A 1.83 μJ/Classification, 8-Channel, Patient-Specific Epileptic Seizure Classification SoC Using a Non-Linear Support Vector Machine

A Reconfigurable Near-Sensor Processor for Anomaly Detection in Limb Prostheses

This paper presents a reconfigurable near-sensor anomaly detection processor to real-time monitor the potential anomalous behaviors of amputees with limb prostheses. The processor is low-power, low-latency, and suitable for equipment on the prostheses and comprises a reconfigurable Variational Autoencoder (VAE), a scalable Self-Organizing Map (SOM) Array, and a window-size-adjustable Markov Chain, which can implement an integrated miniaturized anomaly detection system. With the reconfigurable VAE, the proposed processor can support up to 64 sensor sampling channels programmable by global configuration, which can meet the anomaly detection requirements in different scenarios. A scalable SOM array allows for the selection of different sizes based on the complexity of the data. Unlike traditional time accumulation-based anomaly detection methods, the Markov Chain is utilized to detect time-series-based anomalous data. The processor is designed and fabricated in a UMC 40-nm LP technology with a core area of 1.49 mm 2 and a power consumption of 1.81 mW. It achieves real-time detection performance with 0.933 average F1 Score for the FSP dataset within 24.22 μs, and 0.956 average F1 Score for the SFDLA-12 dataset within 30.48 μs. The energy dissipation of detection for each input feature is 43.84 nJ with the FSP dataset, and 55.17 nJ with the SFDLA-12 dataset. Compared with ARM Cortex-M4 and ARM Cortex-M33 microcontrollers, the processor achieves energy and area efficiency improvements ranging from 257 ×, 193 × and 11 ×, 8 ×, respectively.

An epilepsy classification based on FFT and fully convolutional neural network nested LSTM

Background and objective

Epilepsy, which is associated with neuronal damage and functional decline, typically presents patients with numerous challenges in their daily lives. An early diagnosis plays a crucial role in managing the condition and alleviating the patients' suffering. Electroencephalogram (EEG)-based approaches are commonly employed for diagnosing epilepsy due to their effectiveness and non-invasiveness. In this study, a classification method is proposed that use fast Fourier Transform (FFT) extraction in conjunction with convolutional neural networks (CNN) and long short-term memory (LSTM) models.

Methods

Most methods use traditional frameworks to classify epilepsy, we propose a new approach to this problem by extracting features from the source data and then feeding them into a network for training and recognition. It preprocesses the source data into training and validation data and then uses CNN and LSTM to classify the style of the data.

Results

Upon analyzing a public test dataset, the top-performing features in the fully CNN nested LSTM model for epilepsy classification are FFT features among three types of features. Notably, all conducted experiments yielded high accuracy rates, with values exceeding 96% for accuracy, 93% for sensitivity, and 96% for specificity. These results are further benchmarked against current methodologies, showcasing consistent and robust performance across all trials. Our approach consistently achieves an accuracy rate surpassing 97.00%, with values ranging from 97.95 to 99.83% in individual experiments. Particularly noteworthy is the superior accuracy of our method in the AB versus (vs.) CDE comparison, registering at 99.06%.

Conclusion

Our method exhibits precise classification abilities distinguishing between epileptic and non-epileptic individuals, irrespective of whether the participant's eyes are closed or open. Furthermore, our technique shows remarkable performance in effectively categorizing epilepsy type, distinguishing between epileptic ictal and interictal states versus non-epileptic conditions. An inherent advantage of our automated classification approach is its capability to disregard EEG data acquired during states of eye closure or eye-opening. Such innovation holds promise for real-world applications, potentially aiding medical professionals in diagnosing epilepsy more efficiently.

Reducing False Alarms in Wearable Seizure Detection With EEGformer: A Compact Transformer Model for MCUs

The long-term, continuous analysis of electroencephalography (EEG) signals on wearable devices to automatically detect seizures in epileptic patients is a high-potential application field for deep neural networks, and specifically for transformers, which are highly suited for end-to-end time series processing without handcrafted feature extraction. In this work, we propose a small-scale transformer detector, the EEGformer, compatible with unobtrusive acquisition setups that use only the temporal channels. EEGformer is the result of a hardware-oriented design exploration, aiming for efficient execution on tiny low-power micro-controller units (MCUs) and low latency and false alarm rate to increase patient and caregiver acceptance.Tests conducted on the CHB-MIT dataset show a 20% reduction of the onset detection latency with respect to the state-of-the-art model for temporal acquisition, with a competitive 73% seizure detection probability and 0.15 false-positive-per-hour (FP/h). Further investigations on a novel and challenging scalp EEG dataset result in the successful detection of 88% of the annotated seizure events, with 0.45 FP/h.We evaluate the deployment of the EEGformer on three commercial low-power computing platforms: the single-core Apollo4 MCU and the GAP8 and GAP9 parallel MCUs. The most efficient implementation (on GAP9) results in as low as 13.7 ms and 0.31 mJ per inference, demonstrating the feasibility of deploying the EEGformer on wearable seizure detection systems with reduced channel count and multi-day battery duration.

Seizure-Cluster-Inception CNN (SciCNN): A Patient-Independent Epilepsy Tracking SoC With 0-Shot-Retraining

Epilepsy tracking System-on-Chips (SoC) usually perform patient-specific classification to deal with the patient-to-patient seizure pattern variation from a surface electroencephalogram (EEG). However, the patient-specific classifier training requires the EEG signals from the target patients a priori, which involves costly and time-consuming hospitalization for the inpatient data collection. To address this issue, this paper presents a patient-independent epilepsy tracking SoC that is trained with pre-existing databases and can be directly deployed to the target patients without collecting their data and performing cumbersome patient-specific training beforehand. The proposed SoC adopts a Seizure-Cluster-Inception Convolutional Neural Network (SciCNN) Neural Processor (SNP) to reduce SRAM access rate by 179.05× with the Kernel-Wise Pipeline (KWP). The 22-Ch. SoC achieves event-based sensitivity of 90.3%/90.4%/83.3% and specificity of 93.6%/95.7%/88.6% on unseen patients from CHB-MIT database/EU database/local hospital patient, respectively.

NET-TEN: a silicon neuromorphic network for low-latency detection of seizures in local field potentials

Objective. Therapeutic intervention in neurological disorders still relies heavily on pharmacological solutions, while the treatment of patients with drug resistance remains an unresolved issue. This is particularly true for patients with epilepsy, 30% of whom are refractory to medications. Implantable devices for chronic recording and electrical modulation of brain activity have proved a viable alternative in such cases. To operate, the device should detect the relevant electrographic biomarkers from local field potentials (LFPs) and determine the right time for stimulation. To enable timely interventions, the ideal device should attain biomarker detection with low latency while operating under low power consumption to prolong battery life.Approach. Here we introduce a fully-analog neuromorphic device implemented in CMOS technology for analyzing LFP signals in anin vitromodel of acute ictogenesis. Neuromorphic networks have progressively gained a reputation as low-latency low-power computing systems, which makes them a promising candidate as processing core of next-generation implantable neural interfaces.Main results. The developed system can detect ictal and interictal events with ms-latency and with high precision, consuming on average 3.50 nW during the task.Significance. The work presented in this paper paves the way to a new generation of brain implantable devices for personalized closed-loop stimulation for epilepsy treatment.

A Shallow Autoencoder Framework for Epileptic Seizure Detection in EEG Signals

This paper presents a trainable hybrid approach involving a shallow autoencoder (AE) and a conventional classifier for epileptic seizure detection. The signal segments of a channel of electroencephalogram (EEG) (EEG epochs) are classified as epileptic and non-epileptic by employing its encoded AE representation as a feature vector. Analysis on a single channel-basis and the low computational complexity of the algorithm allow its use in body sensor networks and wearable devices using one or few EEG channels for wearing comfort. This enables the extended diagnosis and monitoring of epileptic patients at home. The encoded representation of EEG signal segments is obtained based on training the shallow AE to minimize the signal reconstruction error. Extensive experimentation with classifiers has led us to propose two versions of our hybrid method: (a) one yielding the best classification performance compared to the reported methods using the k-nearest neighbor (kNN) classifier and (b) the second with a hardware-friendly architecture and yet with the best classification performance compared to other reported methods in this category using a support-vector machine (SVM) classifier. The algorithm is evaluated on the Children's Hospital Boston, Massachusetts Institute of Technology (CHB-MIT), and University of Bonn EEG datasets. The proposed method achieves 98.85% accuracy, 99.29% sensitivity, and 98.86% specificity on the CHB-MIT dataset using the kNN classifier. The best figures using the SVM classifier for accuracy, sensitivity, and specificity are 99.19%, 96.10%, and 99.19%, respectively. Our experiments establish the superiority of using an AE approach with a shallow architecture to generate a low-dimensionality yet effective EEG signal representation capable of high-performance abnormal seizure activity detection at a single-channel EEG level and with a fine granularity of 1 s EEG epochs.

Seizure Detection and Prediction by Parallel Memristive Convolutional Neural Networks

During the past two decades, epileptic seizure detection and prediction algorithms have evolved rapidly. However, despite significant performance improvements, their hardware implementation using conventional technologies, such as Complementary Metal-Oxide-Semiconductor (CMOS), in power and area-constrained settings remains a challenging task; especially when many recording channels are used. In this paper, we propose a novel low-latency parallel Convolutional Neural Network (CNN) architecture that has between 2-2,800x fewer network parameters compared to State-Of-The-Art (SOTA) CNN architectures and achieves 5-fold cross validation accuracy of 99.84% for epileptic seizure detection, and 99.01% and 97.54% for epileptic seizure prediction, when evaluated using the University of Bonn Electroencephalogram (EEG), CHB-MIT and SWEC-ETHZ seizure datasets, respectively. We subsequently implement our network onto analog crossbar arrays comprising Resistive Random-Access Memory (RRAM) devices, and provide a comprehensive benchmark by simulating, laying out, and determining hardware requirements of the CNN component of our system. We parallelize the execution of convolution layer kernels on separate analog crossbars to enable 2 orders of magnitude reduction in latency compared to SOTA hybrid Memristive-CMOS Deep Learning (DL) accelerators. Furthermore, we investigate the effects of non-idealities on our system and investigate Quantization Aware Training (QAT) to mitigate the performance degradation due to low Analog-to-Digital Converter (ADC)/Digital-to-Analog Converter (DAC) resolution. Finally, we propose a stuck weight offsetting methodology to mitigate performance degradation due to stuck [Formula: see text] memristor weights, recovering up to 32% accuracy, without requiring retraining. The CNN component of our platform is estimated to consume approximately 2.791 W of power while occupying an area of 31.255 mm² in a 22 nm FDSOI CMOS process.

Group-Chopping: An 8-Channel, 0.04% Gain Mismatch, 2.1 µW 0.017 mm2 Instrumentation Amplifier for Bio-Potential Recording

An 8-channel AFE with a group-chopping instrumentation amplifier (GCIA) is proposed for bio-potential recording applications. The group-chopping technique cascades chopper switches to progressively swap channels and dynamically removes gain mismatch among all channels. An 8-phase non-overlapping clocking scheme is developed and achieves excellent between-channel gain mismatch characteristics. The dynamic offsets among all channels are mitigated by the GCIA as well. The GCIA is the first work that minimizes the gain mismatch across more than two channels. With the help of the group-chopping, combined with an area-efficient open-loop structure, the GCIA shows <0.04% between-channel gain mismatch, the lowest mismatch reported to date. The chip is fabricated in 0.18µm 1P6M CMOS, occupies only 0.017 mm²/Ch., consumes 2.1 μW/Ch. under 0.5 V supply and achieves an NEF of 2.1.

Seizures initiate in zones of relative hyperexcitation in a zebrafish epilepsy model

Seizures are thought to arise from an imbalance of excitatory and inhibitory neuronal activity. While most classical studies suggest excessive excitatory neural activity plays a generative role, some recent findings challenge this view and instead argue that excessive activity in inhibitory neurons initiates seizures. We investigated this question of imbalance in a zebrafish seizure model with two-photon imaging of excitatory and inhibitory neuronal activity throughout the brain using a nuclear-localized calcium sensor. We found that seizures consistently initiated in circumscribed zones of the midbrain before propagating to other brain regions. Excitatory neurons were both more prevalent and more likely to be recruited than inhibitory neurons in initiation as compared with propagation zones. These findings support a mechanistic picture whereby seizures initiate in a region of hyper-excitation, then propagate more broadly once inhibitory restraint in the surround is overcome.

Comparison of Empirical Mode Decomposition, Wavelets, and Different Machine Learning Approaches for Patient-Specific Seizure Detection Using Signal-Derived Empirical Dictionary Approach

Analysis of long-term multichannel EEG signals for automatic seizure detection is an active area of research that has seen application of methods from different domains of signal processing and machine learning. The majority of approaches developed in this context consist of extraction of hand-crafted features that are used to train a classifier for eventual seizure detection. Approaches that are data-driven, do not use hand-crafted features, and use small amounts of patients' historical EEG data for classifier training are few in number. The approach presented in this paper falls in the latter category, and is based on a signal-derived empirical dictionary approach, which utilizes empirical mode decomposition (EMD) and discrete wavelet transform (DWT) based dictionaries learned using a framework inspired by traditional methods of dictionary learning. Three features associated with traditional dictionary learning approaches, namely projection coefficients, coefficient vector and reconstruction error, are extracted from both EMD and DWT based dictionaries for automated seizure detection. This is the first time these features have been applied for automatic seizure detection using an empirical dictionary approach. Small amounts of patients' historical multi-channel EEG data are used for classifier training, and multiple classifiers are used for seizure detection using newer data. In addition, the seizure detection results are validated using 5-fold cross-validation to rule out any bias in the results. The CHB-MIT benchmark database containing long-term EEG recordings of pediatric patients is used for validation of the approach, and seizure detection performance comparable to the state-of-the-art is obtained. Seizure detection is performed using five classifiers, thereby allowing a comparison of the dictionary approaches, features extracted, and classifiers used. The best seizure detection performance is obtained using EMD based dictionary and reconstruction error feature and support vector machine classifier, with accuracy, sensitivity and specificity values of 88.2, 90.3, and 88.1%, respectively. Comparison is also made with other recent studies using the same database. The methodology presented in this paper is shown to be computationally efficient and robust for patient-specific automatic seizure detection. A data-driven methodology utilizing a small amount of patients' historical data is hence demonstrated as a practical solution for automatic seizure detection.

Closed-Loop Neural Prostheses With On-Chip Intelligence: A Review and a Low-Latency Machine Learning Model for Brain State Detection

The application of closed-loop approaches in systems neuroscience and therapeutic stimulation holds great promise for revolutionizing our understanding of the brain and for developing novel neuromodulation therapies to restore lost functions. Neural prostheses capable of multi-channel neural recording, on-site signal processing, rapid symptom detection, and closed-loop stimulation are critical to enabling such novel treatments. However, the existing closed-loop neuromodulation devices are too simplistic and lack sufficient on-chip processing and intelligence. In this paper, we first discuss both commercial and investigational closed-loop neuromodulation devices for brain disorders. Next, we review state-of-the-art neural prostheses with on-chip machine learning, focusing on application-specific integrated circuits (ASIC). System requirements, performance and hardware comparisons, design trade-offs, and hardware optimization techniques are discussed. To facilitate a fair comparison and guide design choices among various on-chip classifiers, we propose a new energy-area (E-A) efficiency figure of merit that evaluates hardware efficiency and multi-channel scalability. Finally, we present several techniques to improve the key design metrics of tree-based on-chip classifiers, both in the context of ensemble methods and oblique structures. A novel Depth-Variant Tree Ensemble (DVTE) is proposed to reduce processing latency (e.g., by 2.5× on seizure detection task). We further develop a cost-aware learning approach to jointly optimize the power and latency metrics. We show that algorithm-hardware co-design enables the energy- and memory-optimized design of tree-based models, while preserving a high accuracy and low latency. Furthermore, we show that our proposed tree-based models feature a highly interpretable decision process that is essential for safety-critical applications such as closed-loop stimulation.

Multi-Level Optimization of an Ultra-Low Power BrainWave System for Non-Convulsive Seizure Detection

We present a systematic evaluation and optimization of a complex bio-medical signal processing application on the BrainWave prototype system, targeted towards ambulatory EEG monitoring within a tiny power budget of 1 mW. The considered BrainWave processor is completely programmable, while maintaining energy-efficiency by means of a Coarse-Grained Reconfigurable Array (CGRA). This is demonstrated through the mapping and evaluation of a state-of-the-art non-convulsive epileptic seizure detection algorithm, while ensuring real-time operation and seizure detection accuracy. Exploiting the CGRA leads to an energy reduction of 73.1%, compared to a highly tuned software implementation (SW-only). A total of 9 complex kernels were benchmarked on the CGRA, resulting in an average 4.7 × speedup and average 4.4 × energy savings over highly tuned SW-only implementations. The BrainWave processor is implemented in 28-nm FDSOI technology with 80 kB of Foundry-provided SRAM. By exploiting near-threshold computing for the logic and voltage-stacking to minimize on-chip voltage-conversion overhead, additional 15.2% and 19.5% energy savings are obtained, respectively. At the Minimum-Energy-Point (MEP) (223 μW, 8 MHz) we report a measured state-of-the-art 90.6% system conversion efficiency, while executing the epileptic seizure detection in real-time.

A 10.13µJ/Classification 2-Channel Deep Neural Network Based SoC for Negative Emotion Outburst Detection of Autistic Children

An electroencephalogram (EEG)-based non-invasive 2-channel neuro-feedback SoC is presented to predict and report negative emotion outbursts (NEOB) of Autistic patients. The SoC incorporates area-and-power efficient dual-channel Analog Front-End (AFE), and a deep neural network (DNN) emotion classification processor. The classification processor utilizes only the two-feature vector per channel to minimize the area and overfitting problems. The 4-layers customized DNN classification processor is integrated on-sensor to predict the NEOB. The AFE comprises two entirely shared EEG channels using sampling capacitors to reduce the area by 30%. Moreover, it achieves an overall integrated input-referred noise, NEF, and crosstalk of 0.55 µV_RMS, 2.71, and -79 dB, respectively. The 16 mm² SoC is implemented in 0.18 um 1P6M, CMOS process and consumes 10.13 μJ/classification for 2 channel operation while achieving an average accuracy of >85% on multiple emotion databases and real-time testing.

Video-Based Detection of Generalized Tonic-Clonic Seizures Using Deep Learning

Timely detection of seizures is crucial to implement optimal interventions, and may help reduce the risk of sudden unexpected death in epilepsy (SUDEP) in patients with generalized tonic-clonic seizures (GTCSs). While video-based automated seizure detection systems may be able to provide seizure alarms in both in-hospital and at-home settings, earlier studies have primarily employed hand-designed features for such a task. In contrast, deep learning-based approaches do not rely on prior feature selection and have demonstrated outstanding performance in many data classification tasks. Despite these advantages, neural network-based video classification has rarely been attempted for seizure detection. We here assessed the feasibility and efficacy of automated GTCSs detection from videos using deep learning. We retrospectively identified 76 GTCS videos from 37 participants who underwent long-term video-EEG monitoring (LTM) along with interictal video data from the same patients, and 10 full-night seizure-free recordings from additional patients. Using a leave-one-subject-out cross-validation approach (LOSO-CV), we evaluated the performance to detect seizures based on individual video frames (convolutional neural networks, CNNs) or video sequences [CNN+long short-term memory (LSTM) networks]. CNN+LSTM networks based on video sequences outperformed GTCS detection based on individual frames yielding a mean sensitivity of 88% and mean specificity of 92% across patients. The average detection latency after presumed clinical seizure onset was 22 seconds. Detection performance increased as a function of training dataset size. Collectively, we demonstrated that automated video-based GTCS detection with deep learning is feasible and efficacious. Deep learning-based methods may be able to overcome some limitations associated with traditional approaches using hand-crafted features, serve as a benchmark for future methods and analyses, and improve further with larger datasets.

Energy-Efficient Integrated Circuit Solutions Toward Miniaturized Closed-Loop Neural Interface Systems

Miniaturized implantable devices play a crucial role in neural interfaces by monitoring and modulating neural activities on the peripheral and central nervous systems. Research efforts toward a compact wireless closed-loop system stimulating the nerve automatically according to the user's condition have been maintained. These systems have several advantages over open-loop stimulation systems such as reduction in both power consumption and side effects of continuous stimulation. Furthermore, a compact and wireless device consuming low energy alleviates foreign body reactions and risk of frequent surgical operations. Unfortunately, however, the miniaturized closed-loop neural interface system induces several hardware design challenges such as neural activity recording with severe stimulation artifact, real-time stimulation artifact removal, and energy-efficient wireless power delivery. Here, we will review recent approaches toward the miniaturized closed-loop neural interface system with integrated circuit (IC) techniques.

An Active Concentric Electrode for Concurrent EEG Recording and Body-Coupled Communication (BCC) Data Transmission

This paper presents a wearable active concentric electrode for concurrent EEG monitoring and Body-Coupled Communication (BCC) data transmission. A three-layer concentric electrode eliminates the usage of wires. A common mode averaging unit (CMAU) is proposed to cancel not only the continuous common-mode interference (CMI) but also the instantaneous CMI of up to 51V_pp. The localized potential matching technique removes the ground electrode. An open-loop programmable gain amplifier (OPPGA) with the pseudo-resistor-based RC-divider block is presented to save the silicon area. The presented work is the first reported so far to achieve the concurrent EEG signal recording and BCC-based data transmission. The proposed chip achieves 100 dB CMRR and 110 dB PSRR, occupies 0.044 mm², and consumes 7.4 μW with an input-referred noise density of 26 nV/√Hz.

From Seizure Detection to Smart and Fully Embedded Seizure Prediction Engine: A Review

Recent review papers have investigated seizure prediction, creating the possibility of preempting epileptic seizures. Correct seizure prediction can significantly improve the standard of living for the majority of epileptic patients, as the unpredictability of seizures is a major concern for them. Today, the development of algorithms, particularly in the field of machine learning, enables reliable and accurate seizure prediction using desktop computers. However, despite extensive research effort being devoted to developing seizure detection integrated circuits (ICs), dedicated seizure prediction ICs have not been developed yet. We believe that interdisciplinary study of system architecture, analog and digital ICs, and machine learning algorithms can promote the translation of scientific theory to a more realistic intelligent, integrated, and low-power system that can truly improve the standard of living for epileptic patients. This review explores topics ranging from signal acquisition analog circuits to classification algorithms and dedicated digital signal processing circuits for detection and prediction purposes, to provide a comprehensive and useful guideline for the construction, implementation and optimization of wearable and integrated smart seizure prediction systems.

A Real-Time Depth of Anesthesia Monitoring System Based on Deep Neural Network With Large EDO Tolerant EEG Analog Front-End

In this article, we present a real-time electroencephalogram (EEG) based depth of anesthesia (DoA) monitoring system in conjunction with a deep learning framework, AnesNET. An EEG analog front-end (AFE) that can compensate ±380-mV electrode DC offset using a coarse digital DC servo loop is implemented in the proposed system. The EEG-based MAC, EEGMAC, is introduced as a novel index to accurately predict the DoA, which is designed for applying to patients anesthetized by both volatile and intravenous agents. The proposed deep learning protocol consists of four layers of convolutional neural network and two dense layers. In addition, we optimize the complexity of the deep neural network (DNN) to operate on a microcomputer such as the Raspberry Pi 3, realizing a cost-effective small-size DoA monitoring system. Fabricated in 110-nm CMOS, the prototype AFE consumes 4.33 μW per channel and has the input-referred noise of 0.29 μVrms from 0.5 to 100 Hz with the noise efficiency factor of 2.2. The proposed DNN was evaluated with pre-recorded EEG data from 374 subjects administrated by inhalational anesthetics under surgery, achieving an average squared and absolute errors of 0.048 and 0.05, respectively. The EEGMAC with subjects anesthetized by an intravenous agent also showed a good agreement with the bispectral index value, confirming the proposed DoA index is applicable to both anesthetics. The implemented monitoring system with the Raspberry Pi 3 estimates the EEGMAC within 20 ms, which is about thousand-fold faster than the BIS estimation in literature.

An On-Chip Processor for Chronic Neurological Disorders Assistance Using Negative Affectivity Classification

Chronic neurological disorders (CND's) are lifelong diseases and cannot be eradicated, but their severe effects can be alleviated by early preemptive measures. CND's, such as Alzheimer's, Autism Spectrum Disorder (ASD), and Amyotrophic Lateral Sclerosis (ALS), are the chronic ailment of the central nervous system that causes the degradation of emotional and cognitive abilities. Long term continuous monitoring with neuro-feedback of human emotions for patients with CND's is crucial in mitigating its harmful effect. This paper presents hardware efficient and dedicated human emotion classification processor for CND's. Scalp EEG is used for the emotion's classification using the valence and arousal scales. A linear support vector machine classifier is used with power spectral density, logarithmic interhemispheric power spectral ratio, and the interhemispheric power spectral difference of eight EEG channel locations suitable for a wearable non-invasive classification system. A look-up-table based logarithmic division unit (LDU) is to represent the division features in machine learning (ML) applications. The implemented LDU minimizes the cost of integer division by 34% for ML applications. The implemented emotion's classification processor achieved an accuracy of 72.96% and 73.14%, respectively, for the valence and arousal classification on multiple publicly available datasets. The 2 x 3mm² processor is fabricated using a 0.18 μm 1P6M CMOS process with power and energy utilization of 2.04 mW and 16 μJ/classification, respectively, for 8-channel operation.

Charge-Redistribution Based Quadratic Operators for Neural Feature Extraction

This paper presents a SAR converter based mixed-signal multiplier for the feature extraction of neural signals using quadratic operators. After a thorough analysis of design principles and circuit-level aspects, the proposed architecture is explored for the implementation of two quadratic operators often used for the characterization of neural activity, the moving average energy (MAE) operator and the nonlinear energy operator (NEO). Programmable chips for both operators have been implemented in a HV-180 nm CMOS process. Experimental results confirm their suitability for energy computation and action potential detection and the accomplished area×power performance is compared to prior art. The MAE and NEO prototypes, at a sampling rate of 30kS/s, consume 116 nW and 178 nW, respectively, and digitize both the input neural signal and the operator outcome, with no need for digital multipliers.

A TDM-Based 16-Channel AFE ASIC With Enhanced System-Level CMRR for Wearable EEG Recording With Dry Electrodes

A multi-channel analog front-end (AFE) ASIC for wearable EEG recording application is presented in this article. Two techniques, namely chopping stabilization (CS) and time-division-multiplexing (TDM) are combined in a unified manner to improve the input-referred noise and the system level common-mode rejection ratio (CMRR) for multi-channel AFE. With the proposed TDM/CS structure, multiple channels can share single second-stage amplifier for significant reduction in chip size and power consumption. Dual feedback loops for input impedance boosting as well as electrode offset cancellation are incorporated in the system. Implemented in a 0.18-μm CMOS process, the AFE consumes 24 μW under 1 V supply. The input referred noise is 0.63 μV_rms in 0.5 Hz-100 Hz and the input impedance is boosted to 560 MΩ at 50 Hz. The measured amplifier intrinsic CMRR and system-level AFE CMRR are 89 dB and 82 dB, respectively.

A Noninvasive Glucose Monitoring SoC Based on Single Wavelength Photoplethysmography

Conventional glucose monitoring methods for the growing numbers of diabetic patients around the world are invasive, painful, costly and, time-consuming. Complications aroused due to the abnormal blood sugar levels in diabetic patients have created the necessity for continuous noninvasive glucose monitoring. This article presents a wearable system for glucose monitoring based on a single wavelength near-infrared (NIR) Photoplethysmography (PPG) combined with machine-learning regression (MLR). The PPG readout circuit consists of a switched capacitor Transimpedance amplifier with 1 MΩ gain and a 10-Hz switched capacitor LPF. It allows a DC bias current rejection up to 20 μA with an input-referred current noise of 7.3 pA/√Hz. The proposed digital processor eliminates motion artifacts, and baseline drifts from PPG signal, extracts six distinct features and finally predicts the blood glucose level using Support Vector Regression with Fine Gaussian kernel (FGSVR) MLR. A novel piece-wise linear (PWL) approach for the exponential function is proposed to realize the FGSVR on-chip. The overall system is implemented using a 180 nm CMOS process with a chip area of 4.0 mm² while consuming 1.62 mW. The glucose measurements are performed for 200 subjects with R² of 0.937. The proposed system accurately predicts the sugar level with a mean absolute relative difference (mARD) of 7.62%.

An Energy-Efficient CMOS Dual-Mode Array Architecture for High-Density ECoG-Based Brain-Machine Interfaces

This article presents an energy-efficient electrocorticography (ECoG) array architecture for fully-implantable brain machine interface systems. A novel dual-mode analog signal processing method is introduced that extracts neural features from high- γ band (80-160 Hz) at the early stages of signal acquisition. Initially, brain activity across the full-spectrum is momentarily observed to compute the feature weights in the digital back-end during full-band mode operation. Subsequently, these weights are fed back to the front-end and the system reverts to base-band mode to perform feature extraction. This approach utilizes a distinct optimized signal pathway based on power envelope extraction, resulting in 1.72× power reduction in the analog blocks and up to 50× potential power savings for digitization and processing (implemented off-chip in this article). A prototype incorporating a 32-channel ultra-low power signal acquisition front-end is fabricated in 180 nm CMOS process with 0.8 V supply. This chip consumes 1.05  μW (0.205  μW for feature extraction only) power and occupies 0.245 [Formula: see text] die area per channel. The chip measurement shows better than 76.5-dB common-mode rejection ratio (CMRR), 4.09 noise efficiency factor (NEF), and 10.04 power efficiency factor (PEF). In-vivo human tests have been carried out with electroencephalography and ECoG signals to validate the performance and dual-mode operation in comparison to commercial acquisition systems.

Low-Power Hardware Implementation of a Support Vector Machine Training and Classification for Neural Seizure Detection

In this paper, a low power support vector machine (SVM) training, feature extraction, and classification algorithm are hardware implemented in a neural seizure detection application. The training algorithm used is the sequential minimal optimization (SMO) algorithm. The system is implemented on different platforms: such as field programmable gate array (FPGA), Xilinx Virtex-7 and application specific integrated circuit (ASIC) using hardware-calibrated UMC 65 nm CMOS technology. The implemented training hardware is introduced as an accelerator intellectual property (IP), especially in the case of large number of training sets, such as neural seizure detection. Feature extraction and classification blocks are implemented to achieve the best trade-off between sensitivity and power consumption. The proposed seizure detection system achieves a sensitivity around 96.77% when tested with the implemented linear kernel classifier. A power consumption evaluation is performed on both the ASIC and FPGA platforms showing that the ASIC power consumption is improved by a factor of 2X when compared with the FPGA counterpart.

A Resource-Optimized VLSI Implementation of a Patient-Specific Seizure Detection Algorithm on a Custom-Made 2.2 cm 2 Wireless Device for Ambulatory Epilepsy Diagnostics

A patient-specific epilepsy diagnostic solution in the form of a wireless wearable ambulatory device is presented. First, the design, VLSI implementation, and experimental validation of a resource-optimized machine learning algorithm for epilepsy seizure detection are described. Next, the development of a mini-PCB that integrates a low-power wireless data transceiver and a programmable processor for hosting the seizure detection algorithm is discussed. The algorithm uses only EEG signals from the frontal lobe electrodes while yielding a seizure detection sensitivity and specificity competitive to the standard full EEG systems. The experimental validation of the algorithm VLSI implementation proves the possibility of conducting accurate seizure detection using quickly-mountable dry-electrode headsets without the need for uncomfortable/painful through-hair electrodes or adhesive gels. Details of design and optimization of the algorithm, the VLSI implementation, and the mini-PCB development are presented and resource optimization techniques are discussed. The optimized implementation is uploaded on a low-power Microsemi Igloo FPGA, requires 1237 logic elements, consumes 110 μW dynamic power, and yields a minimum detection latency of 10.2 μs. The measurement results from the FPGA implementation on data from 23 patients (198 seizures in total) shows a seizure detection sensitivity and specificity of 92.5% and 80.1%, respectively. Comparison to the state of the art is presented from system integration, the VLSI implementation, and the wireless communication perspectives.

A 9.2-g Fully-Flexible Wireless Ambulatory EEG Monitoring and Diagnostics Headband With Analog Motion Artifact Detection and Compensation

An 8-channel wearable wireless device for ambulatory surface EEG monitoring and analysis is presented. The entire multi-channel recording, quantization, and motion artifact removal circuitries are implemented on a 4-layer polyimide flexible substrate. The recording electrodes and active shielding are also integrated on the same substrate, yielding the smallest form factor compared to the state of the art. Thanks to the dry non-contact electrodes, the system is quickly mountable with minimal assistance required, making it an ideal ambulatory front- and temporal-lobe EEG monitoring device. The flexible main board is connected to a rechargeable battery on one end and to a 13 × 17 mm ² rigid board on the other end. The mini rigid board hosts a low-power programmable FPGA and a BLE 5.0 transceiver, which add diagnostic capability and wireless connectivity features to the device, respectively. Design considerations for a wearable EEG monitoring and diagnostic device are discussed in details. The theory of the novel fully-analog method for motion artifact detection and removal is described and the detailed circuit implementation is presented. The device performance in terms of voltage gain (260 V/V), bandwidth (DC-300 Hz), motion artifact removal, and wireless communication throughput (up to 1 Mbps) is experimentally validated. The entire wearable solution with the battery weighs 9.2 grams.

A Patient-Specific Single Sensor IoT-Based Wearable Fall Prediction and Detection System

Falls in older adults are a major cause of morbidity and mortality and are a key class of preventable injuries. This paper presents a patient-specific (PS) fall prediction and detection prototype system that utilizes a single tri-axial accelerometer attached to the patient's thigh to distinguish between activities of daily living (ADL) and fall events. The proposed system consists of two modes of operation: 1) fast mode for fall predication (FMFP) predicting a fall event (300-700 msec) before occurring and 2) slow mode for fall detection (SMFD) with a 1-sec latency for detecting a fall event. The nonlinear support vector machine classifier (NLSVM)-based FMFP algorithm extracts seven discriminating features for the pre-fall case to identify a fall risk event and alarm the patient. The proposed SMFD algorithm utilizes a Three-cascaded 1-sec sliding frames classification architecture with a linear regression-based offline training to identify a single and optimal threshold for each patient. Fall incidence will trigger an alarming notice to the concern healthcare providers via the Internet. Experiments are performed with 20 different subjects (age above 65 years) and a total number of 100 associated falls and ADL recordings indoors and outdoors. The accuracy of the proposed algorithms is furthermore validated via MobiFall Dataset. FMFP achieves sensitivity and specificity of 97.8% and 99.1%, respectively, while SMFD achieves sensitivity and specificity of 98.6% and 99.3%, respectively, for a total number of 600 measured falls and ADL cases from 77 subjects.

Patient-specific seizure detection in long-term EEG using signal-derived empirical mode decomposition (EMD)-based dictionary approach

OBJECTIVE

The objective of the work described in this paper is the development of a computationally efficient methodology for patient-specific automatic seizure detection in long-term multi-channel EEG recordings.

APPROACH

A novel patient-specific seizure detection approach based on a signal-derived empirical mode decomposition (EMD)-based dictionary approach is proposed. For this purpose, we use an empirical framework for EMD-based dictionary creation and learning, inspired by traditional dictionary learning methods, in which the EMD-based dictionary is learned from the multi-channel EEG data being analyzed for automatic seizure detection. We present the algorithm for dictionary creation and learning, whose purpose is to learn dictionaries with a small number of atoms. Using training signals belonging to seizure and non-seizure classes, an initial dictionary, termed as the raw dictionary, is formed. The atoms of the raw dictionary are composed of intrinsic mode functions obtained after decomposition of the training signals using the empirical mode decomposition algorithm. The raw dictionary is then trained using a learning algorithm, resulting in a substantial decrease in the number of atoms in the trained dictionary. The trained dictionary is then used for automatic seizure detection, such that coefficients of orthogonal projections of test signals against the trained dictionary form the features used for classification of test signals into seizure and non-seizure classes. Thus no hand-engineered features have to be extracted from the data as in traditional seizure detection approaches.

MAIN RESULTS

The performance of the proposed approach is validated using the CHB-MIT benchmark database, and averaged accuracy, sensitivity and specificity values of 92.9%, 94.3% and 91.5%, respectively, are obtained using support vector machine classifier and five-fold cross-validation method. These results are compared with other approaches using the same database, and the suitability of the approach for seizure detection in long-term multi-channel EEG recordings is discussed.

SIGNIFICANCE

The proposed approach describes a computationally efficient method for automatic seizure detection in long-term multi-channel EEG recordings. The method does not rely on hand-engineered features, as are required in traditional approaches. Furthermore, the approach is suitable for scenarios where the dictionary once formed and trained can be used for automatic seizure detection of newly recorded data, making the approach suitable for long-term multi-channel EEG recordings.

VLSI Design of SVM-Based Seizure Detection System With On-Chip Learning Capability

Portable automatic seizure detection system is very convenient for epilepsy patients to carry. In order to make the system on-chip trainable with high efficiency and attain high detection accuracy, this paper presents a very large scale integration (VLSI) design based on the nonlinear support vector machine (SVM). The proposed design mainly consists of a feature extraction (FE) module and an SVM module. The FE module performs the three-level Daubechies discrete wavelet transform to fit the physiological bands of the electroencephalogram (EEG) signal and extracts the time-frequency domain features reflecting the nonstationary signal properties. The SVM module integrates the modified sequential minimal optimization algorithm with the table-driven-based Gaussian kernel to enable efficient on-chip learning. The presented design is verified on an Altera Cyclone II field-programmable gate array and tested using the two publicly available EEG datasets. Experiment results show that the designed VLSI system improves the detection accuracy and training efficiency.

Automatic Detection of Epilepsy and Seizure Using Multiclass Sparse Extreme Learning Machine Classification

An automatic detection system for distinguishing normal, ictal, and interictal electroencephalogram (EEG) signals is of great help in clinical practice. This paper presents a three-class classification system based on discrete wavelet transform (DWT) and the nonlinear sparse extreme learning machine (SELM) for epilepsy and epileptic seizure detection. Three-level lifting DWT using Daubechies order 4 wavelet is introduced to decompose EEG signals into delta, theta, alpha, and beta subbands. Considering classification accuracy and computational complexity, the maximum and standard deviation values of each subband are computed to create an eight-dimensional feature vector. After comparing five multiclass SELM strategies, the one-against-one strategy with the highest accuracy is chosen for the three-class classification system. The performance of the designed three-class classification system is tested with publicly available epilepsy dataset. The results show that the system achieves high enough classification accuracy by combining the SELM and DWT and reduces training and testing time by decreasing computational complexity and feature dimension. With excellent classification performance and low computation complexity, this three-class classification system can be utilized for practical epileptic EEG detection, and it offers great potentials for portable automatic epilepsy and seizure detection system in the future hardware implementation.