Search

During the past two decades, a number of two-terminal switching devices have been demonstrated in the literature. They typically exhibit hysteric behavior in the current-to-voltage characteristics. These devices have often been also referred to as memristive devices. Their capacity to switch and exhibit electrical hysteresis has made them well-suited for applications such as data storage, in-memory computing, and in-sensor computing or in-memory sensing. The aim of this perspective paper is to is twofold. Firstly, it seeks to provide a comprehensive examination of the existing research findings in the field and engage in a critical discussion regarding the potential for the development of new non-Von-Neumann computing machines that can seamlessly integrate sensing and computing within memory units. Secondly, this paper aims to demonstrate the practical application of such an innovative approach in the realm of cancer medicine. Specifically, it explores the modern concept of employing multiple cancer markers simultaneously to enhance the efficiency of diagnostic processes in cancer medicine.

This paper addresses two key challenges toward an integrated forward error-correcting biosensor based on our previously reported self-assembled quick-response (QR) code. The first challenge involves the choice of the paper substrate for printing and self-assembling the QR code. We have compared four different substrates that includes regular printing paper, Whatman filter paper, nitrocellulose membrane and lab synthesized bacterial cellulose. We report that out of the four substrates bacterial cellulose outperforms the others in terms of probe (gold nanorods) and ink retention capability. The second challenge involves remote activation of the analyte sampling and the QR code self-assembly process. In this paper, we use light as a trigger signal and a graphite layer as a light-absorbing material. The resulting change in temperature due to infrared absorption leads to a temperature gradient that then exerts a diffusive force driving the analyte toward the regions of self-assembly. The working principle has been verified in this paper using assembled biosensor prototypes where we demonstrate higher sample flow rate due to light induced thermal gradients.

This paper extends our previous work on silver-enhancement based self-assembling structures for designing reliable, self-powered biosensors with forward error correcting (FEC) capability. At the core of the proposed approach is the integration of paper-based microfluidics with quick response (QR) codes that can be optically scanned using a smart-phone. The scanned information is first decoded to obtain the location of a web-server which further processes the self-assembled QR image to determine the concentration of target analytes. The integration substrate for the proposed FEC biosensor is polyethylene and the patterning of the QR code on the substrate has been achieved using a combination of low-cost ink-jet printing and a regular ballpoint dispensing pen. A paper-based microfluidics channel has been integrated underneath the substrate for acquiring, mixing and flowing the sample to areas on the substrate where different parts of the code can self-assemble in presence of immobilized gold nanorods. In this paper we demonstrate the proof-of-concept detection using prototypes of QR encoded FEC biosensors.

This paper presents a wireless, self-powered, affinity-based biosensor based on the integration of paper-based microfluidics with our previously reported method for self-assembling radio-frequency (RF) antennas. At the core of the proposed approach is a silver-enhancement technique that grows portions of a RF antenna in regions where target antigens hybridize with target specific affinity probes. The hybridization regions are defined by a network of nitrocellulose based microfluidic channels which implement a self-powered approach to sample the reagent and control its flow and mixing. The integration substrate for the biosensor has been constructed using polyethylene and the patterning of the antenna on the substrate has been achieved using a low-cost ink-jet printing technique. The substrate has been integrated with passive radio-frequency identification (RFID) tags to demonstrate that the resulting sensor-tag can be used for continuous monitoring in a food supply-chain where direct measurement of analytes is typically considered to be impractical. We validate the proof-of-concept operation of the proposed sensor-tag using IgG as a model analyte and using a 915 MHz Ultra-high-frequency (UHF) RFID tagging technology.

Sensitivity is an essential consideration for microfluidic paper-based analytical devices (μPADs) when these devices are used for low concentration sample detection. Very recently, ion concentration polarization (ICP)-based μPADs are emerging as novel tools for bio-sample preconcentration. In this study, we develop an enclosed paper-based microfluidic platform as a preconcentrator based on the ICP effect. This paper chip is fabricated by a Parafilm embedding technique and holds many advantages over traditional open-channel μPADs, which usually suffer from sample contamination and evaporation problems. The enclosed structure minimizes sample evaporation, reduces contamination risk, and increases the mechanical strength of the paper channel. The experiment results show that more than 100-fold concentration enhancement can be achieved in the enclosed paper device, which is comparable with most reported paper-based ICP devices. Additionally, other improved ICP performance has been observed on the enclosed device, including better concentration plug profile, increased concentration durability.

This study presents a novel high density surface electromyography (EMG) decomposition method, named as 2CFastICA, because it incorporates two key algorithms: kernel constrained FastICA and correlation constrained FastICA. The former focuses on overcoming the local convergence of FastICA without requiring the peel-off strategy used in the progressive FastICA peel-off (PFP) framework. The latter further refines the output of kernel constrained FastICA by correcting possible erroneous or missed spikes. The two constrained FastICA algorithms supplement each other to warrant the decomposition performance. The 2CFastICA method was validated using simulated surface EMG signals with different motor unit numbers and signal to noise ratios (SNRs). Two source validation was also performed by simultaneous high density surface EMG and intramuscular EMG recordings, showing a matching rate (MR) of (97.2 ± 3.5) % for 170 common motor units. In addition, a different form of two source validation was also conducted taking advantages of the high density surface EMG characteristics of patients with amyotrophic lateral sclerosis, showing a MR of (99.4 ± 0.9) % for 34 common motor units from interference and sparse datasets. Both simulation and experimental results indicate that 2CFastICA can achieve similar decomposition performance to PFP. However, the efficiency of decomposition can be greatly improved by 2CFastICA since the complex signal processing procedures associated with the peel-off strategy are not required any more. Along with this paper, we also provide the MATLAB open source code of 2CFastICA for high density surface EMG decomposition.

This paper introduces a wirelessly powered scattered neural recording wearable system that can facilitate continuous, untethered, and long-term electroencephalogram (EEG) recording. The proposed system, including 32 standalone EEG recording devices and a central controller, is incorporated in a wearable form factor. The standalone devices are sparsely distributed on the scalp, allowing for flexible placement and varying quantities to provide extensive spatial coverage and scalability. Each standalone device featuring a low-power EEG recording application-specific integrated circuit (ASIC) wirelessly receives power through a 60 MHz inductive link. The low-power ASIC design (84.6 μW) ensures sufficient wireless power reception through a small receiver (Rx) coil. The 60 MHz inductive link also serves as the data carrier for wireless communication between standalone devices and the central controller, eliminating the need for additional data antennas. All these efforts contribute to the miniaturization of standalone devices with dimensions of 12×12×5 mm 3 , enhancing device wearability. The central controller applies the pulse width modulation (PWM) scheme on the 60 MHz carrier, transmitting user commands at 4 Mbps to EEG recording ASICs. The ASIC employs a novel synchronized PWM demodulator to extract user commands, operating signal digitization and data transmission. The analog frontend (AFE) amplifies the EEG signal with a gain of 45 dB and applies band-pass filtering from 0.03 Hz to 400 Hz, with an input-referred noise (IRN) of 3.62 μV RMS . The amplified EEG signal is then digitized by a 10-bit successive approximation register (SAR) analog-to-digital converter (ADC) with a peak signal-to-noise and distortion ratio (SNDR) of 55.4 dB. The resulting EEG data is transmitted to an external software-defined radio (SDR) Rx through load-shift-keying (LSK) backscatter at 3.75 Mbps. The system's functionality is fully evaluated in human experiments.

This paper presents an innovative T/Y-maze-based wireless power transmission (WPT) system designed to monitor spatial reference memory and learning behavior in freely moving rats. The system facilitates uninterrupted optical/electrical stimulation and neural recording experiments through the integration of wireless headstages or implants in T/Y maze setups. Utilizing an array of resonators covering the entire underneath of the mazes, the wireless platform ensures scalability with various configurations. The array is designed to ensure a natural localization mechanism to localize the Tx power toward the location of the Rx coil. The system is analyzed and modeled using ANSYS HFSS software to optimize design. The primary goal was to achieve uniform wireless power delivery throughout the mazes through a comparative study of different transmitter (Tx) array configurations, such as float, series, and parallel resonators. The calculated Specific Absorption Rate (SAR) in rat tissue model equals 1.7 W/kg at the power carrier frequency of 13.56 MHz. A prototype of the proposed maze-based WPT design, featuring 8 Tx resonators, a Tx coil and power amplifier, and a headstage power harvesting unit, is successfully implemented and its performance characterized for all three resonator configurations. The implemented T maze-based WPT system has a total length of 128 cm. In the overlapping Tx resonators configuration, a homogeneity of 94% is achieved for the measured power transfer efficiency at over 30%, while continuously delivering over 60 mW for series configuration.

In this paper, a high gain amplifier with phase compensation loop is presented. A structure of parallel gate cross-coupled transistors to both ends of differential pair drain and source is designed to improves the load impedance, which obtains sufficient gain and further reduces power consumption. A novel capacitor bootstrap load circuit is proposed. The capacitor bootstrap topology is constructed by the drain source resistance of the transistor working in the cut-off region, where the gate source parasitic capacitor of the transistor is in parallel with the bootstrap capacitor rather than the existing series structure, thereby only a small bootstrap capacitor is required. By avoiding the use of large capacitors, chip area can be effectively reduced without compromising performance such as gain and bandwidth. The amplifier is fabricated using 10-μm n-type a-IGZO TFT technology. Measurement results show that the proposed amplifier achieves a voltage gain of 43.5dB and a common mode rejection ratio of 61.2dB while maintaining low power consumption. The amplifier also exhibits a -3dB bandwidth covering 0.4~2.1KHz, encompassing major bioelectric frequency bands. A real-time ECG signal was successfully captured using the fabricated TFT amplifier and gel electrodes. It has great potential in flexible sensing and acquisition applications such as electro cardiogram (ECG), electro encephalogram (EEG), pulse detection, and other wearable applications.

This paper presents a high-precision CMOS fluorescence photometry sensor using a novel lock-in amplification scheme based on switched-biasing and ping-pong auto-zeroing techniques. The CMOS sensor includes two photodiodes and a lock-in amplifier (LIA) operating at 1 kHz. The LIA comprises a differential low-noise amplifier using a novel switched-biasing ping-pong auto-zeroed scheme, an automatic phase aligner, a programmable gain amplifier, a band-pass filter, a mixer, and an output low-pass filter. The design is fabricated in 0.18-μm CMOS process, and the measurement shows that the LIA can retrieve noisy input signals with a dynamic reserve of 42 dB, while consuming only 0.7 mW from a 1.8 V supply voltage. The measured results show that the LIA can detect a wide range of incident light power from 8 nW to 24 μW. The proposed design is encapsulated in a 3D-printed housing allowing for real-time in vitro biomarker detection. This ambulatory platform uses an LED and a fiber optic to convey the excitation light to the sample and retrieve the fluorescence signal. Experiments with a beads solution diluted in PBS demonstrate that the sensor has a sensitivity of 1:100 k. Experimental results obtained in vitro with NIH3T3 mouse cells tagged with membrane dye show the ability of the prototype to detect different densities of cell culture. The portable prototype, which includes optical filters and a small 30 mm × 36 mm × 30 mm printed circuit board enclosed inside the 3D-printed housing, consumes 36.7 mW and weighs 120 g.

Most operant conditioning circuits predominantly focus on simple feedback process, few studies consider the intricacies of feedback outcomes and the uncertainty of feedback time. This paper proposes a neuromorphic circuit based on operant conditioning with addictiveness and time memory for automatic learning. The circuit is mainly composed of hunger output module, neuron module, excitement output module, memristor-based decision module, and memory and feedback generation module. In the circuit, the process of output excitement and addiction in stochastic feedback is achieved. The memory of interval between the two rewards is formed. The circuit can adapt to complex scenarios with these functions. In addition, hunger and satiety are introduced to realize the interaction between biological behavior and exploration desire, which enables the circuit to continuously reshape its memories and actions. The process of operant conditioning theory for automatic learning is accomplished. The study of operant conditioning can serve as a reference for more intelligent brain-inspired neural systems.

This paper presents an electronic skin (e-skin) taxel array readout chip in 0.18μm CMOS technology, achieving the highest reported spatial resolution of 200μm, comparable to human fingertips. A key innovation is the integration on chip of a 12×16 polyvinylidene fluoride (PVDF)-based piezoelectric sensor array with per-taxel signal conditioning frontend and spiking readout combined with local embedded neuromorphic first-order processing through Complex Receptive Fields (CRFs). Experimental results show that Spiking Neural Network (SNN)-based classification of the chip's spatiotemporal spiking output for input tactile stimuli such as texture and flutter frequency achieves excellent accuracies up to 97.1% and 99.2%, respectively. SNN-based classification of the indentation period applied to the on-chip PVDF sensors achieved 95.5% classification accuracy, despite using only a small 256-neuron SNN classifier, a low equivalent spike encoding resolution of 3-5 bits, and a sub-Nyquist 2.2kevent/s population spiking rate, a state-of-the-art power consumption of 12.33nW per-taxel, and 75μW-5mW for the entire chip is obtained. Finally, a comparison of the texture classification accuracies between two on-chip spike encoder outputs shows that the proposed neuromorphic level-crossing sampling (NLCS) architecture with a decaying threshold outperforms the conventional bipolar level-crossing sampling (LCS) architecture with fixed threshold.

This paper presents a compact and low-cost on-chip sensor and readout circuit. The sensor achieves high-resolution 5-degrees-of-freedom (DoF) tracking (x, y, z, yaw, and pitch). With the help of an external wire wound sensor, it can also achieve high-resolution 6-degrees-of-freedom (DoF) tracking (x, y, z, yaw, pitch, and roll angles). The sensor uses low-frequency magnetic fields to detect the position and orientation of instruments, providing a viable alternative to using X-rays in image-guided surgery. To measure the local magnetic field, a highly miniaturised on-chip magnetic sensor capable of sensing the magnetic field has been developed incorporating an on-chip magnetic sensor coil, analog-front end, continuous-time ΔΣ analog-to-digital converter (ADC), LVDS transmitter, bandgap reference, and voltage regulator. The microchip is fabricated using 65 nm CMOS technology and occupies an area of 1.06 mm2, the smallest reported among similar designs to the best of our knowledge. The 5-DoF system accurately navigates with a precision of 1.1 mm within the volume-of-intrest (VOI) of 15×15×15 cm 3 . The 6-DoF system achieves a navigation accuracy of 0.8 mm and an angular error of 1.1 degrees in the same VOI. These results were obtained at a 20 Hz update rate in benchtop characterisation. The prototype sensor demonstrates accurate position tracking in real-life pre-clinical in-vivo settings within the porcine lung of a live swine, achieving a reported worst-case registration accuracy of 5.8 mm.

To avoid unexpected failures of units in manufacturing systems, failure mode recognition and prognostics are critically important in prognostics health management (PHM). Most existing methods either ignored the effects of various failure modes on remaining useful lifetime (RUL) prediction or implemented failure mode recognition and RUL prediction as two independent tasks, which failed to exploit failure mode information to obtain accurate RUL prediction. In fact, RUL highly depends on failure modes because sensor signals under different failure modes usually present different degradation patterns. To address the issue, this paper proposes a joint learning model of failure mode recognition and RUL prediction for degradation processes based on multiple sensor signals. The proposed joint learning model first extracts features by considering the degradation mechanism to ensure good interpretability for degradation modeling, and then takes the extracted features as inputs to a deep neural network. By conducting failure mode recognition and RUL prediction as a collaborative task, the proposed model can fully characterize the complex relationship among the extracted features, RUL and failure modes, and outputs the recognized failure modes and the predicted RUL of units simultaneously. A case study on the degradation of aircraft gas turbine engines is presented to evaluate the proposed model performance., Note to Practitioners—: The paper aims to develop a joint learning method for failure mode recognition and RUL prediction of operating units. Specifically, the developed method addresses a challenging issue in practice, i.e., how to effectively conduct failure mode recognition and RUL prediction as a joint task based on interpretable extracted degradation features from multiple sensor signals. To implement this method in practice, four steps are included as follows: First , collect multiple sensor signals, failure time, and failure modes of historical units. Second , construct the joint learning model based on features extracted from sensor signals by considering the degradation mechanism. Third , estimate model parameters using the data of historical units. Fourth , recognize the failure mode and predict the RUL of an in-service unit. Since the proposed method is a data-driven neural network with flexible model structure that considers complex data relationships, it is expected to be applicable to many practical situations and use cases, especially for manufacturing systems with complex structures and unknown failure thresholds.

The study of neuron interactions and hardware implementations are crucial research directions in neuroscience, particularly in developing large-scale biological neural networks. The FitzHugh-Nagumo (FHN) model is a popular neuron model with highly biological plausibility, but its complexity makes it difficult to apply at scale. This paper presents a cost-saving and improved precision approximation algorithm for the digital implementation of the FHN model. By converting the computational data into floating-point numbers, the original multiplication calculations are replaced by adding the floating-point exponent part and fitting the mantissa part with piecewise linear. In the hardware implementation, shifters and adders are used, greatly reducing resource overhead. Implementing FHN neurons by this approximation calculations on FPGA reduces the normalized root mean square error (RMSE) to 3.5% of the state-of-the-art (SOTA) while maintaining a performance overhead ratio improvement of 1.09 times. Compared to implementations based on approximate multipliers, the proposed method achieves a 20% reduction in error at the cost of a 2.8% increase in overhead.This model gained additional biological properties compared to LIF while reducing the deployment scale by only 9%. Furthermore, the hardware implementation of nine coupled circular networks with eight nodes and directional diffusion was carried out to demonstrate the algorithm's effectiveness on neural networks. The error decreased to 60% compared to the single neuron of the SOTA. This hardware-friendly algorithm allows for the low-cost implementation of high-precision hardware simulation, providing a novel perspective for studying large-scale, biologically plausible neural networks.

Brain-machine interfaces (BMI) are widely adopted in neuroscience investigations and neural prosthetics, with sensing channel counts constantly increasing. These Investigations place increasing demands for high data rates and low-power implantable devices despite high tissue losses. The Impulse radio ultra-wideband (IR-UWB), a revived wireless technology for short-range radios, has been widely used in various applications. Since the requirements and solutions are application-oriented, in this review paper we focus on neural recording implants with high-data rates and ultra-low power requirements. We examine in detail the working principle, design methodology, performance, and implementations of different architectures of IR-UWB transceivers in a quantitative manner to draw a deep comparison and extract the bottlenecks and possible solutions concerning the dedicated application. Our analysis shows that current solutions rely on enhanced or combined modulation techniques to improve link margin. An in-depth study of prior-art publications that achieved Gbps data rates concludes that edge-combination architecture and non-coherent detectors are remarkable for transmitter and receiver, respectively. Although the aim to minimize power and improve data rate - defined as energy efficiency (pJ/b) - extending communication distance despite high tissue losses and limited power budget, good narrow-band interference (NBI) tolerance coexisted in the same frequency band of UWB systems, and compatibility with energy harvesting designs are among the critical challenges remained unsolved. Furthermore, we expect that the combination of artificial intelligence (AI) and the inherent advantages of UWB radios will pave the way for future improvements in BMIs.

This paper presents an energy-efficient wireless power receiver for implantable electrical stimulation applications, which can achieve one-step adiabatic bipolar-supply that is generated by a hybrid single-stage dual-output regulating (SSDOR) rectifiers. The structure using only four switches overcomes the disadvantages that the two output voltage values in the traditional dual-output rectifiers are close to each other. A constant-current (CC) controlled adiabatic dynamic voltage scaling (DVS) technique is proposed to minimize the voltage headroom of the stimulating drivers and improve the stimulation efficiency significantly. In addition, the receiver adopts only one general constant on-time (COT) low-frequency control to adjust the stimulation current, reducing both the power consumption and the complexity of the control circuits. The proposed receiver has been fabricated in a 0.18 μm BCD process with ±6 V voltage compliance and 2.5 mA maximum stimulating current. With a current range from ±1.5 mA to ±2.5 mA, the measured maximum average headroom voltage is only 80 mV and the peak total efficiency of the receiver is 85.6%. The functionalities of the proposed receiver have been successfully verified through in vitro experiments.

Intracortical brain-computer interfaces offer superior spatial and temporal resolutions, but face challenges as the increasing number of recording channels introduces high amounts of data to be transferred. This requires power-hungry data serialization and telemetry, leading to potential tissue damage risks. To address this challenge, this paper introduces an event-based neural compressive telemetry (NCT) consisting of 8 channel-rotating Δ-ADCs, an event-driven serializer supporting a proposed ternary address event representation protocol, and an event-based LVDS driver. Leveraging a high sparsity of extracellular spikes and high spatial correlation of the high-density recordings, the proposed NCT achieves a compression ratio of >11.4×, while consumes only 1 μW per channel, which is 127× more efficient than state of the art. The NCT well preserves the spike waveform fidelity, and has a low normalized RMS error <23% even with a spike amplitude down to only 31 μV.

Recently, deep learning-based methods have emerged as the preferred approach for ultrasound data analysis. However, these methods often require large-scale annotated datasets for training deep models, which are not readily available in practical scenarios. Additionally, the presence of speckle noise and other imaging artifacts can introduce numerous hard examples for ultrasound data classification. In this paper, drawing inspiration from self-supervised learning techniques, we present a pre-training method based on mask modeling specifically designed for ultrasound data. Our study investigates three different mask modeling strategies: random masking, vertical masking, and horizontal masking. By employing these strategies, our pre-training approach aims to predict the masked portion of the ultrasound images. Notably, our method does not rely on externally labeled data, allowing us to extract representative features without the need for human annotation. Consequently, we can leverage unlabeled datasets for pre-training. Furthermore, to address the challenges posed by hard samples in ultrasound data, we propose a novel hard sample mining strategy. To evaluate the effectiveness of our proposed method, we conduct experiments on two datasets. The experimental results demonstrate that our approach outperforms other state-of-the-art methods in ultrasound image classification. This indicates the superiority of our pre-training method and its ability to extract discriminative features from ultrasound data, even in the presence of hard examples., Competing Interests: No potential confilct of interest was reported by all authors., (© 2024 The Authors.)

In this paper, a novel fixed-window level-crossing analog-to-digital converter (LCADC) is proposed for the ECG monitoring application. The proposed circuit is implemented using fewer comparators and reference levels compared to the conventional structure, which results in a decrease in complexity and occupied silicon area. Also, the power consumption is reduced considerably by decreasing the activity of the comparator. Simulation results show a 5-fold reduction in activity by applying the standard ECG signals to the proposed structure. The proposed circuit is implemented in 0.18 μm CMOS technology using a 0.9 V supply voltage. Measurement results show a 5.9 nW power consumption and a 7.4-bit resolution. The circuit occupies a 0.05846 mm 2 silicon area. A typical level-crossing-based R-peak-detection algorithm is applied to the output samples of the LCADC, which shows the effectiveness of using this type of sampling.

This paper presents a novel resonance-based, adaptable, and flexible inductive wireless power transmission (WPT) link for powering implantable and wearable devices throughout the human body. The proposed design provides a comprehensive solution for wirelessly delivering power, sub-micro to hundreds of milliwatts, to deep-tissue implantable devices (3D space of human body) and surface-level wearable devices (2D surface of human skin) safely and seamlessly. The link comprises a belt-fitted transmitter (Belt-Tx) coil equipped with a power amplifier (PA) and a data demodulator unit, two resonator clusters (to cover upper-body and lower-body), and a receiver (Rx) unit that consists of Rx load and resonator coils, rectifier, microcontroller, and data modulator units for implementing a closed-loop power control (CLPC) mechanism. All coils are tuned at 13.56 MHz, Federal Communications Commission (FCC)-approved industrial, scientific, and medical (ISM) band. Novel customizable configurations of resonators in the clusters, parallel for implantable devices and cross-parallel for wearable devices and vertically oriented implants, ensure uniform power delivered to the load, PDL, enabling natural Tx power localization toward the Rx unit. The proposed design is modeled, simulated, and optimized using ANSYS HFSS software. The Specific Absorption Rate (SAR) is calculated under 1.5 W/kg, indicating the design's safety for the human body. The proposed link is implemented, and its performance is characterized. For both the parallel cluster (implant) and cross-parallel cluster (wearable) scenarios, the measured results indicate: 1) an upper-body PDL exceeding 350 mW with a Power Transfer Efficiency (PTE) reaching 25%, and 2) a lower-body PDL surpassing 360 mW with a PTE of up to 20%, while covering up to 92% of the human body.

Non-invasive, closed-loop brain modulation offers an accessible and cost-effective means of evaluating and modulating one's mental and physical well-being, such as Parkinson's disease, epilepsy, and sleep disorders. However, wearable EEG systems pose significant challenges for the analog front-end (AFE) circuits in view of μV-level EEG signals of interest, multiple sources of interference, and ill-defined skin contact. This paper presents a direct-digitization AFE tailored for dry-electrode scalp EEG recording, characterized by wide input dynamic range (DR) and high input impedance. The AFE utilizes a second-order 5-bit delta-delta sigma (Δ-ΔΣ) ADC to shape DC electrode offset (DEO) and low-frequency disturbances while retaining high accuracy. A non-inverting pseudo-differential instrumentation amplifier (IA) embedded in the ADC ensures high input impedance (Zin) and common-mode rejection ratio (CMRR). Fabricated in a standard 0.18-μm CMOS process, the AFE delivers 700-mVpp input signal range, 95.3-dB DR, 87-dB SNDR, and 800-MΩ input impedance at 50 Hz while consuming 88.4μW from a 1.2 V supply. The benefits of high DR and high input impedance have been validated by dry-electrode EEG measurement.

This paper presents a low-power frequency-domain functional near-infrared spectroscopy (FD-fNIRS) readout circuit for the absolute value measurement of tissue optical characteristics. The paper proposes a mixer-first analog front-end (AFE) structure and a 1-bit Σ-Δ phase-to-digital converter (PDC) to reduce the required circuit bandwidth and the laser modulation frequency, thereby saving power while maintaining high resolution. The proposed chip achieves sub-0.01° phase resolution and consumes 6.8 mW of power. Nine optical solid phantoms are produced to evaluate the chip. Compared to a self-built high-precision measurement platform that combines a network analyzer with an avalanche photodiode (APD) module, the maximum measuring errors of the absorption coefficient and reduced scattering coefficient are 10.6% and 12.3%, respectively.

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, respectively. 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.

This paper presents a low-noise bioimpedance (bio-Z) spectroscopy interface for electrical impedance myography (EIM) over the 1 kHz to 2 MHz frequency range. The proposed interface employs a sinusoidal signal generator based on direct-digital-synthesis (DDS) to improve the accuracy of the bio-Z reading, and a quadrature low-intermediate frequency (IF) readout to achieve a good noise-to-power efficiency and the required data throughput to detect muscle contractions. The readout is able to measure baseline and time-varying bio-Z by employing robust and power-efficient low-gain IAs and sixth-order single-bit bandpass (BP) ΔΣ ADCs. The proposed bio-Z spectroscopy interface is implemented in a 180 nm CMOS process, consumes 344.3 - 479.3 μW, and occupies 5.4 mm 2 area. Measurement results show 0.7 m Ω/√{Hz} sensitivity at 15.625 kHz, 105.8 dB SNR within 4 Hz bandwidth, and a 146.5 dB figure-of-merit. Additionally, recording of EIM in time and frequency domain during contractions of the bicep brachii muscle demonstrates the potential of the proposed bio-Z interface for wearable EIM systems.

In the paper, we present an integrated flow cytometer with a 2D array of magnetic sensors based on dual-frequency oscillators in a 65-nm CMOS process, with the chip packaged with microfluidic controls. The sensor architecture and the presented array signal processing allows uninhibited flow of the sample for high throughput without the need for hydrodynamic focusing to a single sensor. To overcome the challenge of sensitivity and specificity that comes as a trade off with high throughout, we perform two levels of signal processing. First, utilizing the fact that a magnetically tagged cell is expected to excite sequentially an array of sensors in a time-delayed fashion, we perform inter-site cross-correlation of the sensor spectrograms that allows us to suppress the probability of false detection drastically, allowing theoretical sensitivity reaching towards sub-ppM levels that is needed for rare cell or circulating tumor cell detection. In addition, we implement two distinct methods to suppress correlated low frequency drifts of singular sensors-one with an on-chip sensor reference and one that utilizes the frequency dependence of the susceptibility of super-paramagnetic magnetic beads that we deploy as tags. We demonstrate these techniques on a 7×7 sensor array in 65 nm CMOS technology packaged with microfluidics with magnetically tagged dielectric particles and cultu lymphoma cancer cells.

This paper presents a novel charge balancing (CB) with a current-control (CC) mode and a voltage-control (VC) mode for implantable biphasic stimulators, which can achieve one-step accurate anodic pulse generating. Compared with the conventional short-pulse-injection-based CB, the proposed method could reduce the balancing time and avoid inducing undesired artifact. The CC operation compensates the majority stimulation charge at high speed, while the VC operation guarantees a high CB precision. In order to eliminate the oscillation during the mode transition, a smooth CC-VC transition method is adopted. In addition, a digital auxiliary monitoring loop is introduced against the variations of the tissue-electrode interface impedance during the stimulation process to meet long-term CB requirement. The proposed stimulator has been fabricated in a 0.18 μm BCD process with 10 V voltage compliance, and the measured CB precision is less than 3 mV. The functionalities of the proposed CB have been verified successfully through in vitro experiments.

Recent years have witnessed significant advances brought by microfluidic biochips in automating biochemical protocols. Accurate preparation of fluid samples is an essential component of these protocols, where concentration prediction and generation are critical. Equipped with the advantages of convenient fabrication and control, microfluidic mixers demonstrate huge potential in sample preparation. Although finite element analysis (FEA) is the most commonly used simulation method for accurate concentration prediction of a given microfluidic mixer, it is time-consuming with poor scalability for large biochip sizes. Recently, machine learning models have been adopted in concentration prediction, with great potential in enhancing the efficiency over traditional FEA methods. However, the state-of-the-art machine learning-based method can only predict the concentration of mixers with fixed input flow rates and fixed sizes. In this paper, we propose a new concentration prediction method based on graph neural networks (GNNs), which can predict output concentrations for microfluidic mixters with variable input flow rates. Moreover, a transfer learning method is proposed to transfer the trained model to mixers of different sizes with reduced training data. Experimental results show that, for microfluidic mixers with fixed input flow rates, the proposed method obtains an average reduction of 88% in terms of prediction errors compared with the state-of-the-art method. For microfluidic mixers with variable input flow rates, the proposed method reduces the prediction error by 85% on average. Besides, the proposed transfer learning method reduces the training data by 84% for extending the pre-trained model for microfluidic mixers of different sizes with acceptable prediction error.

Over the past two decades Biomedical Engineering has emerged as a major discipline that bridges societal needs of human health care with the development of novel technologies. Every medical institution is now equipped at varying degrees of sophistication with the ability to monitor human health in both non-invasive and invasive modes. The multiple scales at which human physiology can be interrogated provide a profound perspective on health and disease. We are at the nexus of creating "avatars" (herein defined as an extension of "digital twins") of human patho/physiology to serve as paradigms for interrogation and potential intervention. Motivated by the emergence of these new capabilities, the IEEE Engineering in Medicine and Biology Society, the Departments of Biomedical Engineering at Johns Hopkins University and Bioengineering at University of California at San Diego sponsored an interdisciplinary workshop to define the grand challenges that face biomedical engineering and the mechanisms to address these challenges. The Workshop identified five grand challenges with cross-cutting themes and provided a roadmap for new technologies, identified new training needs, and defined the types of interdisciplinary teams needed for addressing these challenges. The themes presented in this paper include: 1) accumedicine through creation of avatars of cells, tissues, organs and whole human; 2) development of smart and responsive devices for human function augmentation; 3) exocortical technologies to understand brain function and treat neuropathologies; 4) the development of approaches to harness the human immune system for health and wellness; and 5) new strategies to engineer genomes and cells., (© 2023 The Authors.)

Goal: Cervical cancer is one of the most common cancers in women worldwide, ranking among the top four. Unfortunately, it is also the fourth leading cause of cancer-related deaths among women, particularly in developing countries where incidence and mortality rates are higher compared to developed nations. Colposcopy can aid in the early detection of cervical lesions, but its effectiveness is limited in areas with limited medical resources and a lack of specialized physicians. Consequently, many cases are diagnosed at later stages, putting patients at significant risk. Methods: This paper proposes an automated colposcopic image analysis framework to address these challenges. The framework aims to reduce the labor costs associated with cervical precancer screening in undeserved regions and assist doctors in diagnosing patients. The core of the framework is the MFEM-CIN hybrid model, which combines Convolutional Neural Networks (CNN) and Transformer to aggregate the correlation between local and global features. This combined analysis of local and global information is scientifically useful in clinical diagnosis. In the model, MSFE and MSFF are utilized to extract and fuse multi-scale semantics. This preserves important shallow feature information and allows it to interact with the deep feature, enriching the semantics to some extent. Conclusions: The experimental results demonstrate an accuracy rate of 89.2% in identifying cervical intraepithelial neoplasia while maintaining a lightweight model. This performance exceeds the average accuracy achieved by professional physicians, indicating promising potential for practical application. Utilizing automated colposcopic image analysis and the MFEM-CIN model, this research offers a practical solution to reduce the burden on healthcare providers and improve the efficiency and accuracy of cervical cancer diagnosis in resource-constrained areas., Competing Interests: The authors affirm that this study was carried out without any commercial or financial associations that would be interpreted as indicative of potential conflicts of interest., (© 2024 The Authors.)

Deep learning (DL) has been used for electromyographic (EMG) signal recognition and achieved high accuracy for multiple classification tasks. However, implementation in resource-constrained prostheses and human-computer interaction devices remains challenging. To overcome these problems, this paper implemented a low-power system for EMG gesture and force level recognition using Zynq architecture. Firstly, a lightweight network model structure was proposed by Ultra-lightweight depth separable convolution (UL-DSC) and channel attention-global average pooling (CA-GAP) to reduce the computational complexity while maintaining accuracy. A wearable EMG acquisition device for real-time data acquisition was subsequently developed with size of 36mm×28mm×4mm. Finally, a highly parallelized dedicated hardware accelerator architecture was designed for inference computation. 18 gestures were tested, including force levels from 22 healthy subjects. The results indicate that the average accuracy rate was 94.92% for a model with 5.0k parameters and a size of 0.026MB. Specifically, the average recognition accuracy for static and force-level gestures was 98.47% and 89.92%, respectively. The proposed hardware accelerator architecture was deployed with 8-bit precision, a single-frame signal inference time of 41.9μs, a power consumption of 0.317W, and a data throughput of 78.6 GOP/s.

This paper presents an arterial distension monitoring scheme using a field-programmable gate array (FPGA)-based inference machine in an ultrasound scanner circuit system. An arterial distension monitoring requires a precise positioning of an ultrasound probe on an artery as a prerequisite. The proposed arterial distension monitoring scheme is based on a finite state machine that incorporates sequential support vector machines (SVMs) to assist in both coarse and fine adjustments of probe position. The SVMs sequentially perform recognitions of ultrasonic A-mode echo pattern for a human carotid artery. By employing sequential SVMs in combination with convolution and average pooling, the number of features for the inference machine is significantly reduced, resulting in less utilization of hardware resources in FPGA. The proposed arterial distension monitoring scheme was implemented in an FPGA (Artix7) with a resource utilization percentage less than 9.3%. To demonstrate the proposed scheme, we implemented a customized ultrasound scanner consisting of a single-element transducer, an FPGA, and analog interface circuits with discrete chips. In measurements, we set virtual coordinates on a human neck for 9 human subjects. The achieved accuracy of probe positioning inference is 88%, and the Pearson coefficient (r) of arterial distension estimation is 0.838.

This paper demonstrates an inductively coupled high-accuracy localization system for miniature ingestible devices. It utilizes an inductance double capacitances-series capacitance (LCC-S) compensation architecture that enables mutual inductance measurement at primary side that is positioned outside the human body and less constrained by power budget and size than the miniature ingestible. Depending on the secondary circuit architecture, only limited and simple cooperative measurements are needed from the ingestible secondary side, which saves power and area in the miniature device. The errors in the system are modeled thoroughly, providing insights about system require-ments for a particular localization accuracy target for efficient design and to identify key building blocks with large influence on overall performance. The model shows that sub-centimeter localization root-mean-square error (RMSE) can be achieved with a modest external ADC (18bit) using three primary coils and three secondary coils. The localization is verified along a complete small intestine tract with realistic dimensions. The proposed model is verified by simulation and experiment showing that at the selected frequency range up to 5 MHz the body has no influence on the accuracy. The use of 0.9% saline as phantom is proposed which guarantees the analysis validity for all body types.

Intracardiac wireless communication is crucial for the development of multi-chamber leadless cardiac pacemakers (LCP). However, the time-varying characteristics of intracardiac channel pose major challenges. As such, mastering the dynamic conduction properties of the intracardiac channel and modeling the equivalent time-varying channel are imperative for realizing LCP multi-chamber pacing. In this paper, we present a limiting volume variational approach based on the electrical properties of cardiac tissues and trends in chamber volume variation. This approach was used to establish a quasi-static and a continuous time-varying equivalent circuit model of an intracardiac channel. An equivalence analysis was conducted on the model, and a discrete time-varying equivalent circuit phantom grounded on the cardiac cycle was subsequently established. Moreover, an ex vivo cardiac experimental platform was developed for verification. Results indicate that in the frequency domain, the congruence between phantom and ex vivo experimental outcomes is as high as 94.3%, affirming the reliability of the equivalent circuit model. In the time domain, the correlation is up to 75.3%, corroborating its effectiveness. The proposed time-varying equivalent circuit model exhibits stable and standardized dynamic attributes, serving as a powerful tool for addressing time-varying challenges and simplifying in vivo or ex vivo experiments.

This paper presents a tri-modal self-adaptive photoplethysmography (PPG) sensor interface IC for concurrently monitoring heart rate, SpO 2 , and pulse transit time, which is a critical intermediate parameter to derive blood pressure. By implementing a highly-reconfigurable analog frontend (AFE) architecture, flexible signal chain timing control, and flexible dual-LED drivers, this sensor interface provides wide operating space to support various PPG-sensing use cases. A heart-beat-locked-loop (HBLL) scheme is further extended to achieve time-multiplexed dual-input pulse transit time extraction based on two PPG sensors placed at fingertip and chest. A selfadaptive calibration scheme is proposed to automatically match the chip's operating point with the current use case, guaranteeing a sufficient signal-to-noise ratio for the user while consuming minimum system power. This paper proposes a DC offset cancellation (DCOC) approach comprised by a logarithmic transimpedance amplifier and an 8-bit SAR ADC, achieving a measured 38 nA residue error and 8.84 μA maximum input current. Fabricated in a 65nm CMOS process, the proposed trimodal PPG sensor interface consumes 2.3 - 5.7 μW AFE power and 1.52 mm2 die area with 102dB (SpO 2 mode), 110 - 116 dB (HR & PTT mode) dynamic range. A SpO 2 test case and a HR & PTT test case are both demonstrated in the paper, achieving 18.9 μW and 43.7 μW system power, respectively.

A wirelessly powered and data communication system is presented which is implemented as a full system, designed for multisite implanted biomedical applications. The system is capable of receiving wireless power and data communication for each implant separately, using inductive links with different resonance frequencies. To achieve this, dual-band coils are presented in the system. In addition, the system provides bi-directional data communication, utilizing amplitude and load shift keying (ASK and LSK) modulation schemes over a single inductive link. The system employs a digitally assisted active rectifier and an automatic resonance tuning system, to improve the power transfer efficiency (PTE) through various coupling coefficients, while minimizing the reverse current and power dissipation. The power control unit enables closed-loop monitoring to prevent high or low power delivery, and it can detect inefficient or excessive wireless power transmission or prevent temperature elevation by limiting the voltage to a safe level. A separated-V b ASK demodulator is presented in the paper which is utilized within the data conversion chain, serving both the external and implanted units. The whole system is fabricated using a standard 180-nm 1.8/3.3 V CMOS process with a core area of 0.82 mm 2 . The system is tested with coupled multisite inductive links and offers the maximum overall PTE of 31.2%, from the Tx coil to the implant load.

Robot-assisted motor training is applied for neurorehabilitation in stroke patients, using motor imagery (MI) as a representative paradigm of brain-computer interfaces to offer real-life assistance to individuals facing movement challenges. However, the effectiveness of training with MI may vary depending on the location of the stroke lesion, which should be considered. This paper introduces a multi-task electroencephalogram-based heterogeneous ensemble learning (MEEG-HEL) specifically designed for cross-subject training. In the proposed framework, common spatial patterns were used for feature extraction, and the features according to stroke lesions are shared and selected through sequential forward floating selection. The heterogeneous ensembles were used as classifiers. Nine patients with chronic ischemic stroke participated, engaging in MI and motor execution (ME) paradigms involving finger tapping. The classification criteria for the multi-task were established in two ways, taking into account the characteristics of stroke patients. In the cross-subject session, the first involved a direction recognition task for two-handed classification, achieving a performance of 0.7419 (±0.0811) in MI and 0.7061 (±0.1270) in ME. The second task focused on motor assessment for lesion location, resulting in a performance of 0.7457 (±0.1317) in MI and 0.6791 (±0.1253) in ME. Comparing the specific-subject session, except for ME on the motor assessment task, performance on both tasks was significantly higher than the cross-subject session. Furthermore, classification performance was similar to or statistically higher in cross-subject sessions compared to baseline models. The proposed MEEG-HEL holds promise in improving the practicality of neurorehabilitation in clinical settings and facilitating the detection of lesions.

As an essential cognitive function, attention has been widely studied and various indices based on EEG have been proposed for its convenience and easy availability for real-time attention monitoring. Although existing indices based on spectral power of empirical frequency bands are able to describe the attentional state in some way, the reliability still needs to be improved. This paper proposed a subject-specific attention index based on the weighted spectral power. Unlike traditional indices, the ranges of frequency bands are not empirical but obtained from subject-specific change patterns of spectral power of electroencephalograph (EEG) to overcome the great inter-subject variance. In addition, the contribution of each frequency component in the frequency band is considered different. Specifically, the ratio of power spectral density (PSD) function in attentional and inattentional state is utilized to calculate the weight to enhance the effectiveness of the proposed index. The proposed subject-specific attention index based on the weighted spectral power is evaluated on two open datasets including EEG data of a total of 44 subjects. The results of the proposed index are compared with 3 traditional attention indices using various statistical analysis methods including significance tests and distribution variance measurements. According to the experimental results, the proposed index can describe the attentional state more accurately. The proposed index respectively achieves accuracies of 86.21% and 70.00% at the 1% significance level in both the t-test and Wilcoxon rank-sum test for two datasets, which obtains improvements of 41.38% and 20.00% compared to the best result of the traditional indices. These results indicate that the proposed index provides an efficient way to measure attentional state.

Machine vision and artificial intelligence hold promise across healthcare applications. In this paper, we focus on the emerging research direction of infant action recognition, and we specifically consider the task of reaching which is an important developmental milestone. We develop E-babyNet, a lightweight yet effective neural-network-based framework for infant action recognition that leverages the spatial and temporal correlation of bounding boxes of infants' hands and objects to reach for to determine the onset and offset of the reaching action. E-babyNet consists of two main layers based on two LSTM and a Bidirectional LSTM (BiLSTM) model, respectively. The first layer provides a pre-evaluation of the reaching action for each hand by providing onset and offset keyframes. Then, the biLSTM model merges the previous outputs to deliver an outcome of the reaching actions detection for each frame including the reaching hand. We evaluated our approach against four other lightweight structures using a dataset comprising 5,865 annotated images resulting in 16,337 bounding boxes from 375 distinctive infant reaching actions performed while sitting by different subjects in unconstrained (home/clinic) environments. Results illustrate the effectiveness of our approach and ability to provide reliable reaching action detection and offer onset and offset keyframes with a precision of one frame. Moreover, the biLSTM layer can handle the transition between reaching actions and help reduce false detections.

Timely diagnosis of mild traumatic brain injury (mTBI) remains challenging due to the rapid recovery of acute symptoms and the absence of evidence of injury in static neuroimaging scans. Furthermore, while longitudinal tracking of mTBI is essential in understanding how the diseases progresses/regresses over time for enhancing personalized patient care, a standardized approach for this purpose is not yet available. Recent functional neuroimaging studies have provided evidence of brain function alterations following mTBI, suggesting mTBI-detection models can be built based on these changes. Most of these models, however, rely on manual feature engineering, but the optimal set of features for detecting mTBI may be unknown. Data-driven approaches, on the other hand, may uncover hidden relationships in an automated manner, making them suitable for the problem of mTBI detection. This paper presents a data-driven framework based on Siamese Convolutional Neural Network (SCNN) to detect mTBI and to monitor the recovery state from mTBI over time. The proposed framework is tested on the cortical images of Thy1-GCaMP6s mice, obtained via widefield calcium imaging, acquired in a longitudinal study. Results show that the proposed model achieves a classification accuracy of 96.5%. To track the state of the injured brain over time, a reference distance map is constructed, which together with the SCNN model, are employed to assess the recovery state in subsequent sessions after injury, revealing that the recovery progress varies among subjects. The promising results of this work suggest that a similar approach could be potentially applicable for monitoring recovery from mTBI, in humans.

This paper reports a microfabricated triaxial capacitive force sensor. The sensor is fully encapsulated with inert and biocompatible glass (fused silica) material. The sensor comprises two glass plates, on which four capacitors are located. The sensor is intended for subdermal implantation in fingertips and palms and providing tactile sensing capabilities for patients with paralyzed hands. Additional electronic components, such as passives and IC chips, can also be integrated with the sensor in a hermetic glass package to achieve an implantable tactile sensing system. Through attachment to a human palm, the sensor has been shown to respond appropriately to typical hand actions, such as squeezing or picking up a bottle.

Recent usage of Virtual Reality (VR) technology in surgical training has emerged because of its cost-effectiveness, time savings, and cognition-based feedback generation. However, the quantitative evaluation of its effectiveness in training is still not studied thoroughly. This paper demonstrates the effectiveness of a VR-based surgical training simulator in laparoscopic surgery and investigates how stochastic modeling represented as Continuous-time Markov-chain (CTMC) can be used to explicit the training status of the surgeon. By comparing the training in real environments and in VR-based training simulators, the authors also explore the validity of the VR simulator in laparoscopic surgery. The study further aids in establishing learning models of surgeons, supporting continuous evaluation of training processes for the derivation of real-time feedback by CTMC-based modeling.

Steady-state visual evoked potential (SSVEP), one of the most popular electroencephalography (EEG)-based brain-computer interface (BCI) paradigms, can achieve high performance using calibration-based recognition algorithms. As calibration-based recognition algorithms are time-consuming to collect calibration data, the least-squares transformation (LST) has been used to reduce the calibration effort for SSVEP-based BCI. However, the transformation matrices constructed by current LST methods are not precise enough, resulting in large differences between the transformed data and the real data of the target subject. This ultimately leads to the constructed spatial filters and reference templates not being effective enough. To address these issues, this paper proposes multi-stimulus LST with online adaptation scheme (ms-LST-OA)., Methods: The proposed ms-LST-OA consists of two parts. Firstly, to improve the precision of the transformation matrices, we propose the multi-stimulus LST (ms-LST) using cross-stimulus learning scheme as the cross-subject data transformation method. The ms-LST uses the data from neighboring stimuli to construct a higher precision transformation matrix for each stimulus to reduce the differences between transformed data and real data. Secondly, to further optimize the constructed spatial filters and reference templates, we use an online adaptation scheme to learn more features of the EEG signals of the target subject through an iterative process trial-by-trial., Results: ms-LST-OA performance was measured for three datasets (Benchmark Dataset, BETA Dataset, and UCSD Dataset). Using few calibration data, the ITR of ms-LST-OA achieved 210.01±10.10 bits/min, 172.31±7.26 bits/min, and 139.04±14.90 bits/min for all three datasets, respectively., Conclusion: Using ms-LST-OA can reduce calibration effort for SSVEP-based BCIs.

Alzheimer's Disease (AD) accounts for the majority of dementia, and Mild Cognitive Impairment (MCI) is the early stage of AD. Early and accurate diagnosis of dementia plays a vital role in more targeted treatments and effectively halting disease progression. However, the clinical diagnosis of dementia requires various examinations, which are expensive and require a high level of expertise from the doctor. In this paper, we proposed a classification method based on multi-modal data including Electroencephalogram (EEG), eye tracking and behavioral data for early diagnosis of AD and MCI. Paradigms with various task difficulties were used to identify different severity of dementia: eye movement task and resting-state EEG tasks were used to detect AD, while eye movement task and delayed match-to-sample task were used to detect MCI. Besides, the effects of different features were compared and suitable EEG channels were selected for the detection. Furthermore, we proposed a data augmentation method to enlarge the dataset, designed an extra ERPNet feature extract layer to extract multi-modal features and used domain-adversarial neural network to improve the performance of MCI diagnosis. We achieved an average accuracy of 88.81% for MCI diagnosis and 100% for AD diagnosis. The results of this paper suggest that our classification method can provide a feasible and affordable way to diagnose dementia.

Upper limb functional impairments persisting after stroke significantly affect patients' quality of life. Precise adjustment of robotic assistance levels based on patients' motion intentions using sEMG signals is crucial for active rehabilitation. This paper systematically reviews studies on continuous prediction of upper limb single joints and multi-joint combinations motion intention using Model-Based (MB) and Model-Free (MF) approaches over the past decade, based on 186 relevant studies screened from six major electronic databases. The findings indicate ongoing challenges in terms of subject composition, algorithm robustness and generalization, and algorithm feasibility for practical applications. Moreover, it suggests integrating the strengths of both MB and MF approaches to improve existing algorithms. Therefore, future research should further explore personalized MB-MF combination methods incorporating deep learning, attention mechanisms, muscle synergy features, motor unit features, and closed-loop feedback to achieve precise, real-time, and long-duration prediction of multi-joint complex movements, while further refining the transfer learning strategy for rapid algorithm deployment across days and subjects. Overall, this review summarizes the current research status, significant findings, and challenges, aiming to inspire future research on predicting upper limb motion intentions based on sEMG.