Department of Electrical and Computer Engineering

Rapid and accessible cardiac biomarker testing is essential for the timely diagnosis and risk assessment of myocardial infarction (MI) and heart failure (HF), two interrelated conditions that frequently coexist and drive recurrent hospitalizations with high mortality. However, current laboratory and point-of-care testing systems are limited by long turnaround times, narrow dynamic ranges for the tested biomarkers, and single-analyte formats that fail to capture the complexity of cardiovascular disease. Here, we present a deep learning-enhanced dual-mode multiplexed vertical flow assay (xVFA) with a portable optical reader and a neural network-based quantification pipeline. This optical sensor integrates colorimetric and chemiluminescent detection within a single paper-based cartridge to complementarily cover a large dynamic range (spanning ~6 orders of magnitude) for both low- and high-abundance biomarkers, while maintaining quantitative accuracy. Using 50 µL of serum, the optical sensor simultaneously quantifies cardiac troponin I (cTnI), creatine kinase-MB (CK-MB), and N-terminal pro-B-type natriuretic peptide (NT-proBNP) within 23 min. The xVFA achieves sub-pg/mL sensitivity for cTnI and sub-ng/mL sensitivity for CK-MB and NT-proBNP, spanning the clinically relevant ranges for these biomarkers. Neural network models trained and blindly tested on 92 patient serum samples yielded a robust quantification performance (Pearsons r > 0.96 vs. reference assays). By combining high sensitivity, multiplexing, and automation in a compact and cost-effective optical sensor format, the dual-mode xVFA enables rapid and quantitative cardiovascular diagnostics at the point of care.

Many laser-based high-energy-density science (HEDS) facilities have one or more short-pulse (sub- to few-picosecond) laser beams for diagnostics. For the past decade, we have been developing a novel x-ray probing capability using such picosecond lasers interacting with an underdense plasma to produce relativistic electrons. The ultimate goal of these experiments is to demonstrate a new type of x-ray backlighter using the short-pulse ARC laser at the National Ignition Facility (NIF). Before this diagnostic is fielded at the NIF, it is critical to demonstrate the viability and reproducibility of the x-ray source on comparable high-power short-pulse laser systems. We present experiments that were carried out with the OMEGA EP laser at the University of Rochester’s laboratory for laser energetics. In these experiments, high-energy electrons are produced through a combination of the self-modulation instability and direct laser acceleration in an underdense gas jet. These electrons generate directional x-rays with characteristic energies up to several tens of keV as they execute betatron motion in the ion channel, and the inverse Compton scattering process generates even harder x-rays, with characteristic photon energies of 60–240 keV. When implemented on the OMEGA EP laser(s), this x-ray source yields results that are comparable to those obtained recently on the short-pulse Titan laser at the Jupiter Laser Facility at Lawrence Livermore National Laboratory, after accounting for differences in laser energy, peak intensity, focusing f/#, and plasma source. Applications of such an x-ray source for HEDS experiments are discussed.

Certain epigenetic modifications, such as the methylation of CpG sites, can serve as biomarkers for chronological age. Previously, we introduced the BayesAge frameworks for accurate age prediction through the use of locally weighted scatterplot smoothing (LOWESS) to capture the nonlinear relationship between methylation or gene expression and age, and maximum likelihood estimation (MLE) for bulk bisulfite and RNA sequencing data. Here, we introduce MicroBayesAge, a maximum likelihood framework for age prediction using DNA microarray data that provides less biased age predictions compared to commonly used linear methods. Furthermore, MicroBayesAge enhances prediction accuracy relative to previous versions of BayesAge by subdividing input data into age-specific cohorts and employing a new two-stage process for training and testing. Additionally, we explored the performance of our model for sex-specific age prediction which revealed slight improvements in accuracy for male patients, while no changes were observed for female patients.

Compute in memory (CiM) accelerators perform matrix vector multiplications (MVMs) directly inside memory arrays, reducing data movement and improving both energy efficiency and throughput for AI workloads. To reduce the number of conversions, recent designs use multi-bit compute cells. Nevertheless, practical multi-bit CiM still faces a tension between accuracy, efficiency, and re-writeability, since multi-level NVM based designs suffer from nonlinearity and poor re-writeability, while multi-level activation based DRAM / SRAM macros are limited by mismatch and low accuracy. This work introduces a per-cell DAC based CiM macro that combines the density of multi-level NVM with fully re-writable DRAM weights to break the trade-off. Each bit cell embeds a compact 6-bit CTT based current mode DAC, calibrated in-situ through a write verify write loop, together with a 1T(1C) embedded DRAM. Post-layout simulations of a 576×256 macro in 22nm FDSOI project 49.9 8b-TOPS/W, and 8.96 8b-TOPS/mm2, while system level evaluation on CIFAR-10 and CIFAR-100 with ResNet-50 shows less than 0.2% accuracy loss relative to a digital baseline, without the need for any Hardware-aware Training.

Large antenna arrays are critical for reliability and high data rates in wireless networks at millimeter-wave and sub-terahertz bands. While traditional methods for initial beam alignment for analog phased arrays scale beam alignment overhead linearly with the array size, compressive sensing (CS) and machine learning (ML) algorithms can scale logarithmically. CS and ML methods typically utilize pseudo-random or heuristic beam designs as compressive codebooks. However, these codebooks may not be optimal for scenarios with uncertain array impairments or multipath, particularly when measurements are phase-less or power-based. In this work, we propose a novel dictionary learning method to design codebooks for phase-less beam alignment given multipath and unknown impairment statistics. This codebook learning algorithm uses an alternating optimization with block coordinate descent to update the codebooks and Monte Carlo trials over multipath and impairments to incorporate a-priori knowledge of the hardware and environment. Additionally, we discuss engineering considerations for the codebook design algorithm, including a comparison of three proposed loss functions and three proposed beam alignment algorithms used for codebook learning. As one of the three beam alignment methods, we propose transfer learning for ML-based beam alignment to reduce the training time of both the ML model and codebook learning. We demonstrate that codebook learning and our ML-based beam alignment algorithms can significantly reduce the beam alignment overhead in terms of number of measurements required.

Generative models cover various application areas, including image and video synthesis, natural language processing and molecular design, among many others^1-11. As digital generative models become larger, scalable inference in a fast and energy-efficient manner becomes a challenge^12-14. Here we present optical generative models inspired by diffusion models⁴, where a shallow and fast digital encoder first maps random noise into phase patterns that serve as optical generative seeds for a desired data distribution; a jointly trained free-space-based reconfigurable decoder all-optically processes these generative seeds to create images never seen before following the target data distribution. Except for the illumination power and the random seed generation through a shallow encoder, these optical generative models do not consume computing power during the synthesis of the images. We report the optical generation of monochrome and multicolour images of handwritten digits, fashion products, butterflies, human faces and artworks, following the data distributions of MNIST¹⁵, Fashion-MNIST¹⁶, Butterflies-100¹⁷, Celeb-A datasets¹⁸, and Van Gogh's paintings and drawings¹⁹, respectively, achieving an overall performance comparable to digital neural-network-based generative models. To experimentally demonstrate optical generative models, we used visible light to generate images of handwritten digits and fashion products. In addition, we generated Van Gogh-style artworks using both monochrome and multiwavelength illumination. These optical generative models might pave the way for energy-efficient and scalable inference tasks, further exploiting the potentials of optics and photonics for artificial-intelligence-generated content.

Quantum-cascade vertical-external-cavity surface-emitting-lasers (VECSELs) based on disordered amplifying metasurfaces are demonstrated and explored as potential broadband, multi-mode THz sources. The disorder is introduced along one spatial axis of the metasurface by pseudo-randomly varying the width of its resonant ridge antennas. Compared to a quantum-cascade (QC) VECSEL based on a uniform metasurface, the disordered structure supports much more localized transverse modes with reduced spatial overlap within the QC gain material. This localization is hypothesized to facilitate the spatial hole burning of the gain material and, therefore, enable multi-mode lasing, particularly for short cavities on the order of a few wavelengths. Several devices have been fabricated and shown to differ from uniform QC-VECSELs in a few key ways, possessing highly nonlinear light-current characteristics, angle-dependent emission spectra and broadband multi-mode lasing. At most, 17 modes are simultaneously observed, spanning 680 GHz. The number of VECSEL modes is shown to have an inverse relationship with the cavity length, which is attributed to increased diffractive losses in the open plano-plano cavity. As the cavity length is tuned, the device emits over a quasi-continuous band from 3.15 to 3.97 THz or up to 4.24 THz with a longitudinal mode hop.

Unmanned aerial vehicles have undergone substantial development and market growth recently. With research focusing on improving control strategies for better user experience, feedback systems, which are vital for operator awareness of surroundings and flight status, remain underdeveloped. Current bulky manipulators also hinder accuracy and usability. Here, we present an enhanced human-machine interface based on skin-integrated multimodal sensing and feedback devices for closed-loop drone control. This system captures hand gestures for intuitive, rapid, and precise control. An integrated tactile actuator array translates the drone's posture into two-dimensional tactile information, enhancing the operator's perception of the flight situation. Integrated obstacle detection and neuromuscular electrical stimulation-based force feedback system enable collision avoidance and flight path correction. This closed-loop system combines intuitive controls and multimodal feedback to reduce training time and cognitive load while improving flight stability, environmental awareness, and the drone's posture. The use of stretchable electronics also addresses wearability and bulkiness issues in traditional systems, advancing human-machine interface design.

Cardiovascular digital health technologies potentially outperform traditional clinical equipment through their noninvasive, on-body, and portable monitoring with mass cardiac data beyond the confines of inpatient settings. However, existing cardiovascular wearables have difficulty with providing medical-grade accuracy with a chronically comfortable and stable patient/consumer device interface for reliable clinical decision-making. Here, we develop an intensive care unit (ICU)-grade breathable cardiac electronic skin system (BreaCARES) for real-time, wireless, continuous, and comfortable cardiac care. BreaCARES enables a novel digital cardiac care platform for health care, outpatient diagnostics, stable intraoperative monitoring during heart surgery, and continuous and comfortable inpatient postoperative cardiac care, exhibiting ICU-grade accuracy while having superior anti-interference stability, portability, and long-term on-skin biocompatibility to the clinically and commercially available cardiac monitors in cardiovascular ICUs.