Segmentation Of Data Research Articles

Peanut oil price change forecasts through the neural network

Purpose For a wide range of market actors, including policymakers, forecasting changes in commodity prices is crucial. As one of essential edible oil, peanut oil’s price swings are certainly important to predict. In this paper, the weekly wholesale price index for the period of January 1, 2010 to January 10, 2020 is used to address this specific forecasting challenge for the Chinese market. Design/methodology/approach The nonlinear auto-regressive neural network (NAR-NN) model is the forecasting method used. Forecasting performance based on various settings, such as training techniques, delay counts, hidden neuron counts and data segmentation ratios, are assessed to build the final specification. Findings With training, validation and testing root mean square errors of 5.89, 4.96 and 5.57, respectively, the final model produces reliable and accurate forecasts. Here, this paper demonstrates the applicability of the NAR-NN approach for commodity price predictions. Originality/value On the one hand, the findings may be used as independent technical price movement predictions. Conversely, they may be included in forecast combinations with forecasts derived from other models to form viewpoints of commodity price patterns for policy research.

Farside helioseismology with Sun-as-a-star data: the solar cycle as seen with 7-day-long BiSON timeseries

Abstract We present results from fitting p-mode spectra derived from 7-d segments of Sun-as-a-star helioseismic observations from the Birmingham Solar Oscillations Network covering 32 yr. The results show a clear dependence of the mode frequencies on solar activity, and the frequency dependence of the sensitivity to activity can also be seen. Because we use data segments that cover less than half of a solar rotation, we are able to test for the effect of activity on the solar far side. By fitting with a model that takes into account activity on the far side of the Sun, we show that the frequency shifts are sensitive to activity from the whole Sun, not just the side facing the observer. Our results suggest that there is potential to investigate activity-related asteroseismic frequency shifts in solar-like oscillators using short time series of observations.

A Novel Mesoscale Eddy Identification Method Using Enhanced Interpolation and A Posteriori Guidance

Mesoscale eddies are pivotal oceanographic phenomena affecting marine environments. Accurate and stable identification of these eddies is essential for advancing research on their dynamics and effects. Current methods primarily focus on identifying Cyclonic and Anticyclonic eddies (CE, AE), with anomalous eddy identification often requiring secondary analyses of sea surface height anomalies and eddy center properties, leading to segmented data interpretations. This study introduces a deep learning model integrating multi-source fusion data with a Squeeze-and-Excitation (SE) attention mechanism to enhance the identification accuracy for both normal and anomalous eddies. Comparative ablation experiments validate the model’s effectiveness, demonstrating its potential for more nuanced, multi-source, and multi-class mesoscale eddy identification. This approach offers a promising framework for advancing mesoscale eddy identification through deep learning.

Physics-Based Synthetic Data Model for Automated Segmentation in Catalysis Microscopy.

In catalysis research, the amount of microscopy data acquired when imaging dynamic processes is often too much for nonautomated quantitative analysis. Developing machine learned segmentation models is challenged by the requirement of high-quality annotated training data. We thus substitute expert-annotated data with a physics-based sequential synthetic data model. We study environmental scanning electron microscopy (ESEM) data collected from isopropanol oxidation to acetone over cobalt oxide as an example. Upon applying a temperature program during the reaction a phase transition occurs, reducing the catalyst selectivity toward acetone. This is accompanied on the micrometer ESEM scale by the formation of cracks between the pores of the catalyst surface. We aim to generate synthetic data to train a neural network capable of semantic segmentation (pixel-wise labeling) of this ESEM data. This analysis will lead to insights into this phase transition. To generate synthetic data that approximates this transition, our algorithm composes the ESEM images of the room-temperature catalyst with dynamically evolving synthetic cracks satisfying physical construction principles, gathered from qualitative knowledge accessible in the ESEM data. We mimic the surface crack growth propagation along surface paths, avoiding close vicinity to nearby pores. This physics-based approach results in a lowered rate of false positives compared to a random approach.

BOston Neonatal Brain Injury Data for Hypoxic Ischemic Encephalopathy (BONBID-HIE): I. MRI and Lesion Labeling

Hypoxic ischemic encephalopathy (HIE) is a brain injury that occurs in 1 ~ 5/1000 term neonates. Accurate identification and segmentation of HIE-related lesions in neonatal brain magnetic resonance images (MRIs) is the first step toward identifying high-risk patients, understanding neurological symptoms, evaluating treatment effects, and predicting outcomes. We release the first public dataset containing neonatal brain diffusion MRI and expert annotation of lesions from 133 patients diagnosed with HIE. HIE-related lesions in brain MRI are often diffuse (i.e., multi-focal), and small (over half the patients in our data having lesions occupying <1% of the brain volume (including ventricles)). Segmentation for HIE MRI data is remarkably different from, and arguably more challenging than, other segmentation tasks such as brain tumors with focal and relatively large lesions. We hope that this dataset can help fuel the development of MRI lesion segmentation methods for HIE and small diffuse lesions in general.

Data-Driven Machine Learning Strategy for Designing Metal-Ion-Doped γ-Bi2MoO6 Photocatalysts to Enhance Degradation Performance.

Doped semiconductors are often used to improve photocatalytic efficiency and address the challenges of easy recombination of electron-hole pairs and poor photoluminescence. However, the reproducibility and complexity of experimental studies result in time-consuming and less cost-effective studies, and it is difficult to gain insights into the intrinsic properties of doped photocatalysts to control their performance. Introducing a machine learning approach, we constructed a photocatalytic model of transition-metal- and rare earth metal-ion-doped γ-Bi2MoO6. We selected 18 factors of preparation conditions and dopant ion properties, and constructed 806 data sets through literature collection for correlation analysis, paving the way for a more efficient and cost-effective research process. The results of our study are promising. The trained and improved XGboost model demonstrated high resistance to the variability caused by data segmentation, with a cross-validated model showing a coefficient of determination of 0.942. Through the combination of characteristic importance and Shapley additive explanation analysis, the importance and correlation trends of preparation conditions and dopant ion properties are obtained, especially the positive correlation trend of excitation time and preparation time and the negative correlation trend of atomic mass and bandwidth. Model prediction and experimental validation are used to demonstrate the effectiveness and behavioral prediction ability, and the Zn and Cd elements are successfully predicted for doping modification means. This study contributes to the modification and preparation of γ-Bi2MoO6 materials and provides a solid foundation for the efficient design of photocatalysts.

Quantitative Structural and Compositional Elucidation of Real-World Pharmaceutical Tablet Using Large Field-of-View, Correlative Microscopy-Tomography Techniques and AI-Enabled Image Analysis.

The purpose of this study is to present a correlative microscopy-tomography approach in conjunction with machine learning-based image segmentation techniques, with the goal of enabling quantitative structural and compositional elucidation of real-world pharmaceutical tablets. Specifically, the approach involves three sequential steps: 1) user-oriented tablet constituent identification and characterization using correlative mosaic field-of-view SEM and energy dispersive X-ray spectroscopy techniques, 2) phase contrast synchrotron X-ray micro-computed tomography (SyncCT) characterization of a large, representative volume of the tablet, and 3) constituent segmentation and quantification of the imaging data through user-guided, iterative supervised machine learning and deep learning. This approach was implemented on a real-world tablet containing 15% API and multiple common excipients. A representative volumetric tablet image was obtained using SyncCT at a 0.36-µm resolution, from which constituent particles and pores were fully segmented and quantified. As validation, the derived tablet formulation composition and porosity agreed with the experimental values, despite the micrometer-scale particle and pore sizes. The approach also revealed the formation of ordered mixture inside the tablet. Notably, the image-derived size distributions of both the agglomerated microcrystalline cellulose and its primary particulate units matched the laser diffraction-based measurements of the as-is material. Key pore attributes including the pore size distribution, spatial anisotropy, and pore interconnectivity were also qualified. Overall, this study demonstrated that the correlative microscopy-tomography approach, by leveraging phase contrast SyncCT and AI-based image analysis, can deliver new, practically-useful structural and compositional information and facilitate more efficient formulation and process development of tablets.

Improved Edge Detection using Morphological Operation to Segmentation of Fingernail Images

Accurate segmentation of fingernail images is essential for biomedical applications like dermatological diagnostics and nail disease assessments. This study compares traditional methods (Sobel and Canny edge detectors) with an improved method using adaptive thresholding and morphological closing for fingernail image segmentation. The methodology includes data collection, preprocessing, edge detection, segmentation, and evaluation. A dataset of 500 fingernail images (free of nail polish) was captured using a digital camera. Preprocessing involves grayscale conversion to simplify analysis and Gaussian smoothing to reduce noise while preserving key features. For segmentation, thresholding and K-means clustering isolate the fingernail from the background. Evaluation combines qualitative and quantitative analyses. Qualitative results demonstrate that the improved method consistently outperforms traditional techniques under diverse conditions. Quantitative evaluation, based on accuracy, recall, F1 score, and Intersection over Union (IoU), further supports these findings. The Sobel method achieves 0.80 accuracy, 0.77 recall, 0.87 F1 score, and 0.77 IoU. The Canny method achieves 0.82 accuracy, 0.78 recall, 0.88 F1 score, and 0.78 IoU. In contrast, the improved method achieves 0.97 accuracy, 0.98 recall, 0.99 F1 score, and 0.98 IoU. The results clearly show that the improved method, using adaptive thresholding and morphological closing, provides superior segmentation performance. Additionally, the approach remains computationally efficient, making it suitable for real-time applications in medical diagnostics.

Crop Disease Detection by Deep Joint Segmentation and Hybrid Classification Model: A CAD‐Based Agriculture Development System

ABSTRACTPrecise detection of crop disease at the early stage is a crucial task, which will reduce the spreading of disease by taking preventive measures. The main goal of this research is to propose a hybrid classification system for detecting crop disease by utilising Modified Deep Joint (MDJ) segmentation. The detection of crop diseases involves five stages. They are data acquisition, pre‐processing, segmentation, feature extraction and disease detection. In the initial stage, image data of diverse crops is gathered in the data acquisition phase. According to the work, we are considering Apple and corn crops with benchmark datasets. The input image is subjected to pre‐processing by utilising the median filtering process. Subsequently, the pre‐processed image under goes a segmentation process, where Modified Deep Joint segmentation is proposed in this work. From the segmented image, features like shape, colour, texture‐based features and Improved Median Binary Pattern (IMBP)‐based features are extracted. Finally, the extracted features are given to the hybrid classification system for identifying the crop diseases. The hybrid classification model includes Bidirectional Long Short‐Term Memory (Bi‐LSTM) and Deep Belief Network (DBN) classifiers. The outcome of both the classifiers is the score, which is subjected to an improved score level fusion model, which determines the final detection results. Finally, the performance of the proposed hybrid model is evaluated over existing methods for various metrics. At a training data of 90%, the proposed scheme attained an accuracy of 0.965, while conventional methods achieved less accuracy rates.

A new intensity-modulated radiation therapy with deep learning heart rate prediction framework for smart health monitoring

&lt;p&gt;This research paper monitors the patient’s health using sensor data, cloud, and big data Hadoop tools and used to predict heart attack and related results were discussed in detail. The integration of big data, and wearable sensors in pervasive computing has significantly enhanced healthcare services. This proposal focuses on developing an advanced healthcare monitoring system tailored for tracking the activities of elderly individuals. The wearable sensors are placed on humans at a right angle, left arm, right arm, and chest to collect the data. The large data are split into smaller segments using the map and reduce process of big data Hadoop tools. The intensity-modulated radiation therapy (IMRT) approach is used for the mapping phase and deep convolutional neural network (DCNN), deep belief network (DBN), and long short-term memory (LSTM) and proposed deep learning heart rate prediction (DLHRP) algorithms are used for the combiner/reduce phase. The reduction process combines similar segments of data to predict identical classes to predict the severity of human conditions. The proposed IMRT-DLHRP system has improved performance of 96.34% accuracy compared with 84.25%, 89.47%, and 91.58% compared to DCNN, DBN, and LSTM respectively, therefore proposed framework has significant improvement over existing approaches.&lt;/p&gt;

Performance Evaluation of Deep Learning Models for Practical 3D Tooth Scan Data Segmentation: Application of MeshSegNet and DilatedToothSegNet Models

Performance Evaluation of Deep Learning Models for Practical 3D Tooth Scan Data Segmentation: Application of MeshSegNet and DilatedToothSegNet Models

Очистка полуструктурированных и неструктурированных данных дистанционного зондирования Земли

Abstract. The problem of processing semi-structured and unstructured Earth remote sensing (ERS) data obtained by various methods, including satellites and unmanned aerial vehicles, is touched upon. The purpose of the article is to study and implement algorithms for effective purification and preprocessing of semi-structured and unstructured Earth remote sensing data. The importance of cleaning these data from noise, artifacts and errors is emphasized in order to increase their accuracy and significance in applied scientific research and practical application. Key data preprocessing techniques are considered, including noise removal, distortion correction, classification, segmentation and standardization of data, reinforcing theoretical positions with practical examples in Python using libraries such as GDAL, OpenCV and scikit-image. Examples of detection of aerospace and transport objects using machine learning and deep learning are given, emphasizing the importance of accuracy, completeness and F1-Score metrics in assessing the quality of data purification. The practical significance of the study lies in evaluating the effectiveness of data purification methods used to restore images during remote sensing of the Earth.

Semantic segmentation of urban airborne LiDAR data of varying landcover diversity using XGBoost

AbstractSemantic segmentation of aerial LiDAR dataset is a crucial step for accurate identification of urban objects for various applications pertaining to sustainable urban development. However, this task becomes more complex in urban areas characterised by the coexistence of modern developments and natural vegetation. The unstructured nature of point cloud data, along with data sparsity, irregular point distribution, and varying sizes of urban objects, presents challenges in point cloud classification. To address these challenges, development of robust algorithmic approach encompassing efficient feature sets and classification model are essential. This study incorporates point‐wise features to capture the local spatial context of points in datasets. Furthermore, an ensemble machine learning model based on extreme boosting is utilised, which integrates sequential training for weak learners, to enhance the model’s resilience. To thoroughly investigate the efficacy of the proposed approach, this study utilises three distinct datasets from diverse geographical locations, each presenting unique challenges related to class distribution, 3D terrain intricacies, and geographical variations. The Land‐cover Diversity Index is introduced to quantify the complexity of landcover in 3D by measuring the degree of class heterogeneity and the frequency of class variation in the dataset. The proposed approach achieved an accuracy of 90% on the regionally complex, higher landcover diversity dataset, Trivandrum Aerial LiDAR Dataset. Furthermore, the results of the study demonstrate improved overall predictive accuracy of 91% and 87% on data segments from two benchmark datasets, DALES and Vaihingen 3D.

RapidBenthos: Automated segmentation and multi‐view classification of coral reef communities from photogrammetric reconstruction

Abstract Underwater photogrammetry is routinely used to monitor large areas of complex and heterogeneous ecosystems, such as coral reefs. However, deriving data on benthic components (i.e. sand, rubble, coral and algae) from photogrammetry products has remained challenging due to the highly time‐consuming process of manual data extraction. We developed a machine learning approach to quantify benthic community composition in coral reefs from orthomosaics, which requires no manual delineation of benthic components for training or implementation. The current study presents RapidBenthos, an automated workflow that segments and classifies large‐area images. Our pipeline (1) uses a pre‐trained segmentation model, eliminating the need for manually generated fine‐scale segmented training data, and (2) classifies the resulting segments from multiple views using the underlying survey images, allowing for classification to fine taxonomic levels. Within a test photomosaic built from a coral reef area of 40 m−2, the model automatically detected 43 different benthic classes. Validation resulted in an overall classification accuracy of 0.96 and a segmentation accuracy of 0.87, when compared to a manually digitised replica. The RapidBenthos workflow was 195 times faster than manual segmentation and classification. Additional validation of 524 Acropora coral colonies from 11 additional test plots resulted in a segmentation accuracy of 0.92 and classification accuracy of 0.88 to the coarser ‘Acropora’ group. RapidBenthos has the capability to extract an unprecedented level of data from photomosaics of coral reefs or other complex environments, allowing to sustainably scale photogrammetric monitoring technique both in replicate and survey extent, which consequently can lead to new research questions and more informed ecosystem management.

Measuring arterial tortuosity in the cerebrovascular system using Time-of-Flight MRI.

The Circle of Willis (CW) is a critical cerebrovascular structure that supports collateral blood flow to maintain brain perfusion and compensate for eventual occlusions. Increased tortuosity of high-risk vessels within the CW has been implicated as a marker in the progression of cerebrovascular diseases especially in structures like the internal carotid artery (ICA). This is partly due to age-related plaque deposition or arterial stiffening. Producing reliable tortuosity measurements for vessels segmented from magnetic resonance (MR) time-of-flight (TOF) images requires precise curvature estimation, but existent methods struggle with noisy or sparse segmentation data. We introduce an open-source, end-to-end pipeline that uses unit-speed spline fitting for accurate curvature estimation, generating robust curvature-based tortuosity metrics for the ICA combined with an indicator of spline fit quality. We test this with theoretical data and apply this method to TOF data from 22 participants. We report that our metrics are able to capture tortuosity even under heightened noise constraints and discriminate different types of abnormal arterial coiling. We found that our ICA tortuosity measures correlate positively with age and ultrasound measured carotid artery intima media thickness. This ultimately has important translational implications for being able to reliably generate TOF tortuosity measures and estimate cerebrovascular disease burden. We provide open-source code in a GitHub repository.

End-to-end simulation of nanopore sequencing signals with feed-forward transformers.

Nanopore sequencing represents a significant advancement in genomics, enabling direct long-read DNA sequencing at the single-molecule level. Accurate simulation of nanopore sequencing signals from nucleotide sequences is crucial for method development and for complementing experimental data. Most existing approaches rely on predefined statistical models, which may not adequately capture the properties of experimental signal data. Furthermore, these simulators were developed for earlier versions of nanopore chemistry, which limits their applicability and adaptability to the latest flow cell data. To enhance the quality of artificial signals, we introduce seq2squiggle, a novel transformer-based, non-autoregressive model designed to generate nanopore sequencing signals from nucleotide sequences. Unlike existing simulators that rely on static k-mer models, our approach learns sequential contextual information from segmented signal data. We benchmark seq2squiggle against state-of-the-art simulators on real experimental R9.4.1 and R10.4.1 data, evaluating signal similarity, basecalling accuracy, and variant detection rates. Seq2squiggle consistently outperforms existing tools across multiple datasets, demonstrating superior similarity to real data and offering a robust solution for simulating nanopore sequencing signals with the latest flow cell generation. seq2squiggle is freely available on GitHub at: github.com/ZKI-PH-ImageAnalysis/seq2squiggle. Supplementary data are available at Bioinformatics online.

Research on efficient numerical simulation method for integration fracking with production in shale oil reservoir with multi-source data

AbstractHorizontal well hydraulic fracturing technology has significantly enhanced the productivity of shale reservoirs. However, our understanding of the expansion patterns within the complex fracture network and fluid seepage mechanisms under field conditions remains inadequate. Here, this work develops a dynamic geomechanical (DG) model to simulate the complete sequence of operations in hydraulic fracturing. This study utilizes a construction procedure that closely mirrors field practices to establish the DG model. Furthermore, the numerical simulation results of the DG model are calibrated with field data. This work adopts a numerical simulation method that integrates unsteady seepage model for multi-stage fractured horizontal wells with the dilation-recompaction model to develop the DG model. It systematically constructs the geological model of the shale reservoir by utilizing segmented logging data and by segmenting production data. The time series evolution system is developed through an iterative process involving discrete time steps. Results show that the DG model can perform history matching on a multi-stage basis, enabling comprehensive and detailed analysis of the entire reservoir. This process effectively replicates the distribution relationship between each reconstruction zone and the overall productivity. Furthermore, the DG model is capable of accurately simulating the dynamic process of injected high-pressure fluids into the reservoir to fracture the rock and the dynamic evolution law of reservoir properties. Hydraulic fracturing creates a fracture zone that centers on the well’s border and spreads outward radially. The injection volume and failure pressure are significantly correlated with the scale of shale reservoir reconstruction. Following the injection of 790.5 m³ of fracturing fluid in the first stage, the fracture half-length can reach around 148 m, essentially fulfilling the design specifications. Permeability can reach up to 86 mD at this moment, and it can even be maintained at the level of 46 mD during production. In conclusion, the DG model broadens the focus of study on the development of shale reservoirs and lays the groundwork for improving productivity and optimizing hydraulic fracturing design.

Segmentation for mammography classification utilizing deep convolutional neural network

BackgroundMammography for the diagnosis of early breast cancer (BC) relies heavily on the identification of breast masses. However, in the early stages, it might be challenging to ascertain whether a breast mass is benign or malignant. Consequently, many deep learning (DL)-based computer-aided diagnosis (CAD) approaches for BC classification have been developed.MethodsRecently, the transformer model has emerged as a method for overcoming the constraints of convolutional neural networks (CNN). Thus, our primary goal was to determine how well an improved transformer model could distinguish between benign and malignant breast tissues. In this instance, we drew on the Mendeley data repository’s INbreast dataset, which includes benign and malignant breast types. Additionally, the segmentation anything model (SAM) method was used to generate the optimized cutoff for region of interest (ROI) extraction from all mammograms. We implemented a successful architecture modification at the bottom layer of a pyramid transformer (PTr) to identify BC from mammography images.ResultsThe proposed PTr model using a transfer learning (TL) approach with a segmentation technique achieved the best accuracy of 99.96% for binary classifications with an area under the curve (AUC) score of 99.98%, respectively. We also compared the performance of the proposed model with other transformer model vision transformers (ViT) and DL models, MobileNetV3 and EfficientNetB7, respectively.ConclusionsIn this study, a modified transformer model is proposed for BC prediction and mammography image classification using segmentation approaches. Data segmentation techniques accurately identify the regions affected by BC. Finally, the proposed transformer model accurately classified benign and malignant breast tissues, which is vital for radiologists to guide future treatment.

Sustaining Rice Farming: Challenges and Solutions for Agribusiness in Barangay Tagum Norte, Trinidad, Bohol

In the Philippines, rice is a vital crop that serves as the primary food source for 80% of the population in Barangay Tagum Norte, Trinidad, Bohol. Despite the potential of rice agribusiness, farmers in this area need help maintaining consistent production due to limited resources, economic uncertainties, and market fluctuations. This research identifies and addresses these challenges to promote sustainability in rice agribusiness. The study employs a descriptive-exploratory qualitative research design, focusing on a case study approach. Ten rice farmers actively involved in agribusiness in Barangay Tagum Norte were interviewed using structured researcher-made guides. Thematic Analysis was utilized to analyze the gathered information, assigning codes to relevant data segments and organizing them into themes or categories. The findings revealed that financial constraints force farmers to sell their rice at lower prices, and borrowing from lending institutions results in high interest rates, leading to solo farming practices and potential efficiency issues. Manpower challenges, machinery accessibility issues, and significant expenses for seeds, fertilizers, and pesticides also impact productivity. Price fluctuations during harvest create economic challenges, threatening the sustainability of agribusiness. Limited market access and information exacerbate these issues. Farmers need collaborative government-private sector initiatives to address their irrigation, machinery, seed, and fertilizer needs. The study concludes that rice farmers face interconnected challenges throughout the production cycle, including financial constraints, labor shortages, machinery access, seed and fertilizer expenses, government-supplied shortages, and market volatility. The research highlights adaptive strategies such as the “Rollover method,” where farmers strategically time last season’s harvest sale to avoid price declines, showcasing a proactive approach to market fluctuations. Keywords: Agribusiness, Challenges to Sustainability, Qualitative Research, Thematic Analysis

Detection of plant root diseases using computational intelligence and advanced image processing techniques

This study presents an advanced Siamese-based Squeeze Googlenet and LeNet-5 (SSGL) model for the early detection of root diseases in plants, which is crucial for agricultural productivity and crop management. The methodology integrates state-of-the-art image processing techniques and Deep Learning (DL) algorithms across several key stages. In the pre-processing phase, techniques like median filtering, histogram equalization, i-CLAHE, and Z-score normalization are applied to enhance image quality and prepare the data for segmentation. The segmentation process utilizes a modified DeepLab V3 model, renowned for its precision in isolating root structures from complex backgrounds, ensuring accurate subsequent analysis. Feature extraction focuses on deriving comprehensive characteristics from segmented root images, including texture features using GLCM, color histograms, geometric metrics like area and perimeter, and shape descriptors such as circularity and rectangularity. A hybrid optimization approach employs the Hybrid Pufferfish-based Coati Optimization (HPCO) algorithm for feature selection, effectively identifying the most relevant features crucial for disease detection. For detection, the SSGL model integrates optimized architectures from SqueezeNet, GoogleNet, and LeNet-5 within a Siamese network framework. This setup enables efficient classification of root images into healthy and diseased categories based on cosine similarity. The proposed SSGL model outperforms other models with the highest accuracy (0.9856), precision (0.9760), sensitivity (0.9764), and F-measure (0.9771), while also achieving the lowest FPR (0.0149) and FNR (0.0109).