AEAzita Emami

Azita Emami

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Azita Emami has a total of 109 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • An Experimental Characterization of Atmospheric Turbulence Effects on Millimeter-Wave Propagation in a Controlled Environment

    by Shawn Sheng, Omkar Pradhan, Roni Goldshmid, Simone Tanelli, Ken Cooper, Azita Emami

    This paper introduces a novel method for experimentally characterizing atmospheric turbulence effects on millimeter-wave propagation in a controlled laboratory environment. A large fan array wind tunnel with 36x36 individually programmable dual-fan units is used to create turbulent flows over a controlled temperature gradient. A W-band (95.4GHz) radar is used to propagate electromagnetic energy through this turbulent flow, while meteorological parameters such as temperature, humidity, and wind speed are measured by co-located sensors. Phase noise self-cancellation and vibration isolation techniques are used to observe scintillation down to millimeter length scales. The effect of temperature gradients and different mean wind speeds are shown to increase and shift the measured power spectrum of the radar signal within the inertial subrange and dissipation regions of turbulence. Results from this experimental setup are shown to be consistent and repeatable across multiple days. This setup can be used to (i) investigate novel techniques for remote sensing of atmospheric turbulence, (ii) develop statistical channel models for millimeter-wave communication systems, and (iii) expand on existing scintillation theory.

  • A 192-Channel 1D CNN-Based Neural Feature Extractor in 65nm CMOS for Brain-Machine Interfaces

    by Steven P. Bulfer, Jorge Gámez, Albert Yan-Huang, Benyamin Haghi, Volnei A. Pedroni, Richard A. Andersen, Azita Emami

    We present a 192-channel 1D convolutional neural network (1D CNN) based neural feature extractor for Brain-Machine Interfaces (BMI) that achieves state-of-the-art decoding stability at $\mathbf{1.8\ \mu W}$ and 12801 $\mathbf{\mu m^{2}}$ per channel in 65nm CMOS technology. Our device is a fully configurable, scalable, area and power efficient solution that supports models with 2-8 feature layers and a total kernel length of up to 256. This architecture reduces caching requirements by $\mathbf{5}{\boldsymbol\times}$ over conventional computation schemes. Channels and layers are individually power-switchable to further optimize power efficiency for a given neural application. We introduce an on-chip model, FENet-66, that achieves the highest cross-validated decoding performance compared to all previously reported feature sets. We show that this model maintains superior stability over time using recorded data from tetraplegic human participants with spinal cord injury. Our features have 18% higher overall average cross-validated R2 decoding performance compared to Spiking Band Power (SBP), with 28% better performance during the 4th year. Our proposed architecture can also extract mean wavelet power features at low power and latency. We show that custom 1D-CNN kernels achieve 10% better performance compared to wavelet features while compressing the neural data stream by $\mathbf{38}{\boldsymbol\times}$. The models and hardware were validated in real time with a human subject in online closed-loop center-out cursor control experiments with micro-electrode arrays that were implanted for 6 years. Decoders using features generated with this work substantially improve the viability of long-term neural implants compared to other feature extraction methods currently present in low power BMI hardware.

  • Open Access

    Closed-loop Optogenetic Control in a Microplate Reader

    by Hari R. Namboothiri, Krishna Pochana, Bhavya Jaiswal, Azita Emami, Chelsea Y. Hu

    Optogenetics integrates living cells and electronics into powerful cell–silicon systems, but prototyping their dynamics remains challenging. Current tools either require robotic liquid transfers into flow cytometers or rely on custom sensors with narrow dynamic range that limit controller performance. Additionally, current successful optogenetic feedback controllers only operate in chemostats or microfluidic devices that enforce constant growth, because models for growth-aware controller design in batch culture are lacking. Here we present LEMOS, a low-cost LED-embedded microplate that runs inside a commercial microplate reader. Coupled to a growth-aware multiscale model of gene expression for controller tuning, this platform enables rapid design-build-test-learn cycles for cell-silicon systems. We demonstrate closed loop setpoint tracking of gene expression in batch cultures within a standard microplate reader and show how growth dynamics complicates controller selection and tuning. Together, this platform reduces setup overhead and speed up iteration, enabling accurate real-time optogenetic feedback control.


  • Top Cited Papers

    Continuous biochemical profiling of the gastrointestinal tract using an integrated smart capsule

    by Jihong Min, Hyunah Ahn, Heather Lukas, Xiaotian Ma, Rinni Bhansali, Sung-Hyuk Sunwoo, Canran Wang, Yadong Xu, Dickson R. Yao, Gwangmook Kim, et al.

    The gastrointestinal tract contains a wealth of chemical information that can be used to decipher the health of the digestive and nervous systems. Traditional methods of analysis, such as faecal analysis and biopsies, are invasive, costly and incapable of providing real-time metabolic and hormone profiling across the gastrointestinal tract. Commercial ingestible capsule sensors have been developed, but only monitor basic markers, such as pH and pressure, neglecting detailed chemical analysis. Here we report an integrated smart capsule that can simultaneously detect a spectrum of biochemical markers, including electrolytes, metabolites and hormones. The capsule, which is termed PillTrek, is 7 mm in diameter and 25 mm in length, and houses a miniaturized wireless electrochemical workstation capable of executing a range of electrochemical measurement techniques (potentiometry, amperometry, voltammetry and impedimetry), allowing it to interface with a variety of electrochemical sensors and detect various parameters in the gut. Using an array of sensors (serotonin, glucose, pH, ionic strength and temperature), we illustrate the capabilities of the system in vitro and in vivo in animal studies involving rat and rabbit models, monitoring the dynamic profile of these crucial biomarkers and their responsiveness to different dietary intakes.


  • Open Access

    Author Correction: Enhanced control of a brain–computer interface by tetraplegic participants via neural-network-mediated feature extraction

    by Benyamin Haghi, Tyson Aflalo, Spencer Kellis, Charles Guan, Jorge A. Gamez de Leon, Albert Yan Huang, Nader Pouratian, Richard A. Andersen, Azita Emami

  • Enhanced control of a brain–computer interface by tetraplegic participants via neural-network-mediated feature extraction

    by Benyamin Haghi, Tyson Aflalo, Spencer Kellis, Charles Guan, Jorge A. Gamez de Leon, Albert Yan Huang, Nader Pouratian, Richard A. Andersen, Azita Emami

    To infer intent, brain–computer interfaces must extract features that accurately estimate neural activity. However, the degradation of signal quality over time hinders the use of feature-engineering techniques to recover functional information. By using neural data recorded from electrode arrays implanted in the cortices of three human participants, here we show that a convolutional neural network can be used to map electrical signals to neural features by jointly optimizing feature extraction and decoding under the constraint that all the electrodes must use the same neural-network parameters. In all three participants, the neural network led to offline and online performance improvements in a cursor-control task across all metrics, outperforming the rate of threshold crossings and wavelet decomposition of the broadband neural data (among other feature-extraction techniques). We also show that the trained neural network can be used without modification for new datasets, brain areas and participants.


  • A 14.8-μW Power and μ Trms Noise 3-D AC Magnetic Sensor in CMOS for Biomedical Applications

    by Saransh Sharma, Hayward Melton, Liliana B. Edmonds, Olivia Addington, Mikhail G. Shapiro, Azita Emami

    Magnetic sensors are used extensively in applications related to automotives, navigation, medical electronics, and consumer products. Hall sensors are commonly used due to their compatibility with the standard CMOS process but suffer from poor 3-D sensitivity and consume high power. Here, we present the first-of-its-kind 3-D ac magnetic sensor in the CMOS process with high-resolution and ultralow-power operation. The sensor is comprised of three orthogonal and highly dense metal coils ( X , Y , and Z ) implemented in the 65-nm node, which generate a voltage in response to ac magnetic fields by electromagnetic induction. The X and Y sensor coils are realized in the vertical plane of the CMOS chip by using interconnect vias as part of the coil structure, while the Z sensor is realized in the horizontal plane. The induced voltage by the three coils is processed by a low-noise instrumentation amplifier (IA), sharp bandpass filter (BPF), programmable gain amplifier (PGA), differential peak detect and hold circuit, 12-bit successive approximation register (SAR) analog-to-digital converter (ADC), and serializer. The entire circuitry consumes only 14.8 $\mu$ W to yield $\mu$ T-level resolution. The sensor is successfully demonstrated for 3-D localization and tracking of catheters with 500- $\mu$ m mean accuracy in a surgical operation room (OR), which shows significant potential toward clinical translation.


  • Patterned thin film enzyme electrodes via spincoating and glutaraldehyde vapor crosslinking: towards scalable fabrication of integrated sensor-on-CMOS devices

    by Dvin Adalian, Xiomi Madero, Samson Chen, Musab Jilani, Richard D. Smith, Songtai Li, Christin Ahlbrecht, Juan Cardenas, Abhinav Agarwal, Azita Emami, et al.

    Effective continuous glucose monitoring solutions require consistent sensor performance over the lifetime of the device, a manageable variance between devices, and the capability of high volume, low cost production. Here we present a novel and microfabrication-compatible method of depositing and stabilizing enzyme layers on top of planar electrodes that can aid in the mass production of sensors while also improving their consistency. This work is focused on the fragile biorecognition layer as that has been a critical difficulty in the development of microfabricated sensors. We test this approach with glucose oxidase (GOx) and evaluate the sensor performance with amperometric measurements of in vitro glucose concentrations. Spincoating was used to deposit a uniform enzyme layer across a wafer, which was subsequently immobilized via glutaraldehyde vapor crosslinking and patterned via liftoff. This yielded an approximately 300 nm thick sensing layer which was applied to arrays of microfabricated platinum electrodes built on blank wafers. Taking advantage of their planar array format, measurements were then performed in high-throughput parallel instrumentation. Due to their thin structure, the coated electrodes exhibited subsecond stabilization times after the bias potential was applied. The deposited enzyme layers were measured to provide a sensitivity of 2.3 ± 0.2 μA mM−1 mm−2 with suitable saturation behavior and minimal performance shift observed over extended use. The same methodology was then demonstrated directly on top of wireless CMOS potentiostats to build a monolithic sensor with similar measured performance. This work demonstrates the effectiveness of the combination of spincoating and vapor stabilization processes for wafer scale enzymatic sensor functionalization and the potential for scalable fabrication of monolithic sensor-on-CMOS devices.


  • Open Access

    A Field-Deployable Arsenic Sensor Integrating Bacillus Megaterium with CMOS Technology

    by Chelsea Y. Hu, John McManus, Fatemeh Aghlmand, Elin Larsson, Azita Emami, Richard M. Murray

    Bacteria innately monitor their environment by dynamically regulating gene expression to respond to fluctuating conditions. Through synthetic biology, we can harness this natural capability to design cell-based sensors. Bacillus megaterium , a soil bacterium, stands out due to its remarkable heavy metal tolerance and sporulation ability, making it an ideal candidate for heavy metal detection with low transportation costs. However, challenges persist: the synthetic biology toolkit for this strain is underdeveloped and conventional whole-cell sensors necessitate specialized laboratory equipment to read the output. In our study, we genetically modified B. megaterium for arsenic detection, establishing a detection threshold below the EPA recommendation of 10 ppb for drinking water in both vegetative cell form and spore form. Additionally, we integrated both engineered B. megaterium living cells and spores with CMOS chip for field-deployable arsenic detection. We show that the limit of detection of our integrated sensor is applicable in soil and air arsenic contamination testing. As a proof of concept, this work paves the way for deploying our sensor in resource-limited settings, ensuring real-time arsenic detection in challenging environments. Abstract Figure:


  • Open Access

    EKGNet: A 10.96μW Fully Analog Neural Network for Intra-Patient Arrhythmia Classification

    by Benyamin Haghi, Lin Ma, Sahin Lale, Anima Anandkumar, Azita Emami

    We present an integrated approach by combining analog computing and deep learning for electrocardiogram (ECG) arrhythmia classification. We propose EKGNet, a hardware-efficient and fully analog arrhythmia classification architecture that archives high accuracy with low power consumption. The proposed architecture leverages the energy efficiency of transistors operating in the subthreshold region, eliminating the need for analog-to-digital converters (ADC) and static random access memory (SRAM). The system design includes a novel analog sequential Multiply-Accumulate (MAC) circuit that mitigates process, supply voltage, and temperature variations. Experimental evaluations on PhysioNet's MIT-BIH and PTB Diagnostics datasets demonstrate the effectiveness of the proposed method, achieving average balanced accuracy of 95% and 94.25% for intra-patient arrhythmia classification and myocardial infarction (MI) classification, respectively. This innovative approach presents a promising avenue for developing low-power arrhythmia classification systems with enhanced accuracy and transferability in biomedical applications.