Number Of Neurons Research Articles

On the turnpike to design of deep neural networks: Explicit depth bounds

It is well-known that the training of Deep Neural Networks (DNN) can be formalized in the language of optimal control. In this context, this paper leverages classical turnpike properties of optimal control problems to attempt a quantifiable answer to the question of how many layers should be considered in a DNN. The underlying assumption is that the number of neurons per layer—i.e., the width of the DNN—is kept constant. Pursuing a different route than the classical analysis of approximation properties of sigmoidal functions, we prove explicit bounds on the required depths of DNNs based on asymptotic reachability assumptions and a dissipativity-inducing choice of the regularization terms in the training problem. Numerical results obtained for the two spiral task data set for classification indicate that the proposed constructive estimates can provide non-conservative depth bounds.

Design and implementation of deep neural network hardware chip and its performance analysis

The artificial neural network (ANN) with a single layer has a limited capacity to process data. Multiple neurons are connected in the human brain, and the actual capacity of the brain lies in the interconnectedness of multiple neurons. As a specified generalization of ANN deep learning makes use of two or more hidden layers, which implies that a greater number of neurons are required to construct the model. A network that has more than one hidden layer, also known as two or more hidden layers, is referred to as a deep neural network, and the process of training such networks is referred to as deep learning. The research article focuses on the design of a multilayer or deep neural network presented for the target field programmable gate array (FPGA) device spartan-6 (xc6stx4-2t9g144) FPGA. The simulation is carried out using Xilinx ISE and ModelSim software. There are two hidden layers in which (2×1) multiplexer blocks are utilized for processing twenty neurons into the output of ten neurons in the first hidden layer and demultiplexers (1×2) and vice versa. The hardware utilization is estimated on FPGA to compute the performance of the deep neural hardware chip based on memory, flip flops, delay, and frequency parameters. The design is scalable and applicable to various FPGA devices, which makes the work novel. FPGA-based neuromorphic hardware acceleration platform with high speed and low power for discrete spike processing on hardware with great real-time performance.

Gastrodin reduces Aβ brain levels in an Alzheimer's disease mouse model by inhibiting P-glycoprotein ubiquitination

BackgroundStudies have demonstrated the potential of gastrodin in the treatment of Alzheimer's disease (AD), however, its mechanism of action remains elusive. Currently, the Amyloid-β (Aβ) cascade hypothesis continues to be the prevailing theory regarding AD etiology. The ubiquitination of P-glycoprotein (P-gp) at the blood-brain barrier (BBB) contributes to the accumulation of Aβ in the brain during AD. PurposeTo investigate the mechanism of gastrodin intervention in AD. MethodsThe molecular docking, molecular dynamics simulations, and microscale thermophoresis (MST) were employed to identify the action target of gastrodin. The western blot (WB) was performed to detect the protein expression level, the ubiquitination level of P-gp was determined using co-immunoprecipitation (CO-IP) assay. P-gp transport activity was detected using an NBD-CSA fluorescence assay. Trans-Epithelial Electrical Resistance (TEER) was used to detect cell resistance. Fluorescein-labeled dextran experiments were performed to determine the individual cell permeability. The immunofluorescence (IF) was employed to detect Aβ deposition, the Morris Water Maze test was used to assess behavioral changes in APP/PS1 mice and the levels of Aβ40 and Aβ42 expression were quantified using enzyme-linked immunosorbent assay. ResultsThe FBXO15 was the target of gastrodin-mediated inhibition of P-gp ubiquitination. Gastrodin increased the P-gp expression, cell resistance, and P-gp transport activity of BEND.3 cells upon treatment with Aβ40 through mechanisms involving the reduction of FBXO15 and P-gp binding and the inhibition of P-gp ubiquitination. And gastrodin could effectively improve memory function and increase number of neurons in APP/PS1 mice, reduce the accumulation of Aβ40 and Aβ42, and enhance P-gp expression in a dose-dependent manner. ConclusionAβ40 induces the ubiquitination and proteasomal degradation of BBB P-gp, however, gastrodin inhibits the ubiquitination of P-gp by binding to FBXO15, thereby increasing P-gp protein expression and enhancing its transport function.

Eighty-six billion and counting: do we know the number of neurons in the human brain?

Through statistical analysis and comparison of different studies, Alain Goriely challenges the widely accepted figure of 86 billion neurons in the human brain, and argues that the actual number is uncertain, with estimates ranging between 61 and 99 billion.

Effect and mechanism of Donepezil hydrochloride on Alzheimer’s disease

Objective: To explore the effect and mechanism of donepezil hydrochloride on Alzheimer’s disease (AD).Methods: Thirty-six 3-month-old SD rats were selected as the research subjects and randomly divided into a normal control (NC) group and a model group. The model group was given continuous intraperitoneal injection of D-galactose solution (120 mg/kg/d) combined with aluminum trichloride (10 mg/kg/d) by gavage for 60 days. Morris water maze test was conducted to test the learning and memory abilities of the rats, and AD rats were selected. After modeling, AD rats were divided into a normal saline (NS) group and a drug treatment (DT) group. The DT group was given donepezil hydrochloride (1.0 mg/kg) by gavage intervention, and the NC group and the NS group were given equal volumes of physiological saline by gavage. After 4 weeks of intervention, Morris water maze test was conducted to detect the escape latency, the number of platform crossings and residence time in the target quadrant, the rats were euthanized, with serum and hippocampal tissues collected, and hippocampal tissue homogenate (10%) was prepared by using NS. Enzyme linked immunosorbent assay (ELISA) was used to detect the inflammatory and oxidative stress indicators in rat hippocampal tissues and serum, and hematoxylin eosin (HE) staining was used to observe pathological damage in rat hippocampal tissues.Results: Compared with the NC group, the NS group showed a significant increase in escape latency (p < .05), a significant decrease in the number of platform crossings and residence time in the target quadrant (p < .05), and a significant decrease in the activities of superoxide dismutase (SOD) and glutathione peroxidase (GSH-Px) in serum and hippocampal tissues (p < .05). The content of malondialdehyde (MDA) was significantly increased (p < .05), and the levels of interleukin-6 (IL-6) and tumor necrosis factor-α (TNF-α) were significantly increased (p < .05). Compared with the NS group, the DT group can significantly reduce the escape latency (p < .05), increase the number of platform crossings and residence time in the target quadrant (p < .05), significantly increase the activities of SOD and GSH-Px in serum and hippocampal tissues and reduce the content of MDA (p < .05) and the levels of IL-6 and TNF-α (p < .05). After intervention with donepezil hydrochloride, the number of neurons in the hippocampus were significantly increased.Conclusions: Donepezil hydrochloride can improve the learning and memory abilities of rats, reduce the levels of oxidative stress and inflammation in the brain and serum, and improve pathological damage in the hippocampus.

Protective effects of geraniol in a male rat model of Alzheimer's disease: A behavioral, biochemical, and histological study.

Alzheimer's disease (AD) as a neurodegenerative disease can cause behavioral impairments due to oxidative stress. Aging and oxidative conditions are some AD risk factors. We assessed the influence of geraniol (GR), an acyclic monoterpene alcohol, on behavioral functions, hippocampal oxidative status, and histological alterations in AD rats induced by amyloid-β (Aβ). Male Wistar rats (n = 70) were randomly allocated to the control, sham, AD, control-GR (100 mg/kg; per oral: P.O.), AD-GR (100 mg/kg; P.O.; treatment), GR-AD (100 mg/kg; P.O.; pretreatment), and GR-AD-GR (100 mg/kg; P.O.; pretreatment + treatment) groups. GR administration was done for four continuous weeks. After treatments, novel object recognition (NOR) and Morris water maze (MWM) tests assessed the animals' behavior. Then, hippocampal specimens were collected for biochemical assessment. Finally, the number of intact neurons was identified in the hippocampus using hematoxylin and eosin staining. Aβ microinjection increased learning and memory deficits in both NOR and MWM tests, oxidative stress status, and neuronal loss. Oral GR administration improved behavioral deficits and reduced oxidative stress status and neuronal loss in the Aβ-infused animals. GR ameliorates behavioral impairments through a decrease in neuronal degeneration and oxidative stress.

The Mechanism of Propofol in Treating Depression-like Behaviors Based on Network Pharmacology and Experimental Validation

Objective: To explore the potential mechanism of propofol in improving antidepressant-like behavior through network pharmacology and animal experimental validation, providing theoretical and experimental support for the development of novel antidepressant drugs based on propofol. Methods: 1. The targets of propofol and the disease targets of depression were screened through Swiss Target prediction, Super-pred, SEA, Drugbank, CTD, GeneCards, OMIM, and TTD databases. A protein-protein interaction (PPI) network was constructed, and the core targets were screened and visualized using CytoScape 3.10.2 software. Additionally, GO and KEGG enrichment analyses of the intersection targets were performed using the Metascape database. 2. Thirty male C57BL/6 mice were randomly divided into the CON group, LH model group, and PRO treatment group, with 10 mice in each group. The sucrose preference test, forced swim test, and tail suspension test were conducted to determine whether the depression model was successfully established and to assess the improvement effect of propofol on antidepressant-like behavior. 3. The pathological morphological changes of hippocampal neurons in the three groups of mice were observed through HE staining experiments. The expression of mGluR5 protein in the hippocampus was analyzed by Western blot. Results: 1. Network pharmacology analysis indicated that propofol may exert antidepressant effects by acting on core genes such as ALB, ESR1, NFKB1, HSP90AB1, and EGFR. These core target genes were mainly enriched in biological processes such as synaptic transmission and membrane potential regulation; they involved cellular components such as GABA receptor complexes and synaptic membranes; and they included molecular functions such as neurotransmitter receptor activity. Additionally, propofol may exert antidepressant effects through signaling pathways such as neuroactive ligand-receptor interaction, morphine addiction, and GABAergic synapse. 2. Compared with the CON group, the sucrose preference rate of mice in the LH model group significantly decreased, while the immobility time in the forced swim test and tail suspension test significantly increased. Compared with the LH model group, the sucrose preference rate of mice in the PRO treatment group significantly increased, and the immobility time significantly decreased. 3. The HE results showed that compared with the CON group, the number of neurons in the LH group decreased, with loose arrangement, pycnotic and deeply stained nuclei with blurred boundaries. However, the number of necrotic neurons in the PRO group significantly decreased. The Western blot results showed that compared with the CON group, the expression level of mGluR5 protein in the LH model group significantly increased, while it significantly decreased in the PRO treatment group compared with the LH model group. Conclusion: Propofol may exert antidepressant effects by regulating the expression of mGluR5, improving antidepressant-like behavior, and reducing hippocampal neuronal necrosis.

Substantia nigra alterations in mice modeling Parkinson’s disease

Substantia nigra alterations in mice modeling Parkinson’s disease

Imaging the large-scale and cellular response to focal traumatic brain injury in mouse neocortex.

Traumatic brain injury (TBI) affects neural function at the local injury site and also at distant, connected brain areas. However, the real-time neural dynamics in response to injury and subsequent effects on sensory processing and behaviour are not fully resolved, especially across a range of spatial scales. We used in vivo calcium imaging in awake, head-restrained male and female mice to measure large-scale and cellular resolution neuronal activation, respectively, in response to a mild/moderate TBI induced by focal controlled cortical impact (CCI) injury of the motor cortex (M1). Widefield imaging revealed an immediate CCI-induced activation at the injury site, followed by a massive slow wave of calcium signal activation that travelled across the majority of the dorsal cortex within approximately 30s. Correspondingly, two-photon calcium imaging in the primary somatosensory cortex (S1) found strong activation of neuropil and neuronal populations during the CCI-induced travelling wave. A depression of calcium signals followed the wave, during which we observed the atypical activity of a sparse population of S1 neurons. Longitudinal imaging in the hours and days after CCI revealed increases in the area of whisker-evoked sensory maps at early time points, in parallel to decreases in cortical functional connectivity and behavioural measures. Neural and behavioural changes mostly recovered over hours to days in our M1-TBI model, with a more lasting decrease in the number of active S1 neurons. Our results in unanaesthetized mice describe novel spatial and temporal neural adaptations that occur at cortical sites remote to a focal brain injury.

Therapeutic potential of Simvastatin in ALS: Enhanced axonal integrity and motor neuron survival through Apoa4 and Alb modulation.

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease characterized by the selective death of motor neurons in the spinal cord, brainstem, and motor cortex. This study investigates the effects of simvastatin on the G93A-copper/zinc superoxide dismutase (G93ASOD1) transgenic mouse model of ALS. The experiment included three groups: C57BL/6 wild-type (WT) mice, C57BL/6J SOD1G93A mice treated with PBS (SOD1G93A + PBS), and C57BL/6J SOD1G93A mice treated with simvastatin (SOD1G93A + simvastatin). The primary endpoints were survival rates, body weight changes, performance in pole climbing and suspension tests, and neurological deficit scores. Pathological changes were assessed using H&E staining, transmission electron microscopy, Nissl staining, and Masson staining. Proteomic and metabolomic analyses were performed to identify differentially expressed proteins (DEPs) and metabolites. Quantitative real-time polymerase chain reaction (qRT-PCR) and western blotting were used to measure gene expression. Although there were no significant differences in survival rates, body weight, pole climbing, and suspension test performance, or neurological deficit scores between the SOD1G93A + simvastatin and SOD1G93A + PBS groups, simvastatin treatment improved axonal organization within the spinal cord, increased the number of neurons, and reduced cytoplasmic swelling and gastrocnemius fibrosis. A total of 47 DEPs and 13 differential metabolites were identified between the SOD1G93A + PBS and SOD1G93A + simvastatin groups. Notably, the expression levels of Apoa4 and Alb were elevated in the SOD1G93A + simvastatin group compared to the SOD1G93A + PBS group. Our results suggest that simvastatin may have potential therapeutic effects in ALS, likely involving the modulation of Apoa4 and Alb expression.

Tlx Promotes Stroke-Induced Neurogenesis and Neuronal Repair in Young and Aged Mice.

Stroke is one of the leading causes of chronic disability in humans. It has been proposed that the endogenous neural stem/progenitor cells generate new neurons in the damaged area. Still, the contribution of these cells is negligible because a low number of newborn mature neurons are formed. Tlx conventional knock-out mice, Tlx-CreERT2 mice, and Tlx-overexpressing (Tlx-OE) mice were specifically chosen for their unique genetic characteristics, which were crucial for the experiments. Permanent and transient middle cerebral artery occlusion was used to induce stroke in the mice. Immunostainings for doublecortin and GFP/BrdU/NeuN were performed to study neurogenesis and fate mapping. The rotarod test was performed to assess motor deficits. Here, we show that stroke-induced neurogenesis is dramatically increased with the additional expression of two copies of the nuclear receptor-coding gene tailless (Tlx, also known as Nr2e1), which has been shown to be a master regulator of subventricular zone (SVZ) neural stem cells (NSCs). We show that Tlx expression is upregulated after stroke, and stroke-induced neurogenesis is blocked when Tlx is inactivated. Tlx overexpression in NSCs leads to massive induction of neurogenesis via stroke. More newborn mature neurons are formed in Tlx-overexpressing mice, leading to improved coordination and motor function recovery. Most importantly, we also demonstrate that this process is sustained in aged mice, where stroke-induced neurogenesis is nearly undetectable in wild-type animals. This study provides the first stem cell-specific genetic evidence that endogenous NSCs can be exploited by manipulating their master regulator, Tlx, and thus suggests a novel therapeutic strategy for neuronal repair.

Neural network embedding of functional microconnectome

Abstract Our brains operate as a complex network of interconnected neurons. To gain a deeper understanding of this network architecture, it is essential to extract simple rules from its intricate structure. This study aimed to compress and simplify the architecture, with a particular focus on interpreting patterns of functional connectivity in 2.5 hr of electrical activity from a vast number of neurons in acutely sliced mouse brains. Here, we combined two distinct methods together: automatic compression and network analysis. Firstly, for automatic compression, we trained an artificial neural network named NNE (neural network embedding). This allowed us to reduce the connectivity to features, be represented only by 13% of the original neuron count. Secondly, to decipher the topology, we concentrated on the variability among the compressed features and compared them with 15 distinct network metrics. Specifically, we introduced new metrics that had not previously existed, termed as indirect-adjacent degree and neighboring hub ratio. Our results conclusively demonstrated that these new metrics could better explain approximately 40%–45% of the features. This finding highlighted the critical role of NNE in facilitating the development of innovative metrics, because some of the features extracted by NNE were not captured by the currently existed network metrics.

A proof of convergence and equivalence for 1D finite element methods and ReLU neural networks

This paper investigates the convergence and equivalence properties of the Finite Element Method (FEM) and Rectified Linear Unit Neural Networks (ReLU NNs) in solving differential equations. We provide a detailed comparison of the two approaches, highlighting their mutual capabilities in function space approximation. Our analysis proves the subset and superset inclusions that establish the equivalence between FEM and ReLU NNs for approximating piecewise linear functions. Furthermore, a comprehensive numerical evaluation is presented, demonstrating the error convergence behavior of ReLU NNs as the number of neurons per basis function varies. Our results show that while increasing the number of neurons improves approximation accuracy, this benefit diminishes beyond a certain threshold. The maximum observed error between FEM and ReLU NNs is 10-4, reflecting excellent accuracy in solving partial differential equations (PDEs). These findings lay the groundwork for integrating FEM and ReLU NNs, with important implications for computational mathematics and engineering applications.

Research on the Quality Grading Method of Ginseng with Improved DenseNet121 Model

Ginseng is an important medicinal plant widely used in traditional Chinese medicine. Traditional methods for evaluating the visual quality of ginseng have limitations. This study presents a new method for grading ginseng’s appearance quality using an improved DenseNet121 model. We enhance the network’s capability to recognize various channel features by integrating a CA (Coordinate Attention) mechanism. We also use grouped convolution instead of standard convolution in dense layers to lower the number of model parameters and improve efficiency. Additionally, we substitute the ReLU (Rectified Linear Unit) activation function with the ELU (Exponential Linear Unit) activation function, which reduces the problem of neuron death related to ReLU and increases the number of active neurons. We compared several network models, including DenseNet121, ResNet50, ResNet101, GoogleNet, and InceptionV3, to evaluate their performance against our method. Results showed that the improved DenseNet121 model reached an accuracy of 95.5% on the test set, demonstrating high reliability. This finding provides valuable support for the field of ginseng grading.

Decoding Continuous Tracking Eye Movements from Cortical Spiking Activity.

Eye movements are the primary way primates interact with the world. Understanding how the brain controls the eyes is therefore crucial for improving human health and designing visual rehabilitation devices. However, brain activity is challenging to decipher. Here, we leveraged machine learning algorithms to reconstruct tracking eye movements from high-resolution neuronal recordings. We found that continuous eye position could be decoded with high accuracy using spiking data from only a few dozen cortical neurons. We tested eight decoders and found that neural network models yielded the highest decoding accuracy. Simpler models performed well above chance with a substantial reduction in training time. We measured the impact of data quantity (e.g. number of neurons) and data format (e.g. bin width) on training time, inference time, and generalizability. Training models with more input data improved performance, as expected, but the format of the behavioral output was critical for emphasizing or omitting specific oculomotor events. Our results provide the first demonstration, to our knowledge, of continuously decoded eye movements across a large field of view. Our comprehensive investigation of predictive power and computational efficiency for common decoder architectures provides a much-needed foundation for future work on real-time gaze-tracking devices.

Predicting Operational Events in Mechanized Weed Control Operations by Offline Multi-Modal Data and Machine Learning Provides Highly Accurate Classification in Time Domain

Monitoring of operations has become a critical activity in forestry, aiming to provide the data required by planning and production management. Conventional methods, on the other hand, come at a high expense of resources. A neural network was trained, validated, and tested in this study based on multi-modal data to classify relevant operational events in mechanized weed control operations. The architecture of a neural network was tuned in terms of the number of hidden layers and neurons, and the regularization term was set at various values to obtain optimally tuned models for three data modalities: triaxial acceleration data coupled with speed extracted from GNSS signals (AS), triaxial acceleration (A), and speed alone (S). In the training and validation phase, the models based on AS and A achieved a very high classification accuracy, accounting for 92 to 93% when considering four relevant events. In the testing phase, which was run on unseen data, the classification accuracy reached figures of 91 to 92%, indicating a good generalization ability of the models. The results point out that multimodal data are able to provide the features for distinguishing events and add spatial context to the monitored operations, standing as a suitable solution for offline, partly automated monitoring. Future studies are required to see how the capabilities of online, real-time technologies such as deep learning coupled with computer vision can add more context and improve classification performance.

Using Machine Learning for Climate Modelling: Application of Neural Networks to a Slow-Fast Chaotic Dynamical System as a Case Study

This paper explores the capabilities of two types of recurrent neural networks, unidirectional and bidirectional long short-term memory networks, to build a surrogate model for a coupled fast–slow dynamic system and predicting its nonlinear chaotic behaviour. The dynamical system in question, comprising two versions of the classical Lorenz model with a small time-scale separation factor, is treated as an atmosphere–ocean research simulator. In numerical experiments, the number of hidden layers and the number of nodes in each hidden layer varied from 1 to 5 and from 16 to 256, respectively. The basic configuration of the surrogate model, determined experimentally, has three hidden layers, each comprising between 16 and 128 nodes. The findings revealed the advantages of bidirectional neural networks over unidirectional ones in terms of forecasting accuracy. As the forecast horizon increases, the accuracy of forecasts deteriorates, which was quite expected, primarily due to the chaotic behaviour of the fast subsystem. All other things being equal, increasing the number of neurons in hidden layers facilitates the improvement of forecast accuracy. The obtained results indicate that the quality of short-term forecasts with a lead time of up to 0.75 model time units (MTU) improves most significantly. The predictability limit of the fast subsystem (“atmosphere”) is somewhat greater than the Lyapunov time.

Loss of Dnah5 Downregulates Dync1h1 Expression, Causing Cortical Development Disorders and Congenital Hydrocephalus.

Dnah5 is associated with primary ciliary dyskinesia in humans. Dnah5-knockout (Dnah5-/- mice develop acute hydrocephalus shortly after birth owing to impaired ciliary motility and cerebrospinal fluid (CSF) stagnation. In contrast to chronic adult-onset hydrocephalus observed in other models, this rapid ventricular enlargement indicates additional factors beyond CSF stagnation. Herein, we investigated the contributors to rapid ventricular enlargement in congenital hydrocephalus. Dnah5-/- mice were generated using CRISPR/Cas9. The expression of dynein, N-cadherin, and nestin in the cerebral cortex was assessed using microarrays and immunostaining. Real-time PCR and Western blotting were performed for gene and protein quantification, respectively. All Dnah5-/- mice developed hydrocephalus, confirmed by electron microscopy, indicating the absence of axonemal outer dynein arms. Ventricular enlargement occurred rapidly, with a 25% reduction in the number of mature neurons in the motor cortex. Dync1h1 expression was decreased, while cytoplasmic dynein levels were 56.3% lower. Levels of nestin and N-cadherin in the lateral ventricular walls decreased by 31.7% and 33.3%, respectively. Reduced cytoplasmic dynein disrupts neurogenesis and axonal growth and reduces neuron cortical density. Hydrocephalus in Dnah5-/- mice may result from cortical maldevelopment due to cytoplasmic dynein deficiency, further exacerbating ventricular enlargement due to CSF stagnation caused by impaired motile ciliary function.

Robust neural networks using stochastic resonance neurons

Various successful applications of deep artificial neural networks are effectively facilitated by the possibility to increase the number of layers and neurons in the network at the expense of the growing computational complexity. Increasing computational complexity to improve performance makes hardware implementation more difficult and directly affects both power consumption and the accumulation of signal processing latency, which are critical issues in many applications. Power consumption can be potentially reduced using analog neural networks, the performance of which, however, is limited by noise aggregation. Following the idea of physics-inspired machine learning, we propose here a type of neural network using stochastic resonances as a dynamic nonlinear node and demonstrate the possibility of considerably reducing the number of neurons required for a given prediction accuracy. We also observe that the performance of such neural networks is more robust against the impact of noise in the training data compared to conventional networks.

Effects of Shugan Tiaoshen acupuncture on anxiety-like behavior and PKC/ERK/CREB pathway in the bed nucleus of the stria terminalis in rats with post-traumatic stress disorder

To observe the effect of Shugan Tiaoshen acupuncture (acupuncture for soothing the liver and regulating the spirit) on the protein kinase C/extracellular signal-regulated kinase/cAMP response element-binding protein (PKC/ERK/CREB) signaling pathway in the bed nucleus of the stria terminalis (BNST) of rats with post-traumatic stress disorder (PTSD), and to explore the mechanism of acupuncture on alleviating anxiety and fear in PTSD. Fifty SPF-grade male SD rats were randomly divided into a blank group (10 rats) and a PTSD model group (40 rats). The PTSD model was induced by using a combination of closed electric shock and forced exhaustive swimming. Thirty successfully modeled rats were randomly assigned to a model group, a medication group, and an acupuncture group, with 10 rats in each group. The rats in the medication group were treated with paroxetine hydrochloride solution by gavage, once daily for 12 consecutive days. The rats in the acupuncture group were treated with acupuncture at "Baihui" (GV 20) and bilateral "Neiguan" (PC 6), "Shenmen" (HT 7), "Taichong" (LR 3). "Baihui" (GV 20) was needled daily, while the other acupoints were alternately needled on the left side on odd days and the right side on even days, once daily for 12 consecutive days. Anxiety and fear behaviors changes were assessed by using the open field test and elevated plus maze test. Histological changes in the BNST were observed by using HE staining and Nissl staining. The expression of PKC, phosphorylated PKC (p-PKC), ERK1/2, phosphorylated ERK1/2 (p-ERK1/2), and p-CREB proteins in the BNST were detected by using Western blot. Compared with the blank group, the model group showed decreased time and total distance spent in the center of the open field and on the open arms of the elevated plus maze (P<0.05); the BNST tissues in the model group exhibited a reduced number of neurons, disorganized cell arrangement, cell shrinkage, nuclear condensation, abnormal neuronal structure, uneven Nissl staining, and reduced Nissl bodies. The model group showed increased protein expression of p-PKC and p-PKC/PKC ratio (P<0.05) and decreased protein expression of p-ERK1/2, p-CREB, and p-ERK1/2/ERK1/2 ratio (P<0.05). Compared with the model group, the medication group and the acupuncture group showed increased time and total distance spent in the center of the open field and on the open arms of the elevated plus maze (P<0.05); the BNST tissues showed increased number of neurons, more organized cell arrangement, improved neuronal structure, and increased Nissl bodies; the medication group and the acupuncture group also showed decreased p-PKC protein expression and p-PKC/PKC ratio (P<0.05) and increased p-ERK1/2, p-CREB protein expression, and p-ERK1/2/ERK1/2 ratio (P<0.05). Shugan Tiaoshen acupuncture could alleviate anxiety and fear behaviors in PTSD rats, and improve neuronal damage in the BNST. The mechanism may be related to the regulation of the PKC/ERK/CREB signaling pathway in the BNST.