1,721,031 research outputs found
Investigation and Development of CMOS Pixelated Nanocapacitor Biosensors for Quantitative Analyses
Full text linkOn the Response of Nanoelectrode Impedance Spectroscopy Measures to Plant, Animal, and Human Viruses
No full textA simplified lumped geometrical and electrical model for the high-frequency impedance spectroscopy (HFIS) response of nanoelectrodes to T=3 capsids and full viruses is developed starting from atomistic descriptions, in order to test the theoretical response of a realistic HFIS CMOS biosensor platform to different viruses. Capacitance spectra are computed for plant (cowpea chlorotic mottle virus), animal (rabbit haemorrhagic disease virus), and human (hepatitis A virus) viruses. A few common features of the spectra are highlighted, and the role of virus charge, pH, and ionic strength on the expected signal is discussed. They suggest that the frequency of highest sensitivity at nearly physiological concentrations (100 mM) is within reach of existing HFIS platform designs
Complete Cardiorespiratory Monitoring via Wearable Ultra Low Power Ultrasound
No full textThe continuous monitoring of heart (HR) and respiration (RR) rates is crucial for long-term health assessment and physical activity tracking. HR can be extracted reliably by wearable sensors based on electrocardiography (ECG) and featuring multi-electrode setups. In turn, photoplethysmography (PPG) offers similar HR detection accuracy levels with minimal setups but is affected by motion artifacts. Furthermore, while PPG and ECG may also be utilized for indirect RR assessment, they typically result in moderate accuracy. In this context, ultrasound (US) is a valid tool for complete cardiorespiratory monitoring with a single sensor. In fact, instead of probing electro-optical signals near the skin’s surface, its penetrating ability allows accurate tracking of both ventilation and heartbeat frequencies. However, existing solutions based on the US do not typically offer concurrent HR and RR detection or are not available in wearable form factors and power budgets. In this paper, we demonstrate full cardiorespiratory assessment (HR and RR measurements) by means of a single chest-worn wearable ultra low power (ULP) US probe. The proposed methodology, based on the frequency analysis of sequences of A-mode scans, enables a reliable HR and RR assessment with less than 3.9% and 4.6% errors when compared to ECG (for HR measurements) and manual counting (for RR measurements), respectively. The proposed approach does not require precise positioning of the probe on the chest, offering robustness with respect to user misplacements of the sensor. The system consumes only 17 mW, thereby enabling long-term monitoring
Open-Source Fully-Programmable Flow Phantom for Doppler Ultrasound
No full textCarotid Doppler ultrasound is a vital diagnostic tool for the early detection of alterations in vascular function, and wearable devices have been recently proposed to enable continuous monitoring. Developing and optimizing such devices and algorithms necessitates the use of phantoms, which are artificial structures designed to mimic physiological features such as vessels, soft tissues, and blood. These validation processes often rely on commercial solutions, which are expensive, complex, and offer limited customization options. While academic alternatives are more flexible, they often utilize commercial pumping solutions with restricted programmability or high cost. To address these limitations, we introduce DOPFLOW, a novel Doppler ultrasound phantom system. DOPFLOW features a modular design comprising a hydraulic pump, a control board, and a phantom body mimicking Common Carotid Arteries. The system supports user-defined pulsatile patterns with maximum flow speeds exceeding 1 m/s, and it has been tested with a commercial ultrasound Doppler machine, demonstrating a velocity estimation error of less than 15%. Combined with the system's low cost (< 400 $) and open-source design, these capabilities make DOPFLOW a promising tool for developing next-generation Doppler ultrasound devices
Enhancing Performance, Calibration Time and Efficiency in Brain-Machine Interfaces through Transfer Learning and Wearable EEG Technology
No full textBrain-Machine Interfaces (BMIs) have emerged as a transformative force in assistive technologies, empowering individuals with motor impairments by enabling device control and facilitating functional recovery. However, the persistent challenge of inter-session variability poses a significant hurdle, requiring time-consuming calibration at every new use. Compounding this issue, the low comfort level of current devices further restricts their usage. To address these challenges, we propose a comprehensive solution that combines a tiny CNN-based Transfer Learning (TL) approach with a comfortable, wearable EEG headband. The novel wearable EEG device features soft dry electrodes placed on the headband and is capable of on-board processing. We acquire multiple sessions of motor-movement EEG data and achieve up to 96% inter-session accuracy using TL, greatly reducing the calibration time and improving usability. By executing the inference on the edge every 100ms, the system is estimated to achieve 30h of battery life. The comfortable BMI setup with tiny CNN and TL pave the way to future on-device continual learning, essential for tackling inter-session variability and improving usability
ENBIOS-2D Lab
No full textENBIOS-2D Lab is a tool to illustrate and to study simple Ion Sensitive Field Effect Transistor structures in two dimensions. Together with its companion tool ENBIOS-1D Lab, it is meant for use as a teaching tool in support of undergraduate or graduate courses on the basic physics of transduction in ion and particle sensors, and to assist early stage researchers getting familiar with some basic concepts in the field.
At the present stage, ENBIOS-2D Lab supports simulation and visualization of DC I-V characteristics, impedance/admittance spectra as well as DC and AC potential/carrier/ion distributions in simple two-dimensional ISFET structures. A broader set of case studies will become available with future releases of the tool.
The companion ENBIOS-1D Lab tool offers the possibility to simulate simple Electrolyte/Insulator/Semiconductor systems in one-dimension.
The physical system is modelled with the Poisson/Boltzmann (DC) and Poisson/Nernst/Planck - Poisson/Drift/Diffusion (AC small signal) equations coupled to the site-binding charge model equations at the Electrolyte/Insulator interfaces. Dedicated models are implemented for the frequency and salinity dependence of the electrolyte electrical permittivity and the temperature dependence of the ions' mobility (in water solvent).
ENBIOS-2D Lab is powered by ENBIOS, (Electronic Nano-BIOsensor Simulator), a general purpose three-dimensional Control Volume Finite Element Method (CVFEM) simulator developed in-house at the University of Udine - Italy. ENBIOS simulates in three dimensions (3D) the DC and AC small signal impedance response to ions and micro/nanoparticles of three-dimensional devices made of semiconductor, insulator and electrolyte materials
Hand Gesture Recognition via Wearable Ultra-Low Power Ultrasound and Gradient-Boosted Tree Classifiers
No full textWearable ultrasound (US) is becoming more popular for complementing surface electromyography in the hand gesture recognition (HGR) task. In fact, US allows collecting data from deep musculoskeletal structures with high spatiotemporal resolutions and high signal-to-noise ratio. However, existing wearable US solutions for HGR are not sufficiently low-power for guaranteeing continuous, long-term operation, and they typically rely on data processing and classification approaches not suited for edge computing. In this paper, we present the first armband for hand gesture recognition based on a truly wearable (12 cm 2 , 13 g), ultra-low power (16 mW) ultrasound probe, complemented by a lightweight classification approach based on XGBoost gradient-boosted tree classifier. We demonstrate an average cross-validated classification accuracy of 97% on four different gestures, while achieving low inter-session variability (standard deviation as low as 3%) in the scenario of armband removal and repositioning across experiments. Furthermore, thanks to its low complexity and memory usage (10 KB), the classifier can be executed in real-time on a low-power resource-constrained embedded platform. The system consumes only 16 mW and enables multi-day operation with a 320 mAh battery
WULPUS: a Wearable Ultra Low-Power Ultrasound probe for multi-day monitoring of carotid artery and muscle activity
No full textUltrasound (US) is a promising tool for non-invasive, continuous monitoring of deep and superficial human body structures. Recent research advances demonstrated the applicability of A-mode US for blood flow monitoring, prostheses control, and muscle activity classification. However, despite the growing interest and progress in wearable US, existing commercial and academic research platforms do not yet offer all the key functionalities and performance metrics for wearable, configurable continuous monitoring of physiological parameters, at the same time offering access to raw data (to sustain the development of novel machine learning approaches on heterogeneous US applications). To overcome these limitations, we present WULPUS, a truly wearable ultra-low-power US open research platform. WULPUS consumes less than 25 mW, comes in a compact design (46×25 mm, 13 g), and offers an energy-efficient wireless communication link (Bluetooth low-energy) to commodity devices. The probe features 8 time-multiplexed channels, supports up to 50 Hz frame rate (FR), and provides access to raw US data, facilitating algorithm development for automated analysis
Calibration of High-Frequency Impedance Spectroscopy Measurements with Nanocapacitor Arrays
No full textHigh frequency impedance spectroscopy (HFIS) biosensors based on nano-electrode arrays (NEA)
demonstrated the capability to overcome the screening limits set by the Electrical Double Layer
(EDL), thus enabling label-free detection and imaging of analytes far above the sensor surface
[1,2]. In order to achieve quantitatively accurate results, a precise understanding and modeling of
the signal transduction chain is necessary. With reference to the CMOS array platform in [1],
capacitance is measured by CBCM. Hence, the nanoelectrodes are alternatively charged and
discharged by two switch transistors (Fig.1, a), which are activated by non-overlapping clocks with
typically 1 ns floating time between the two phases. The column readout circuits integrate and
average over multiple cycles the charging current to obtain a capacitance information. The output
signal is interpreted in terms of a switching capacitance (CSW), modeled by charge-pump analysis
of an equivalent C-RC circuit excited by a square wave (EDL capacitance CS in series to a parallel
RECE representing the bulk electrolyte [1]; CS, RE and CE are extracted with the biosensor simulator
ENBIOS [3]), good agreement is obtained between experiments and simulations over a broad
range of frequencies and electrolyte salt concentrations [1]. Residual discrepancies, however,
require explanation and this is the main contribution of our abstract. To this end, we firstly, consider
the role of leakage currents (ILEAK) in the sensor cell (due to subthreshold conduction of the inactive
switch). The leakage current implies overestimating the column current IM (and hence the
capacitance). Due to the large number of cells connected on each column, a value as large as
20pA is estimated for ILEAK, and measurements are corrected by compensating for it. Then, we
consider the voltage waveforms at the nanoelectrode, as obtained by Spice simulations with
Predictive Technology Models (PTM) of the sensor cell readout circuit (Fig.1 (b) for a 10mM
electrolyte). Charge repartition between the nanoelectrode’s node and CGS/CGD capacitance of the
switching transistors during the float time distorts the otherwise square-waveform. For electrolytes
with high salt concentration this effect is mitigated (due to the larger load capacitance). To account
for this effect, we extract the harmonic content of the waveform by Fourier expansion of the
waveform (Fig.1, b). Then, ENBIOS simulations at all harmonic frequencies are used to reconstruct
the capacitance response to the actual waveform (CF). Fig.1 (c) compares experiments (corrected
for leakage) and simulations (CSW or CF). The impact of leakage is modest, whereas CF exhibits an
improved agreement with experiments at high frequency, where waveform glitches are more
relevant. These corrections highlight the importance of leakage and harmonic content of the input
waveforms to achieve quantitatively accurate interpretation of NEA HFIS biosensor experiments.
Further work is necessary to extend these results to electrolytes with physiological salinity
VowelNet: Enhancing Communication with Wearable EEG-Based Vowel Imagery
No full textFor decades, researchers have explored Speech Imagery - silently imagining speech - as a communication aid for those with severe impairments. Despite advancements in classification accuracy, existing methods mainly rely on offline, resource-intensive machine learning techniques that necessitate multiple channels, leading to obtrusive setups and social stigma, preventing their application outside clinical settings. This paper presents, for the first time, vowel imagery classification based on a low channel-count, ultra-low-power wearable EEG system (BioGAP), and VowelNet, a novel lightweight neural network optimized for real-time Speech Imagery processing (up to 6 classes) on compact, low-power System-on-Chips. VowelNet requires 8x fewer channels than current EEG-based speech recognition systems, and it provides accuracies up to 91.1% for vowel-rest classification and up to 61.8% for inter-vowel classification. When running on the edge (GAP9), it enables continuous speech imagery classification for more than 1 day on a small 150 mAh LiPo battery, with an output latency of only 41 ms. This work paves the way for non-stigmatizing and energy-efficient assistive communication devices
- …
