Robust Applications Research Articles

An Ensemble Framework Augmenting Surveillance Cameras for Detecting Intruder Clusters as Potential Mobs

<p>Many developing nations around the world curtail crimes through video surveillance technology, but the crime rate is still high. This is compounded by short-staffed security operatives and a deficiency of security infrastructure to assist security operatives with knowledge-driven decision support systems in the low-resource constraint environment. In a public environment, it is challenging to detect intruder clusters accurately as potential mobs for early warning. Previous research investigated some classical techniques, but their recommendations were insufficient. This research develops a Machine Learning (ML) 3-tiers Ensemble framework, which integrates Gray Level Co-occurrence Matrices (GLCM) principles to enhance the capabilities of surveillance cameras and security operatives to effectively discern and respond to potential mob formations. The University of California San Diego (UCSD) pedestrian datasets that are publicly available were used for the experiments. With an improved overall average precision of 0.98, recall of 0.98, and accuracy of 98.52% on the UCSD dataset, the suggested framework outperforms the widely used methods for the detection of intruder clusters. The reduction in computational time on processors showcases the framework's significant advancements as a promising solution for robust real-time threat assessment applications.</p>

Leveraging Shapley Additive Explanations for Feature Selection in Ensemble Models for Diabetes Prediction

Diabetes, a significant global health crisis, is primarily driven in India by unhealthy diets and sedentary lifestyles, with rapid urbanization amplifying these effects through convenience-oriented living and limited physical activity opportunities, underscoring the need for advanced preventative strategies and technology for effective management. This study integrates Shapley Additive explanations (SHAPs) into ensemble machine learning models to improve the accuracy and efficiency of diabetes predictions. By identifying the most influential features using SHAP, this study examined their role in maintaining high predictive performance while minimizing computational demands. The impact of feature selection on model accuracy was assessed across ten models using three feature sets: all features, the top three influential features, and all except these top three. Models focusing on the top three features achieved superior performance, with the ensemble model attaining a better performance in most of the metrics, outperforming comparable approaches. Notably, excluding these features led to a significant decline in performance, reinforcing their critical influence. These findings validate the effectiveness of targeted feature selection for efficient and robust clinical applications.

Serum Irisin as a Predictor for Peripheral arterial disease: Insights from a Clinical Study

ObjectiveThe purpose of this study was to investigate the correlation between serum irisin concentration and peripheral arterial disease (PAD), and to establish clinical prediction nomograms for PAD occurrence by comparing and analyzing clinical data from patients with PAD patients and healthy controls. MethodsA total of 112 patients with PAD and 90 healthy individuals were recruited for the study. Clinical data from both groups were collected and serum irisin concentration was measured using enzyme-linked immunosorbent assay (ELISA). Correlation analysis was conducted. Risk factors for PAD were identified through univariate and multivariate logistic regression. The clinical prediction nomograms were established and validated. ResultsA total of 202 patients were enrolled in this study, with an average age of 63.98 ± 10.40 years. Of these, 123 were male (60.9%) and 79 were female (39.1%). Hypertension was present in 104 patients (51.5%), diabetes in 59 patients (29.2%), dyslipidemia in 94 patients (46.5%), and 105 patients (52.0%) were smokers. Among them, 90 patients were assigned to the PAD group, which included 78 males (69.6%) and 34 females (30.4%), with an average age of 67.54 ± 10.31 years. In this group, 62 patients (55.4%) had hypertension, 53 (47.3%) had diabetes, 62 (55.4%) had dyslipidemia, and 78 (69.6%) were smokers. The Rutherford classification of these patients showed that 64 (57.1%) were at stage I, 25 (22.3%) at stage II, 16 (14.3%) at stage III, and 7 (6.3%) at stage IV. Serum irisin concentration in patients with PAD showed a significant positive correlation with serum high-density lipoprotein (HDL) (r=0.255) and a significant negative correlation with Rutherford classification (r=-0.374) and smoking status (r=-0.263). Univariate and multivariate logistic regression analyses identified irisin, age, diabetes, dyslipidemia, smoking, creatinine(CR), and neutrophil/lymphocyte ratio (NLR) as independent risk factors for the development of PAD ( P < 0.05). Based on these findings, a clinical prediction nomogram was established. Internal validation of the nomogram demonstrated strong discriminatory ability, with an area under the curve (AUC) of 0.942, indicating the model's excellent performance. Calibration curves and decision curve analyses further confirmed the model’s robust calibration and clinical applicability. ConclusionThis study concluded that serum irisin concentrations were significantly lower in the PAD group compared to the healthy control group, and that serum irisin concentrations in the PAD group were significantly correlated with serum HDL, Rutherford classification, and smoking status. Additionally, Irisin level, age, diabetes, dyslipidemia, smoking, CR, and NLR were identified as independent risk factors for PAD development. The clinical prediction nomogram based on these factors may aid in accurately predicting the risk of PAD development.

Accurate depth of anesthesia monitoring based on EEG signal complexity and frequency features

Accurate monitoring of the depth of anesthesia (DoA) is essential for ensuring patient safety and effective anesthesia management. Existing methods, such as the Bispectral Index (BIS), are limited in real-time accuracy and robustness. Current methods have problems in generalizability across diverse patient datasets and are sensitive to artifacts, making it difficult to provide reliable DoA assessments in real time. This study proposes a novel method for DoA monitoring using EEG signals, focusing on accuracy, robustness, and real-time application. EEG signals were pre-processed using wavelet denoising and discrete wavelet transform (DWT). Features such as Permutation Lempel–Ziv Complexity (PLZC) and Power Spectral Density (PSD) were extracted. A random forest regression model was employed to estimate anesthetic states, and an unsupervised learning method using the Hurst exponent algorithm and hierarchical clustering was introduced to detect transitions between anesthesia states. The method was tested on two independent datasets (UniSQ and VitalDB), achieving an average Pearson correlation coefficient of 0.86 and 0.82, respectively. For the combined dataset, the model demonstrated an R-squared value of 0.70, a RMSE of 6.31, a MAE of 8.38, and a Pearson correlation of 0.84, showcasing its robustness and generalizability. This approach offers a more accurate and reliable real-time DoA monitoring tool that could significantly improve patient safety and anesthesia management, especially in diverse clinical environments.

Machine learning-driven prediction of tensile strength in 3D-printed PLA parts

Additive manufacturing (AM) has become a transformative technology in modern production, enabling complex geometric designs with minimal material waste. A significant aspect of AM, particularly in fused deposition modeling (FDM), is the need for precise prediction of mechanical properties, such as ultimate tensile strength (UTS), which is crucial for industrial applications. This study examines whether simple machine learning (ML) algorithms can accurately predict the UTS of 3D-printed polylactic acid (PLA) parts, and evaluates the effectiveness of ML techniques, especially ensemble methods, in enhancing prediction accuracy. To this end, the study compares simple ML algorithms to identify the most accurate model for predicting the UTS of 3D-printed PLA parts. Subsequently, an average ensemble technique combines four ML algorithms, namely categorical boosting (CatBoost), extreme gradient boosting (XGBoost), gradient boosting machine (GBM), and light gradient boosting machine (LGBM), to predict UTS. In this technique, the average predicted UTS values of CatBoost, XGBoost, GBM, and LGBM are taken as the final predicted UTS value. Additionally, 11 ensemble configurations of these algorithms are analyzed to determine the optimized ensemble configuration. The results show that the CatBoost algorithm, with an R2 of 94.46%, achieved the highest predictive accuracy among individual ML algorithms. Moreover, the CatBoost-XGBoost-GBM-LGBM ensemble was the most accurate configuration, achieving an R2 of 98.05% with less than 10% error in predicting 37 external data points not included in the training and testing sets. This study advances predictive modeling in AM by demonstrating that ML, particularly ensemble techniques, can reliably predict material properties, paving the way for more robust applications of AM in industry

Boosting the Microbial Electrosynthesis of Formate by Shewanella oneidensis MR-1 with an Ionic Liquid Cosolvent.

Microbial electrosynthesis (MES) is a rapidly growing technology at the forefront of sustainable chemistry, leveraging the ability of microorganisms to catalyze electrochemical reactions to synthesize valuable compounds from renewable energy sources. The reduction of CO2 is a major target application for MES, but research in this area has been stifled, especially with the use of direct electron transfer (DET)-based microbial systems. The major fundamental hurdle that needs to be overcome is the low efficiency of CO2 reduction largely attributed to minimal microbial access to CO2 owing to its low solubility in the electrolyte. With their tunable physical properties, ionic liquids present a potential solution to this challenge and have previously shown promise in facilitating efficient CO2 electroreduction by increasing the CO2 solubility. However, the use of ionic liquids in MES remains unexplored. In this study, we investigated the role of 1-ethyl-3-methylimidazolium acetate ([EMIM][Ac]) using Shewanella oneidensis MR-1 as a model DET strain. Electrochemical investigations demonstrated the ability of S. oneidensis MR-1 biocathodes to directly convert CO2 to formate with a faradaic efficiency of 34.5 ± 26.1%. The addition of [EMIM][Ac] to the system significantly increased cathodic current density and enhanced the faradaic efficiency to 94.5 ± 4.3% while concurrently amplifying the product yield from 34 ± 23 μM to 366 ± 34 μM. These findings demonstrate that ionic liquids can serve as efficient, biocompatible cosolvents for microbial electrochemical reduction of CO2 to value-added products, holding promise for more robust applications of MES.

Uncertainty-aware genomic deep learning with knowledge distillation.

Deep neural networks (DNNs) have advanced predictive modeling for regulatory genomics, but challenges remain in ensuring the reliability of their predictions and understanding the key factors behind their decision making. Here we introduce DEGU (Distilling Ensembles for Genomic Uncertainty-aware models), a method that integrates ensemble learning and knowledge distillation to improve the robustness and explainability of DNN predictions. DEGU distills the predictions of an ensemble of DNNs into a single model, capturing both the average of the ensemble's predictions and the variability across them, with the latter representing epistemic (or model-based) uncertainty. DEGU also includes an optional auxiliary task to estimate aleatoric, or data-based, uncertainty by modeling variability across experimental replicates. By applying DEGU across various functional genomic prediction tasks, we demonstrate that DEGU-trained models inherit the performance benefits of ensembles in a single model, with improved generalization to out-of-distribution sequences and more consistent explanations of cis-regulatory mechanisms through attribution analysis. Moreover, DEGU-trained models provide calibrated uncertainty estimates, with conformal prediction offering coverage guarantees under minimal assumptions. Overall, DEGU paves the way for robust and trustworthy applications of deep learning in genomics research.

Robustness in deep learning models for medical diagnostics: security and adversarial challenges towards robust AI applications

The current study investigates the robustness of deep learning models for accurate medical diagnosis systems with a specific focus on their ability to maintain performance in the presence of adversarial or noisy inputs. We examine factors that may influence model reliability, including model complexity, training data quality, and hyperparameters; we also examine security concerns related to adversarial attacks that aim to deceive models along with privacy attacks that seek to extract sensitive information. Researchers have discussed various defenses to these attacks to enhance model robustness, such as adversarial training and input preprocessing, along with mechanisms like data augmentation and uncertainty estimation. Tools and packages that extend the reliability features of deep learning frameworks such as TensorFlow and PyTorch are also being explored and evaluated. Existing evaluation metrics for robustness are additionally being discussed and evaluated. This paper concludes by discussing limitations in the existing literature and possible future research directions to continue enhancing the status of this research topic, particularly in the medical domain, with the aim of ensuring that AI systems are trustworthy, reliable, and stable.

Transferable and Interpretable Prediction of Site-Specific Dehydrogenation Reaction Rate Constants with NMR Spectra.

Accurate and efficient determination of site-specific reaction rate constants over a wide temperature range remains challenging, both experimentally and theoretically. Taking the dehydrogenation reaction as an example, our study addresses this issue by an innovative combination of machine learning techniques and cost-effective NMR spectra. Through descriptor screening, we identified a minimal set of NMR chemical shifts that can effectively determine reaction rate constants. The constructed model performs exceptionally well on theoretical data sets and demonstrates impressive generalization capabilities, extending from small molecules to larger ones. Furthermore, this model shows outstanding performance when applied to limited experimental data sets, highlighting its robust applicability and transferability. Utilizing the Sure Independence Screening and Sparsifying Operator (SISSO) algorithm, we also present an interpretable rate constant-temperature-NMR (k-T-NMR) relationship with a mathematical formula. This study reveals the great potential of combining machine learning with easily accessible spectroscopic descriptors in the study of reaction kinetics, enabling the rapid determination of reaction rate constants and promoting our understanding of reactivity.

Extreme Environment-Adaptable and Ultralong-Life Energy Storage Enabled by Synergistic Manipulation of Interfacial Environment and Hydrogen Bonding

The broad applications of energy storage systems have brought improving demands for stable electrodes with robust tolerance to extreme environmental challenges. MXenes show promising pseudocapacitive behaviors, however, the poor thermodynamical and mechanical stability makes them unfavorable for applications under complex and harsh environments. Herein, we break these limitations by aramid nanofibers (ANF)-driven interfacial nanofilling and hydrogen-bonds effects in MXenes. Theoretical and experimental results unveil that ANF with unique polarity preferentially interacts with H2O molecules and forms hydrogen bonding networks to restrain oxidative and mechanical attack toward MXene, at the same time, the nanofilling enables interfacial mass transport intensification for increment in redox dynamics. As such, the synergistically coupled ANF-MXene microstructure (AM) unlocks superior mechanical properties for facing hash forces, i.e., tensile strength of 115.2 MPa and toughness of 1.8 MJ m-3, and an ultra-long cycling life with a capacitance retention of 90.7% after 40,000 cycles. Besides, the effective IR thermal camouflage performance (IR-emissivity: 20.9%) further renders the power supply working invisibly after fast charge/discharge-driven heat generation. Moreover, the performances can be well maintained when subjected to strong acid/alkali, high-temperature (200°C), and cryogenic (-196°C) treatments. These results highlight the key role of interfacial species synergy in accelerating versatile and robust energy applications.

High-dimensional Bayesian likelihood normalisation for CRESST's background model

Using CaWO4 crystals as cryogenic calorimeters, the CRESST experiment searches for nuclear recoils caused by the scattering of potential Dark Matter particles. A reliable identification of a potential signal crucially depends on an accurate background model. In this work, we introduce an improved normalisation method for CRESST's model of electromagnetic backgrounds, which is an important technical step towards developing a more accurate background model. Spectral templates based on Geant4 simulations are normalised via a Bayesian likelihood fit to experimental background data. Contrary to our previous work, no explicit assumption of partial secular equilibrium is required a priori, which results in a more robust and versatile applicability. This new method also naturally considers the correlation between all background components. Due to these purely technical improvements, the presented method has the potential to explain up to 82.7 % of the experimental background within [1 keV,40 keV], an improvement of at most 18.6 % compared to our previous method. The actual value is subject to ongoing validations of the included physics.

On b-Coloring Analysis of Graphs: An Application to Spatial-Temporal Graph Neural Networks for Multi-Step Time Series Forecasting of Soil Moisture and pH in Companion Farming

Let $G$ be a pair of two sets $(V,E)$ with vertex set $V$ and edge set $E$. A proper coloring of a graph $G$ is a vertex coloring of it such that no two adjacent vertices in $G$ have the same color. By $b-$Coloring, we define a coloring of the vertex of $G$ such that each color class has at least one vertex that adjacent with all other color classes. The $b-chromatic$ number of graph $G$, denoted by $\varphi(G),$ is the largest integer $k$ such that graph $G$ has $b-$Coloring with $k$ colors. In this paper, we will explore some new lemmas or theorems regarding to $\varphi(G)$. Furthermore, to see the robust application of $b-$Coloring of graph, at the end of this paper we will illustrate the implementation of $b-$Coloring on spatial temporal graph neural networks (STGNN) multi-step time series forecasting on soil moisture and $pH$ of companion farming.

In-situ formation of La2O2CO3 uniformly immobilized on biomass-derived carbon aerogel for enhanced phosphate removal

In the face of the dual pressures of eutrophication and phosphorus shortages, it is of utmost importance to efficiently remove and recover phosphate from wastewater. In this study, waste-biomass-derived carbon aerogel (CA) was utilized as a dispersing carrier for lanthanum (La), resulting in the successful in-situ generation of an La2O2CO3 (La3.5-CA-450) phosphate adsorbent on the CA matrix. The results demonstrated that the introduction of CA as an adsorbent led to abundant carbon defects and pore structures, which significantly enhanced its performance. Moreover, through the coordinated pyrolysis of CA and La3+, CO32−, capable of exchanging with PO43−, was generated, thereby contributing to the exceptional phosphate adsorption capacity exhibited by the adsorbent. Specifically, La3.5-CA-450 displayed a high phosphate adsorption capacity of 155.8 mg P/g (272.4 mg P/g La), with the molar ratio of P to La being 1.22:1, indicating a high utilization efficiency for La. The adsorbent exhibited exceptional performances in terms of selectivity, renewability, efficient use in natural water environments, and robust application in a broad range of synthetic availability. This research adheres to the green concept “using waste to treat waste” while providing a novel approach for preparing efficient La-based phosphate adsorbents.

Artificial Intelligence and Artificial Stupidity: The Inseparables

Artificial intelligence (AI) has long been heralded for its ability to simulate human intelligence, enabling machines to perform complex tasks such as decision-making, problem-solving, and data analysis. However, alongside the advancements in AI, the concept of artificial stupidity (AS) has gained attention. AS refers to the limitations and errors made by AI systems, often resulting from incomplete data, biased algorithms, or the inherent restrictions placed on AI to simulate more human-like decision-making. These instances of "stupidity" can lead to nonsensical or harmful outcomes, especially when AI is applied to critical areas such as healthcare, autonomous systems, and legal decision-making. This narrative review explores the duality between AI's potential and its flaws, emphasizing the importance of understanding both AI and AS in developing robust, safe, and ethical AI applications. By addressing the causes of artificial stupidity, such as algorithmic limitations and poor data quality, researchers and developers can improve the reliability and decision-making capabilities of AI systems. There is also the need for human oversight and ethical considerations to mitigate the negative impacts of artificial stupidity, especially in high-stakes environments.

RRBM-YOLO: Research on Efficient and Lightweight Convolutional Neural Networks for Underground Coal Gangue Identification.

Coal gangue identification is the primary step in coal flow initial screening, which mainly faces problems such as low identification efficiency, complex algorithms, and high hardware requirements. In response to the above, this article proposes a new "hardware friendly" coal gangue image recognition algorithm, RRBM-YOLO, which is combined with dark light enhancement. Specifically, coal gangue image samples were customized in two scenarios: normal lighting and simulated underground lighting with poor lighting conditions. The images were preprocessed using the dim light enhancement algorithm Retinexformer, with YOLOv8 as the backbone network. The lightweight module RepGhost, the repeated weighted bi-directional feature extraction module BiFPN, and the multi-dimensional attention mechanism MCA were integrated, and different datasets were replaced to enhance the adaptability of the model and improve its generalization ability. The findings from the experiment indicate that the precision of the proposed model is as high as 0.988, the mAP@0.5(%) value and mAP@0.5:0.95(%) values increased by 10.49% and 36.62% compared to the original YOLOv8 model, and the inference speed reached 8.1GFLOPS. This indicates that RRBM-YOLO can attain an optimal equilibrium between detection precision and inference velocity, with excellent accuracy, robustness, and industrial application potential.

Connecting the Brain with Augmented Reality: A Systematic Review of BCI-AR Systems

The increasing integration of brain–computer interfaces (BCIs) with augmented reality (AR) presents new possibilities for immersive and interactive environments, particularly through the use of head-mounted displays (HMDs). Despite the growing interest, a comprehensive understanding of BCI-AR systems is still emerging. This systematic review aims to synthesize existing research on the use of BCIs for controlling AR environments via HMDs, highlighting the technological advancements and challenges in this domain. An extensive search across electronic databases, including IEEEXplore, PubMed, and Scopus, was conducted following the PRISMA guidelines, resulting in 41 studies eligible for analysis. This review identifies key areas for future research, potential limitations, and offers insights into the evolving trends in BCI-AR systems, contributing to the development of more robust and user-friendly applications.

Advanced Sparse Filtering-Based Domain Adaptation for Fault Diagnosis in Variable Working Conditions

Traditional domain adaptation (DA) methods often encounter challenges with cross-domain feature extraction and the precise assessment of domain differences. To overcome these limitations, we introduce the Enhanced Sparse Filtering-Based Domain Adaptation (ESFBDA) method. This method distinguishes itself by enhancing sparse filtering (SF) with the integration of row-column normalization and a cosine penalty, specifically designed to minimize feature loss—a critical issue in existing DA techniques. Additionally, we employ Bootstrap resampling to refine domain distribution alignment, a novel step that boosts feature similarity and effectiveness in DA. This integrated approach ensures more accurate feature extraction, which is crucial for the classifier's fault detection capability. In our study, through two distinct experiments on WT-planetary gearbox fault diagnosis and bearing fault diagnosis, the ESFBDA method demonstrated remarkable accuracy, significantly surpassing traditional approaches and showcasing its robust applicability across varied diagnostic scenarios.

Infrared thermal imaging for assessing human perspiration and evaluating antiperspirant product efficacy

In humans, perspiration regulates core body temperature. Therefore, objectively evaluating it is essential for studying sweat gland function and mechanisms, particularly in antiperspirant efficacy studies. Various approaches have been developed for measuring human perspiration and evaluating antiperspirant efficacy, but are unsuitable for robust and routine clinical testing applications. This paper shows how infrared thermography, utilizing both high- and low-resolution modes, functions as a multiscale imaging modality. The high-resolution mode extracts physiological parameters (respiratory ~ 0.3 Hz and heart rate ~ 1.0 Hz) and visualizes the reduction of the sweat pore radii (from 359 ± 155 μm to 161 ± 47 μm) after antiperspirant application, consistent with known mechanisms of pore plugging and constriction induced by aluminum salts. The low-resolution mode quantitatively maps sweat retention in underarm clothing. All study participants in a clinical trial showed reduced sweat retention on their T-shirts due to antiperspirants, with reductions ranging from approximately 37–97% and an average reduction of 77.7 ± 22.1% using the developed methodology and tested antiperspirant. Overall, this non-invasive technique presents significant potential for clinical and personal care product evaluations, particularly in the early stages of product development.

Vulnerability assessment of elasmobranch species to fisheries in coastal Kenya: Implications for conservation and management policies

Ecological Risk Assessment (ERA) of species is useful for making informed management decisions based on relative vulnerability of species to fisheries. Understanding the vulnerability of species to fishing gears is important for targeted management measures especially for species known to have delicate life-history strategies such as the elasmobranchs. Productivity and Susceptibility Analysis (PSA), an ERA approach, was applied to 61 elasmobranch species (32 shark and 29 ray species) to assess their relative vulnerability (low, medium and high) to fishing by artisanal, prawn trawl and tuna longline fisheries in Kenya’s coastal waters. Overall, results showed 28 % of rays and 31 % of sharks were highly vulnerable to the fisheries, with the pelagic longline tuna fishery having the largest proportions of rays (50 %) and sharks (54 %) being highly vulnerable. Most of the species were in the medium vulnerability category. Among the artisanal gears, only gillnets had species that were highly vulnerable to the fishery, while relatively fewer species (n = 9) of both sharks and rays were highly vulnerable to the prawn trawlers compared to those in the medium vulnerability category (n = 25). Seventeen species classified as High Vulnerability Species Assemblage (HVSA), were assessed to be at high risk of overfishing by different gears, and are recommended for management as a complex. Of the HVSA group, two species; Pseudoginglymostoma brevicaudatum and Rhina ancylostoma are classified by the IUCN Red List of Threatened Species as Critically Endangered (CR), while another four; Rhinoptera jayakari, Mobula birostris, Stegostoma tigrinum and Carcharhinus plumbeus are Endangered (EN), with the rest being either Vulnerable (Gymnura poecilura, Pastinachus ater, Pateobatis jenkinsii, Taeniurops meyeni, Triaenodon obsesus, Carcharhinus falciformis, and Carcharhinus carcharias), Near Threatened (Maculabatis ambigua, Mustelus mosis, Prionace glauca) or classified as Data Deficient (Pliotrema warreni). Although majority of the species have been placed in the medium vulnerability category, some of them are those that are threatened with extinction globally, calling for a more precautionary management approach. A validation approach has been used to mitigate the problems of PSAs associated with data quality and uncertainty around estimated vulnerability indices. More robust application of PSA will require continuous updating of life-history attributes to improve on sensitivity. A precautionary Shark and Ray Management Plan (SRMP) that takes into account the outputs of the PSA is recommended for management and conservation of the elasmobranch stocks in Kenya’s marine waters.

CellSAM: Advancing Pathologic Image Cell Segmentation via Asymmetric Large-Scale Vision Model Feature Distillation Aggregation Network.

Segment anything model (SAM) has attracted extensive interest as a potent large-scale image segmentation model, with prior efforts adapting it for use in medical imaging. However, the precise segmentation of cell nucleus instances remains a formidable challenge in computational pathology, given substantial morphological variations and the dense clustering of nuclei with unclear boundaries. This study presents an innovative cell segmentation algorithm named CellSAM. CellSAM has the potential to improve the effectiveness and precision of disease identification and therapy planning. As a variant of SAM, CellSAM integrates dual-image encoders and employs techniques such as knowledge distillation and mask fusion. This innovative model exhibits promising capabilities in capturing intricate cell structures and ensuring adaptability in resource-constrained scenarios. The experimental results indicate that this structure effectively enhances the quality and precision of cell segmentation. Remarkably, CellSAM demonstrates outstanding results even with minimal training data. In the evaluation of particular cell segmentation tasks, extensive comparative analyzes show that CellSAM outperforms both general fundamental models and state-of-the-art (SOTA) task-specific models. Comprehensive evaluation metrics yield scores of 0.884, 0.876, and 0.768 for mean accuracy, recall, and precision respectively. Extensive experiments show that CellSAM excels in capturing subtle details and complex structures and is capable of segmenting cells in images accurately. Additionally, CellSAM demonstrates excellent performance on clinical data, indicating its potential for robust applications in treatment planning and disease diagnosis, thereby further improving the efficiency of computer-aided medicine.