This paper presents a time-domain (TD) compute-in-memory (CIM) macro based on a binary-gated dual-edge cell (BGDEC) for binary weight networks (BWNs). The BGDEC substantially mitigates subthreshold nonlinearity through binary gating and improves energy efficiency via dual-edge operation. A low-power asynchronously set truncation TDC (ASTTDC) reduces dynamic power through delay-chain truncation and DFF gating. A parity-alternating sampling scheme (PASS) saves half the DFFs with negligible accuracy loss. Implemented in a 55 nm CMOS process, the macro achieves an energy efficiency of 403.8 to 604.7 TOPS/W at 0.8 V. For the CIFAR-10 dataset with 8-bit inputs and 1-bit weights, this work achieves an inference accuracy of 90.87%–91.14%.

Dynamic Vision Sensor (DVS) is an event-based imaging technology inspired by biological photoreceptors, which holds great promise for edge computing. The event streams produced by DVS are often contaminated by Background Activity (BA) noise and hot-pixel noise, which degrade downstream processing. Existing filters typically use fixed parameters, resulting in poor adaptability to changing illumination. In this paper, we propose a lightweight Adaptive Event-based Filtering Spiking Neural Network (AEFSNN) to address these limitations. Inspired by homeostatic plasticity, AEFSNN dynamically adjusts neuronal thresholds by monitoring the input-to-output spike ratio, allowing the network to autonomously converge to an optimal operating point across different lighting conditions. Furthermore, we introduce a novel neuronal wake-up mechanism that inhibits processing neurons until triggered by valid input, which effectively suppresses redundant events generated by neighboring activity. Experiments show that AEFSNN is more robust under varying illumination. Compared with current filters, our method increases the Signal-to-Noise Ratio (SNR) of the output data by 1.42-2.33 dB. Additionally, the filtered data improves classification accuracy on downstream tasks, validating its practical value for neuromorphic vision systems.

Data-retention flip-flop (DRFF) efficiently maintains data during sleep mode and retains state during transitions between active and sleep mode. This paper proposes a novel source-biased stacked inverter (SBS-Inver ter) and a low-leakage, structure-reused DRFF. The sleep latch circuit constructed using the SBS-Inver ter can effectively reduce the power of DRFF when storing data. Reuse of the structure improves the situation of redundant transistors in certain DRFF. Fine-grained inverter level optimization reduces delay and power during the active mode. The DRFF was implemented using a 55 nm process and subjected to comprehensive analysis. Post-layout simulation results at a supply voltage of 0.4 V indicate that the proposed DRFF's data retention leakage power is only 5.3 pW. At a supply voltage of 0.8 V, the power-delay product is only 0.146 nW*ns@20 MHz. Monte Carlo simulation results considering process, voltage and temperature (PVT) variations show that the proposed DRFF can operate reliably down to a supply voltage of 0.4 V.

Achieving fast and robust collision detection on autonomous agents requires perceiving threats and issuing early warnings with minimal latency. Particularly suited for this, dynamic vision sensors (DVS) capture high-speed motion with microsecond temporal resolution, empowering computational models of the lobula giant movement detector (LGMD) to perform rapid and selective collision detection. Aligning with this bio-inspired approach, the neuromorphic architecture and inherent temporal dynamics of spiking neural networks (SNNs) render them an excellent framework for integrating these advantages. However, existing systems still suffer from pronounced noise, highly variable event rates, and limited collision selectivity in complex motion scenarios. To overcome these limitations, we propose an SNN enhancing both robustness and selectivity. Inspired by retinal adaptation, we introduce an adaptive spatiotemporal filtering (ASTF) mechanism. Leveraging spatiotemporal integration and mixed-threshold neurons with global-local adaptation, the ASTF mechanism suppresses hot pixel and background activity (BA) noise and normalizes the output event rate, ensuring stable downstream processing. To further improve looming selectivity, we propose a depth-modulated spike-timing-dependent plasticity (D-STDP) learning rule. This mechanism incorporates a depth-modulated factor derived from global motion cues, which selectively potentiates synapses for looming motion, depresses them for receding motion, and gates off plasticity for irrelevant stimuli. We evaluate the proposed model on a multi-scenario dataset captured with a DVS, containing geometric, ball, and model vehicle motion. The results demonstrate that the proposed model achieves over 94% accuracy across diverse and challenging conditions, providing a promising approach for the development of bio-inspired collision detection visual systems.

Adder neural networks remove multiplication from convolution, yet their direct L1-distance datapath still requires subtraction, absolute-value generation, and wide accumulation. We address this cost by mapping the online L1 operation to minimum selection and time-domain accumulation. The proposed accelerator processes a 3×3×16 window for 16 output channels with 6-bit weights and activations. Each 6-bit minimum is divided into two 3-bit slices. A dual-mode digital-to-time converter (DM-DTC) encodes the most-significant slice in high-linearity (HL) mode and the least-significant slice in low-power (LP) mode. Readout is performed by a shared-clock time-to-digital converter (SC-TDC), in which one Gray-code time reference serves all paths while local latches preserve independent channel results. The training model reproduces code-dependent DTC nonlinearity, process–voltage–temperature variation, jitter, channel offset, TDC quantization, saturation, and scale mismatch. The architecture thereby combines significance-aware time encoding, channel-scalable readout, and hardware-aware adaptation. Post-layout simulations in 55 nm show that the 0.359 mm2, 13.7 Kb design operates at 0.7–1.2 V and 5–30 MHz, consumes 0.025–0.324 mW, and achieves 43.2–94.3 TOPS/W. The normalized figure of merit is 6.01–13.09 POPS/W·bit2. On CIFAR-10/ResNet-20, hardware errors reduce the baseline accuracy from 92.71% to 86.26%; error-aware training achieves 91.53%.

This paper presents an ultra-low-power 128 x 128 pixel pulse-width modulation (PWM) CMOS image sensor for low power applications. For always-on ultra-low-power imaging, a novel PWM pixel circuit with a tapered reset technology is implemented. Additionally, by turning off all the pixels except those in reset phase and readout phase, power consumption is reduced from 52 nW to 0.1 nW for each pixel. In addition, to overcome the settling latency of subthreshold comparator, we use a programmable ramp generator for voltage-to-time conversion for linear response, achieving a non-linearity of 0.04%. This ultra-low-power CMOS image sensor is designed and fabricated in CMOS 180 nm process technology. Measurement results demonstrate that the proposed CMOS image sensor consumes only 61.6 mu W at 62.5 frames per second (fps) with a fill factor of 58% at 0.8V operation. These performances make the image sensor perfectly suitable for IoT applications and some other edge devices.

PURPOSE:To elucidate the mechanisms underlying pyroptosis in skin due to heavy ion radiation. MATERIALS AND METHODS:Human keratinocytes (HaCaT cells) were irradiated with different doses of X-rays and 12C ions. Clonogenic survival, CCK-8 cell proliferation, and micronucleus assays were performed to assess the radiosensitivity of HaCaT cells. The pyroptosis-related, inflammation-related, and MAPK/NF-κB signaling pathway proteins were detected by Western-blotting. RESULTS:12C ion radiation caused more severe damage to HaCaT cells than X-rays. The former induced pyroptosis in the HaCaT cells in a dose-dependent manner. Heavy ion-induced pyroptosis was activated by the Caspase-4/Caspase-5/Gasdermin D (GSDMD) pathway, which was regulated by the MAPK/NF-κB signaling pathway. CONCLUSIONS:12C ion irradiation induced pyroptosis in human keratinocytes by a non-classical pathway via the activation of MAPK/NF-κB signaling pathway. The findings may be used to guide further research on targeted interventions to reduce skin damage and optimize treatment strategies in the future.

The relentless growth of artificial intelligence of things (AIoT) applications has tightened sub-milliwatt power-budget constraints on embedded vision systems. To address this challenge, this brief presents the first brain-inspired intelligent vision sensor (BIVS) based on a novel spiking processing-in-sensor (PIS) architecture that efficiently converts captured pulsewidth-modulation (PWM) image data into real-time sparse spike features for subsequent intelligent processing, offering an attractive balance between computational accuracy and energy efficiency. A 126x126 BIVS is designed in 180nm CMOS technology and consumes 66.9 mu W at 661fps with a resulting efficiency of 5.02 TOPS/W and FoM of 6.37 pJ/pixel/frame. It also achieves a spike feature map accuracy of 99.73% on the MNIST dataset and a classification accuracy of 99.05% when coupled with a simplified spiking convolutional neural network (SCNN).

This paper introduces a Fully-Static, Contention-Free Single-Phase-Clock Flip-Flop for IoT devices. The design objectives are accomplished through three main components. Firstly, the elimination of invalid toggling of internal nodes significantly reduces power consumption. Secondly, a Fully-Static and a Contention-Free structure are achieved by eliminating floating nodes and contention paths, respectively, enhancing robustness. Lastly, area is reduced through logic merging and topology compression. This Flip-Flop is implemented in a 28nm process. Post-simulation results indicate that, at 0.9V, 1GHz and activity rate of 10%/20%, dynamic power of this design is only 0.44 mu W/ 0.88 mu W, resulting in 86.95%/76.44% power reduction compared to traditional MSFF. This design focuses on static power. Static power of this design achieves 0.07nW/0.77nW at 10MHz, 0%, 0.4V/0.9V, 98.91%/97.59% less than MSFF. 10K Monte Carlo simulation confirms that this design maintains 100% functional integrity within the voltage range of 0.4-0.9V. Layout results indicate this design achieves an extremely low area.

In digital systems, optimising the power and delay of flip-flops (FFs) can remarkably enhance its performance. In this paper, an ultra-low power flip-flop (CPCDFF) based on sense amplifier is proposed. The conditional pre-charge (CP) circuit achieves non-redundant transition, which enables power wastage to be reduced. A lite-latch structure is proposed to effectively reduced the CK-to-Q delay of the FF. The proposed completion detection (CD) technique effectively improves the reliability of the FF at near-threshold voltage. The proposed FF is implemented in 22 nm process, and a comprehensive analysis. At 0.8V supply voltage, the postlayout simulation results show that the power of the proposed FF is only 0.151 mu W@500MHz and 10% data toggle rate; the power delay product is only 1.473 nW*ns@20MHz at 20% data toggle rate. The Monte-Carlo simulation results considering the process, voltage, and temperature (PVT) variations indicated that the proposed FF could operate reliably down to a supply voltage of 0.4 V.

This article presents a mixed-signal general-purpose computing core designed to support real-time inference applications in always-on low-power pulsewidth modulation (PWM) CMOS image sensors (CISs), functioning as a processing-in-sensor (PIS) circuit. This core can be integrated into the columns of CISs to perform low-power edge processing on images, without affecting the pixel fill factor and imaging quality. It employs a coordinated mechanism in which motion detection (MD) triggers the activation of feature extraction (FE), thereby achieving an organic integration of MD and FE functionalities, and maximizing the system's power efficiency. MD is implemented via in-column frame difference (FD), while FE is performed using a programmable-weight 3 x 3 convolution, a rectified linear unit activation function, and a 2 x 2 max-pooling (MP) operation. Both functionalities are computed based on real-time PWM signals from the CIS and the principle of current integration, with partial circuit reuse achieved through different switching operations. A 0.8-V computing core prototype, with an area of 720 x 272 mu m, was fabricated and verified using 0.18-mu m standard CMOS technology. The experimental results at an image frame rate of 250 fps demonstrated an average power consumption of 5.23 mu W for MD and 17.53 mu W for FE. The prototype core computes the first two layers of an ultra lightweight convolutional neural network (CNN) for the task of MNIST digit classification, achieving an accuracy loss of only 0.86% compared to the ideal scenario. This analog computing core can be used in multimode, low-power, edge-intelligent vision sensors.

Previous studies have demonstrated that combination therapy involving radiotherapy and aspirin decreases the survival rate of cancer cells. However, the mechanism by which aspirin exerts its radiation sensitization effect at the in vivo level remains largely unclear. In this study, we employed Caenorhabditis elegans (C. elegans) as a model organism to investigate the effect of aspirin combined with radio/chemo-therapy on tumors at the individual level. Here, we illustrate that high-dose aspirin increases the expression of genes involved in core apoptosis pathways (egl-1, ced-9, ced-4 and ced-3) and induces germ cell apoptosis in C. elegans through mitochondrial outer membrane permeabilization (MOMP) and elevation of reactive oxygen species (ROS) levels. Crucially, aspirin-induces ROS upregulates the expression of genes critical for DNA damage response (hus-1, clk-2 and cep-1) and genes involved in MAPK pathways (lin-45, mek-2, mpk-1, sek-1 and pmk-1), thereby mediating the enhanced sensitivity of radio/chemo-therapy by aspirin. Notably, aspirin fails to induce germ cell apoptosis and enhance radio/chemo-therapy in C. elegans lacking the expression of each of those genes. Furthermore, in a C. elegans tumor-like symptom model, aspirin enhances radio/chemo-therapy sensitivity through ROS induction. However, low-dose aspirin can diminish the apoptotic signal of reproductive cells in C. elegans and exert anti-inflammatory effects. Our research results suggest that the tumor-suppressive and radio/chemo-therapy sensitizing effects of aspirin provide robust experimental evidence for improving the clinical efficacy of tumor radio/chemo-therapy and deepening our understanding of aspirin's mechanism of action in cancer.

Abstract Glioblastomas (GBMs) are intracranial gliomas with the highest aggressiveness. Despite maximal treatment intervention, GBM patients’ median survival duration remains at approximately 14–16 months. Nuclear receptor-binding protein 1 (NRBP1) could stimulate the growth of cells. In this study, we investigated whether NRBP1 promotes malignant glioblastoma phenotypes and its potential mechanisms. High NRBP1 expression correlated with higher-grade glioma and shorter duration of overall and disease-free survival. NRBP1 knockdown via short hairpin RNAs caused suppression of cell proliferation, invasion, migration and triggered apoptotic cell death in vitro, whereas its overexpression, through plasmid transfection, showed the opposite effect. GO enrichment and KEGG analysis revealed that NRBP1 regulated differentially expressed gene clusters involved in the PI3K/Akt signaling pathway. Additionally, NRBP1 regulated epithelial–mesenchymal transition mediated by this pathway. Moreover, MK-2206 and SC79, which are respectively an inhibitor and an activator of PI3K/Akt signaling, reversed the effect of NRBP1 knockdown and overexpression on GBM, respectively. Thus, NRBP1 promotes malignant phenotypes in GBM by activating the PI3K/Akt pathway, thereby serving as a prognostic indicator and new target for GBM treatment.

Apoptosis is a highly regulated cell death program that can be mediated by death receptors in the plasma membrane, as well as the mitochondria and the endoplasmic reticulum. Apoptosis plays a key role in the pathogenesis of a variety of human diseases. Peroxisomes are membrane-bound organelles occurring in the cytoplasm of eukaryotic cells. Peroxisomes engage in a functional interplay with mitochondria. They cooperate with each other to maintain the balance of reactive oxygen species homeostasis in cells. Given the key role of mitochondria in the regulation of apoptosis, there could also be an important relationship between peroxisomes and the apoptotic process. Peroxisome dysfunction severely affects mitochondrial metabolism, cellular morphological stability, and biosynthesis, and thus contributes directly or indirectly to a number of apoptosis-related diseases. This chapter provides an overview of the concept, characteristics, inducing factors, and molecular mechanisms of apoptosis, as well as evidence for apoptosis in cancer, cardiovascular diseases, and neurodegenerative disorders, and discusses the important role of the peroxisome in the apoptosis-associated diseases.

A slope analog-to-digital converter (ADC) amenable to be fully implemented on a digital field programmable gate array (FPGA) without requiring any external active or passive components is proposed in this paper. The amplitude information, encoded in the transition times of a standard LVDS differential input—driven by the analog input and by the reference slope generated by an FPGA output buffer—is retrieved by an FPGA time-to-digital converter. Along with the ADC, a new online calibration algorithm is developed to mitigate the influence of process, voltage, and temperature variations on its performance. Measurements on an ADC prototype reveal an analog input range from 0.3 V to 1.5 V, a least significant bit (LSB) of 2.6 mV, and an effective number of bits (ENOB) of 7.4-bit at 600 MS/s. The differential nonlinearity (DNL) is in the range between −0.78 and 0.70 LSB, and the integral nonlinearity (INL) is in the range from −0.72 to 0.78 LSB.

This paper presents a 19 ps precision and 170 M samples/s time-to-digital converter (TDC) in FPGA. Through the direct count method and tapped delay line method, the coarse count and fine count can be extracted, respectively. The direct count is realized by the 350 M clock and the tapped delay line is constructed by the CARRY4 block. The ones-counter encoder is used to convert the thermometer code with bubble errors into binary code, which is applicable to all the FPGA chips. This work not only explains the schematic of the ones-counter encoder, but also shows how to configure it. Owing to the inconsistency of delay elements caused by process, bin-by-bin calibration is utilized to improve the differential nonlinearities (DNL) and integral nonlinearities (INL) of the TDC. A novel method was developed to compensate the influence of voltage and temperature. As the delay elements vary with voltage and temperature, a frequency counter is used to extrapolate and compensate its effect on the delay line. All of the above strategies use online calibration and improve the precision and sampling rate of TDC. The experimental results show the least significant bit (LSB) achieves 17.4 ps, the DNL is within [−0.90, 1.67] LSB, and the INL is in the range of [−1.90, 3.31] LSB.

随着高级驾驶员辅助系统(advanced driver assistance systems,ADAS)中对高分辨率光检测和测距(light detection and ranging,LIDAR)技术的需求越来越高,文章提出一种应用于激光测距技术的去噪算法,该算法是一种完整的数字信号处理方法,用时间相关的单光子计数来进行飞行时间(time-of-flight,TOF)测距,该方法的核心是将原始光子存储在一个m×n的阵列中.该算法包括粗滤波、细滤波2个过程,在粗滤波中,将每16个光子作为一个组,首先编程为4×4子阵列,在子阵列块中,将汇总所有光子;然后逐块进行筛选.在细滤波中,对通过粗滤波选择的3个子阵列中的每个子阵列的16个数据进行卷积;然后在每个子阵列中逐个滤波来确定最大数量的光子.与在单光子统计直方图中执行逐条滤波的传统算法相比,该算法的效率提高了10倍以上.仿真实验结果表明,文中提出的方法大大提高了运算速度,同时精度仍然很高,可用于激光雷达芯片的实时信号处理.

Compared with traditional two-dimensional culture, a three-dimensional (3D) culture platform can not only provide more reliable prediction results, but also provide a simple, inexpensive and less time-consuming method compared with animal models. A direct in vitro model of the patient’s tumor can help to achieve individualized and precise treatment. However, the existing 3D culture system based on microwell arrays has disadvantages, such as poor controllability, an uneven spheroid size, a long spheroid formation time, low-throughput and complicated operation, resulting in the need for considerable labor, etc. Here, we developed a new type of microdevice based on a 384-well plate/96-well plate microarray design. With our design, cells can quickly aggregate into clusters to form cell spheroids with better roundness. This design has the advantage of high throughput; the throughput is 33 times that of a 384-well plate. This novel microdevice is simple to process and convenient to detect without transferring the cell spheroid. The results show that the new microdevice can aggregate cells into spheroids within 24 h and can support drug and radiation sensitivity analyses in situ in approximately one week. In summary, our microdevices are fast, efficient, high-throughput, simple to process and easy to detect, providing a feasible tool for the clinical validation of individualized drug/radiation responses in patients.