research. METHODS: A four-contact cuff electrode was modeled to evaluate selective recording from a peripheral nerve. Fitzhugh-Nagumo equations were used to model the electromagnetic fields generated by active nerves and electrodes and the ``selectivity index'' used to quantify the selective property of the cuff electrode. RESULTS: The action potentials amplitude and impulse velocity generated by Fitzhugh-Nagumo model are similar to real-life nerve measurements according to the literature. The electrical field distribution caused by the impulse propagation along a specific nerve was the maximum near the corresponding contact. Also, the selectivity was increased with increasing the distance between the active sources and the number of contacts. CONCLUSION: The results of this research showed that Fitzhugh-Nagumo equations could model the nerve excitation accurately and could be used in computer simulation for studying nervous systems. Also, using these equations indicated that multi-contact cuff electrodes could be used in recording peripheral nerve signals in order to discriminate active fascicles in a nerve bundle.

Effect of contacts configuration and location on selective stimulation of cuff electrode.

BACKGROUND: Cuff electrodes have been widely used chronically in different clinical applications. Advancements have been made in selective stimulation by using multi-contact cuff electrodes. Steering anodic current is a strategy to increase selectivity by reshaping and localizing electric fields. There are two configurations for contacts to be implemented in cuff, monopolar and tripolar. A cuff electrode with tripolar configuration can restrict the activation to a more localized region within a nerve trunk compared to a cuff with monopolar configuration and improve the selectivity. Anode contacts in tripolar configuration can be made in two structures, "ring" and "dot". OBJECTIVE: In this study, the stimulation capabilities of these two structures were evaluated. METHODS: The recruitment properties and the selectivity of stimulation were examined by measuring the electric potential produced by stimulation currents. RESULTS: The results of the present study indicated that using dot configuration, the current needed to stimulate fascicles in tripolar topologies would be reduced by 10%. It was also shown that stimulation threshold was increased by moving anode contacts inward the cuff. On the other hand, stimulation threshold was decreased by moving the anode contacts outward the cuff which would decrease selectivity, too. CONCLUSIONS: We conclude that dot configuration is a better choice for stimulation. Also, a cuff inward placement of 10% relative to the cuff length was near optimal.

An autonomous real-time single-channel detection of absence seizures in WAG/Rij rats.

This paper presents a real-time, completely automated and patient independent algorithm for detection of absence seizures in WAG/Rij rats as a valid animal model of human absence epilepsy. Single-channel EEG recordings containing totally 488 seizures from 8 WAG/Rij rats were analyzed using the real-time SWD detection algorithm. The proposed algorithms based on the variation of wavelet power to the background power in two specific frequency bands whose spectral power are highly correlated with SWDs. The wavelet powers of two specific frequency bands are calculated with a pattern-adapted mother wavelet and compared with an adaptive ratio of background power of each frequency band. The results indicate used algorithm is able to detect the whole 488 seizures within less than 1 s with sensitivity of 100%. The average precision for 1200, 1400 and 1600 point of window size was 95.2%, 98.3% and 99.17%, respectively. The present algorithm, with its high sensitivity and specificity, could be used for further studies of absence seizures in humans and rats and could be implemented as real-time system for closed loop deep brain stimulation systems.

Investigation of ECG Changes in Absence Epilepsy on WAG/Rij Rats.

INTRODUCTION: Seizures are symptoms associated with abnormal electrical activity in electroencephalogram (EEG). The present study was designed to determine the effect of absence seizure on heart rate (HR) changes in electrocardiogram (ECG). METHODS: HR alterations were recorded simultaneous with spike and wave discharges (SWD) by EEG in 6 WAG/Rij rats as a well characterized and validated genetic animal epilepsy model. Moreover, 6 control rats were used to distinguish the differences of HR changes between various groups. Electrodes were placed on the skull and under the chest skin, minimizing time delay and signal attenuation. HR was calculated by an adaptable algorithm based on continues wavelet transform (CWT) particular for this study. Three main features of HR; minimum, maximum, and mean values were estimated for pre-ictal and ictal intervals for all seizures. RESULTS: ECG beats detected with sensitivity of 99.9% and positive predictability of 99.8% based on CWT. HR deceleration was found in 86% of the seizures. There were statistically significant (P<0.001) reductions of these values from pre-ictal to ictal intervals. Interictal HR acceleration and ictal deceleration were the major feature of alterations and in 23% of seizures, this decrease had priority to the onsets. DISCUSSION: These findings may lead to design a seizure alarm system based on HR and to obtain new insights about sudden unexpected death in epilepsy (SUDEP) phenomenon and side-effects of antiepileptic drugs (AED).