3D printing is evolving at a fast pace in both the manufacturing and construction sectors. These advancements can greatly benefit these industries. However, the 3D printing of concrete structures presents some challenges due to defects in the 3D concrete printed elements. Hence, this study systematically reviews Artificial Intelligence (AI)-driven techniques, such as Computer Vision and Machine Learning, to identify surface defects that can occur in 3D-printed cementitious material structures. The adopted methodology was the PRISMA statement with the aim of reporting the systematic review and meta-analysis. Two well-known databases, Web of Science and Scopus, were utilised for data extraction of articles published during the past 10 years, between 2014 and May 2025. The initial search provided 110 articles, both conference and journal papers; after screening, only 11 were left for the final review assessment. The smaller number of the final articles shows that much work is still needed in this area. It has been observed that various computer vision and machine learning-based methodologies were employed to classify defects in 3D concrete printed structures. Deep learning algorithms, such as YOLO and RT-DETR, were featured as the most efficient in real-time defect detection and quality monitoring. It was also observed that real-time monitoring systems attached to 3D printers help in reducing the material wastage, which is essential to meet the sustainable goals. However, more work is still required to underline the defects of 3D-printed cementitious material, probably with the involvement of AI image processing tools and techniques. This can help to automate the defects in 3D-printed structures, and by this, the productivity could be enhanced.

Machine vision and artificial intelligence have reshaped quality inspection across modern manufacturing. In this project, we designed a non-contact metrology system that uses machine vision to inspect mechanical parts like bolts, nuts, and pinions. The setup relies on a high-resolution camera along with a controlled backlight to grab crisp, clear images of components. We run these images through image processing tools and machine learning algorithms to pull out accurate measurements and spot any defects. This system not only boosts accuracy and consistency, but it also runs in real time and slashes the need for manual labor. Our tests show it’s faster and more precise than older, manual methods. It's the kind of upgrade fit for Industry 4.0 and advanced industrial automation.

Abstract - Artificial Intelligence (AI) has emerged as a powerful tool in image processing by enabling automated analysis, feature extraction and intelligent decision-making from visual data. With advancements in computational power and data availability, AI-based image processing techniques are increasingly applied across multiple domains such as healthcare, agriculture, astronomy, finance, security and industrial automation. This paper presents a comprehensive survey of recent research works that integrate AI methods with image processing to address domain-specific challenges and improve system performance. Key techniques including Convolutional Neural Networks (CNN), Generative Adversarial Networks (GAN), Support Vector Machines (SVM), Optical Character Recognition (OCR) and hybrid models are systematically reviewed and compared. Their applications in tasks such as medical image segmentation, weed detection, satellite image enhancement, astronomical image preprocessing and document automation are discussed. The survey highlights significant improvements in accuracy, efficiency and automation achieved through AI-driven approaches when compared to traditional image processing techniques. In addition, major challenges such as limited annotated datasets, high computational requirements, generalization issues and ethical concerns related to privacy are identified. Finally, the paper outlines future research directions including explainable AI, federated learning and lightweight models for real-time and edge-based deployment. This study aims to serve as a useful reference for researchers and practitioners seeking to apply AI techniques in image processing across diverse real-world applications. Keywords: artificial intelligence, image processing, convolutional neural networks, generative adversarial networks, computer vision, automation

Low-resolution medical images, especially Chest X-Rays (CXRs), often suffer from noise and blurriness, hindering accurate diagnosis. This study introduces the Noise-Aware Convolutional Neural Network (NA-CNN) architecture to address this issue. The objective is to enhance image quality by eliminating noise and converting low-resolution images into high-resolution images. The methodology involves a Convolutional Neural Network (CNN)-based model integrated with sparse coding reconstruction, adaptive downsampling, and nonlinear mapping. Evaluations were conducted on a high-performance system using the COVID-19 Radiography dataset. The results demonstrated that NA-CNN consistently outperformed the existing CNN model, achieving higher Peak Signal-to-Noise-Ratio (PSNR) and Structural Similarity Index Measure (SSIM) values across various noise levels, indicating superior image quality and structural fidelity. The novelty of this work lies in its innovative architecture that combines a CNN with adaptive techniques, resulting in efficient and high-quality image enhancement. NA-CNN's robustness and efficiency make it a valuable tool for medical image processing, providing significant advancements over existing methods.

ERBB2 is amplified or overexpressed in 15% to 20% of primary invasive breast cancers. Despite the aggressive nature of ERBB2-positive breast cancers, the development of anti-ERBB2 therapy has resulted in better prognoses. HScore is a scoring system that broadens immunohistochemistry (IHC) results into a quantitative range. The aim of this retrospective study was to evaluate whether the HScore can be used to predict all ERBB2 IHC-equivocal cases as positive or negative without the need for fluorescence in situ hybridization confirmation. Image acquisition and processing tools were used on previously collected tissue slides from patients with ERBB2 IHC 2+ breast cancer from 2014 to 2023. After image acquisition, HScore values were calculated using variables from image analysis software. The mean HScore values in ERBB2-positive cases were higher (3.06) than in ERBB2-negative cases (2.62). Receiver operating characteristic curve analysis yielded an area under the curve of 0.668 (95% CI, 0.523-0.812), indicating poor predictive accuracy. The optimal HScore cutoff of 2.6007 provided moderate sensitivity (0.77) but limited specificity (0.58). Although the HScore showed moderate predictive accuracy, it lacked the reliability to replace fluorescence in situ hybridization as the gold standard. Future research should focus on refining the methodology and integrating it with other diagnostic approaches to enhance accuracy and clinical utility.

The rapid development of the digital economy has fundamentally transformed the professional role of economists, expanding it beyond traditional analytical and forecasting functions. Modern economists increasingly operate in datadriven environments where economic information must be processed, interpreted, and communicated visually for decisionmakers, clients, and market stakeholders. The growth of e-commerce, digital marketing, and platform-based business models has intensified the demand for specialists capable of integrating economic analysis with visual communication tools. However, higher education programmes in economics often remain focused on quantitative calculations and theoretical modelling, insufficiently addressing the development of visual and applied digital competencies. The purpose of this article is to substantiate the role of computer graphics and design in the formation of professional competencies of bachelor students in economics and to determine their contribution to analytical thinking, visual communication, and applied digital skills within the framework of the digital economy. The research is based on the analysis and synthesis of contemporary scientific literature on digitalisation of economic education, a systemic approach to defining the place of computer graphics and design within economics-oriented educational programmes, and functional analysis of graphic tools used for economic data visualisation. Additionally, the study generalises teaching experience obtained through a laboratory workshop in the discipline “Computer Graphics and Design”, which focuses on practical tasks related to economic analysis, e-commerce, and marketing communication. The findings demonstrate that computer graphics and design perform analytical, communicative, and applied functions in the professional training of future economists. Mastering tools for image processing, infographic development, mock-up creation, and digital advertising enables students to visualise economic data, interpret complex indicators, and present analytical results in a structured and comprehensible form. The integration of graphic tools into economics education enhances students’ ability to analyse market trends, support managerial decision-making, and design visual content for business communication in digital environments. Practical assignments based on real economic contexts contribute to the development of interdisciplinary skills combining economics, design thinking, and digital literacy. Computer graphics and design should be regarded as an integral component of bachelor-level economics education in the digital economy. Their systematic integration into educational programmes strengthens professional mobility, improves graduates’ competitiveness in the labour market, and increases the practical relevance of economic training. The article proposes incorporating design-oriented laboratory workshops into economics curricula as an effective means of developing visual communication competencies. Further research should focus on assessing learning outcomes related to visual-economic skills and expanding interdisciplinary educational models in economics.

Computer-assisted surgery and planning in percutaneous pelvic screw fixation.

Mechanical power (MP) integrates the contributors to ventilator-induced lung injury and relates to mortality, yet bedside calculation remains challenging. The dynamic mechanical power equation (MPdyn) offers a simple approach but requires validation against the geometric gold standard (MPgeo). To validate MPdyn against MPgeo in volume-controlled (VCV) and pressure-controlled ventilation (PCV) across I:E ratios of 1:2 and 1:1. Prospective observational study in a tertiary ICU. Adults with ARDS (n=36) underwent standardized measurements in VCV and PCV at I:E 1:2 and 1:1. For each setting, three complete pressure-volume (P-V) loop screenshots were captured. MPgeo (J/min) was derived using a Python-based image-processing tool (OpenCV/NumPy/PIL) employing grayscale conversion, thresholding, morphological closing, flood-fill segmentation, and area-ratio computation. MPdyn was computed as MV × [WOBv+(PEEP × 0.098)]. Agreement was assessed with univariable linear regression and Bland-Altman analyses. MPdyn correlated strongly with MPgeo in all modes and I:E ratios (R2≥0.97). Biases (MPdyn - MPgeo) were 0.15J/min (0.8%) for VCV 1:2, 0.32J/min (2.1%) for VCV 1:1, 0.28J/min (1.9%) for PCV 1:2, and 0.52J/min (3.4%) for PCV 1:1. The standard deviation of the bias remained <0.5J/min across analyses. MPdyn demonstrates high agreement with MPgeo in both VCV and PCV modes, supporting its use as a simple, reliable, and bedside-applicable tool for calculating mechanical power in ARDS patients.

Cell detection is ubiquitous in the analysis of microfluidic cell assays. In cell biology, immunology, oncology, and toxicology research, studying cellular response starts with identifying the cells on chip. The large amount of data generated in such assays requires automating image analysis. While multitudes of image processing tools exist, the microfluidic channel network and crowded cell environment make it difficult to identify and track cells by conventional image processing techniques. In contrast, machine learning-based techniques may overcome this challenge. Two important challenges in implementing these techniques are that it often requires tedious image labeling and coding expertise. Here, we present a facile method for cell detection in microfluidic arrays using Faster region-based convolutional neural network (R-CNN) that addresses both challenges. First, image labeling is fast and easy, because Faster R-CNN only needs bounding boxes as labels to generate training data. Second, we provide a ready-to-use model and a guide for training a Faster R-CNN model that does not require coding expertise. We demonstrate that Faster R-CNN does not need trade-offs between precision and user-friendliness: we created a model that detects cells with an average precision over 98% using a few hundred annotations, which takes less than half an hour. We show that shapes created by the microfluidic structure alone or its interplay with cells are not misidentified as cells. We show for the first time cell detection using Faster R-CNN in microfluidic chips; we envision that this approach will have a broad use in many on-chip fundamental biology and drug-discovery assays.

BackgroundComputed tomography (CT) and magnetic resonance imaging (MRI) are essential in clinical diagnosis and treatment planning, but their images are often compromised by limited contrast and insufficient detail, reducing diagnostic clarity. Traditional enhancement methods—such as histogram equalization (HE) can improve visibility but may introduce noise, over-enhancement, or structural distortion. Quantum-inspired computational techniques have recently emerged as promising tools for nonlinear and adaptive image processing. Building on the quantum signal processing (QSP) framework, this study proposes a quantum-inspired enhancement (QIE) algorithm designed to improve medical image contrast while preserving structural details.MethodsWe propose a QIE algorithm that embeds a three-pixel quantum-correlation system within a QSP framework. After normalizing grayscale values, each 3×3 neighborhood is mapped to superposition states; edge-sensitive basis states are selectively accumulated in four orientations to produce the enhanced output. The algorithm was evaluated using T2-weighted magnetic resonance (MR) brain images and CT lung images obtained from 10 different patients. Its performance was compared with four representative classical enhancement methods: HE, contrast-limited adaptive HE (CLAHE), fuzzy HE (FHE), and wavelet-based enhancement (WBE), employing quantitative metrics such as entropy, peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), and contrast-to-noise ratio (CNR). Paired two-sided t-tests (α=0.05) were used.ResultsQIE reached the highest mean entropy on both datasets (CT: 4.37±0.31; MR: 6.45±0.16) vs. HE 4.00±0.25 (P=2.8×10−4) and 5.67±0.16 (P=2.3×10−7) respectively, indicating superior information retention and detail enhancement. Its PSNR and SSIM were significantly better than HE, FHE, and WBE (all P<0.01), reflecting better signal fidelity and structural preservation; vs. CLAHE, QIE PSNR was −3.4 dB lower on CT and −3.3 dB lower on MR (both P<0.001), but SSIM differed by <0.001 (P≥0.13). CNR with QIE (CT: 4.00±3.54; MR: 3.66±2.81) was not statistically different from any method (P≥0.05).ConclusionsThe proposed QIE algorithm demonstrates superior performance in enhancing the contrast and preserving the structural details of medical images. By leveraging quantum-inspired mechanisms, the algorithm shows potential for improving diagnostic accuracy and supporting clinical treatment planning. Future work will explore the application of this algorithm to other imaging modalities, investigate its effectiveness as a preprocessing step for commercial artificial intelligence (AI) models, and study the integration with actual quantum computing platforms.

High throughput DNA fragmentation technology for next generation sequencing have become widely available, but there remains a need for affordable and efficient DNA fragmentation pattern analysis. Commercial electrophoresis platforms, such as the TapeStation, are costly, time-consuming, and have limited batch-processing capabilities. Traditional gel electrophoresis provides a low-cost, high-throughput alternative. However, existing open-source software, such as ImageJ, for gel electrophoresis image analysis typically requires extensive manual pre-processing and yields limited quantitative metrics relevant to DNA fragmentation quality control. Here, we have developed an open-source MATLAB-based software, GelInsight, for bulk analysis of gel electrophoresis images for analysis and quality control of DNA fragmentation patterns. GelInsight integrates automated image and signal processing tools to determine the base pair size distribution of each sample and to calculate key quality control metrics, including multiple peak base pair sizes and base pair size percentage within a specified range. A user-friendly graphical user interface facilitates efficient data interaction and comprehensive visualization of the analytical outputs. The quantification accuracy of GelInsight, including peak base-pair accuracy and relative area measurements, is consistent with both existing open-source software (within 2 ± 2 bp) and commercial assays (within 64 ± 24 bp). Overall, this automated tool streamlines gel image analysis and enhances reproducibility and quantitative rigor in assessing DNA fragmentation patterns.

ABSTRACT Fire sources in actual compartment fires exhibit a predominantly uneven spatial distribution. In this context, the impact of the height and location of the fire source, as well as the heat release rate (HRR) of the fire, on the thermal characteristics of Externally Venting Flames (EVF) is experimentally investigated. A systematic parametric analysis is performed, involving 33 separate fire tests, using a 1/4 scale compartment‐façade experimental test rig to evaluate the impact of three fire power levels, three horizontal burner positions and four burner elevations. A comprehensive array of measurement sensors is utilized to accurately characterize the thermal and geometric properties of the developing EVF. In general, experimental results indicate that increasing the burner height results in decreasing temperatures outside the compartment, while increasing the distance between the burner and the opening leads to increased flame projection. Flame intermittency contours are generated using a custom image processing tool to determine the mean EVF height and horizontal projection. The external dimensions of the EVF envelope, as well as its vertical temperature distribution, are compared to the respective values predicted by relevant fire engineering design correlations proposed in the Eurocode design guidelines (EN 1991‐1‐2). In most cases, large discrepancies are observed between the experimental data and the Eurocode predictions, ranging from 22% for the EVF height up to 41% for the EVF projection, while the EVF temperature average deviation is 16%; however, the Eurocode correlations are found to produce, in general, conservative results, with the exception of the EVF temperatures close to the opening's lintel. The obtained results can provide an experimental foundation to assist in further improving and extending currently available fire safety design engineering correlations.

Rice is the staple food of half of the world's population. It provides security for food in many developing nations. The rice crop is usually short, and the deficiency in nutrition is a major problem. The deficiency in nutrients in rice plants is due to soil having low fertility, unbalanced pH, or incorrect application of fertilizers. These factors contribute to nutrition deficiency and affect the crop's growth. The deficiency in nutrients is estimated by observing the crop, leaf's appearance, and leaf's growth pattern. In this work, we aim to analyze the crop with image processing tools in computer vision to accurately estimate and solve the problem at an earlier stage. The ResNet50 model is further customized for accurately diagnosing the deficiency from leaf images of rice plants. The customized model gets an accuracy, F1 score and FPR are of 95.52 %, 95 %, and 2.24 % respectively. It also has better reliability with an MCC of 0.9329 and Kappa of 0.8993 with an inference time of 9 s. The model, thus, provides an early-time efficient and accurate solution to the problem, demonstrating the robust feature learning capabilities of the modified architecture on raw, unaugmented imagedata.

Monitoring of natural resources is a major challenge that remote sensing tools help to facilitate. The Sissili province in Burkina Faso is a territory that includes significant areas dedicated for the preservation of forest resources. The development of satellite image processing tools offers an opportunity for better monitoring of these resources. The aim of this research is to evaluate the performance of classification algorithms in order to determine which one is most suitable for assessing and monitoring land use in the Sissili province. The methodology used is based on the comparative application of three classification algorithms: Maximum Likelihood, Random Forests, and Support Vector Machine. These algorithms were tested on land use units in the Municipality of Sissili, to determine their respective performance. Performance measures were based on the production of the confusion matrix for each algorithm, the calculation of overall accuracy, and the calculation of Kappa coefficient for the three algorithms. The results show that the Random Forest algorithm is the most effective, with a Kappa coefficient of 0.92 and an overall accuracy of 95.14%. This algorithm is followed by SVM with a Kappa coefficient of 0.73 and a maximum overall accuracy of 90.37%. The least effective algorithm for classifying land use units is Maximum Likelihood, with a Kappa coefficient of 0.36 and an overall accuracy of 52.95%. These results clearly demonstrate the superior effectiveness of machine learning algorithms, specifically Random Forest, in classifying land use units in the Sissili province.

BackgroundArterial spin labelling (ASL) is an established magnetic resonance imaging (MRI) technique for non‐invasive assessment of cerebral blood flow (CBF). However, the lack of expertise and the costly computational resources required to analyze ASL data are major barriers to its use in resource‐constrained settings (RCS), particularly in Africa. ASL‐MRICloud was recently introduced as the only cloud‐based open‐source option for ASL processing that requires no installation on local computers, making it suited for ASL analysis in RCS.MethodsIn this work, we implemented ASL‐MRICloud in Google Colab to perform data analysis at scale and minimal cost, with the aim of enhancing population studies reproducibility in RCS. This work was performed as a training exercise of the CONNExIN (COmprehensive Neuroimaging aNalysis Experience In resource‐constraiNed Settings) Program, a neuroimage analysis training program for African researchers. A team of CONNExIN participants leveraged the Open Science Initiative for Perfusion Imaging ASL (OSIPI‐ASL) MRI Challenge dataset (n = 10) to test the implemented ASL‐MRICloud Google Colab. The pipeline included a data preparation step for data conversion and generation of parameter files to be used for data processing. ASL processing and CBF quantification were then performed using automated steps in ASL‐MRICloud to generate whole‐brain CBF maps and extract preset regional values. The pipeline was validated by comparing the CBF maps and regional values to their ground‐truth, ASL‐MRICloud developer analysis, and results from other established ASL processing tools (Oxford ASL and Quantiphyse). The Google Colab implementation was provided to four CONNExIN teams to replicate the challenge data processing and analyze the dataset of 75 subjects from the PREVENT‐AD (PResymptomatic EValuation of Experimental or Novel Treatments for AD) study.ResultsPreliminary results from the simulated OSIPI‐ASL dataset generated from our pipeline are shown in Figure 1. The pipeline, documentation, and results of the analysis from the four teams will be made publicly available on Protocol.io.ConclusionUsing simulated (OSIPI‐ASL) and real‐world (PREVENT‐AD) data, we aim to assess the feasibility of a resource‐efficient image processing tool for reproducible ASL data analysis. Once implemented, we will share our approach for wider use in RCS to enable inclusive and reproducible imaging research.

Monitoring the condition of aquatic ecosystems is becoming increasingly important under growing anthropogenic pressure and climate change, which intensify eutrophication processes in inland waters. The Zakarpattia region, characterized by complex relief, numerous reservoirs, and high biodiversity, is particularly sensitive to hydroecological changes. Traditional methods of water quality control are resource-intensive and lack spatial and temporal efficiency. Therefore, the application of modern remote sensing techniques is highly relevant, as they enable comprehensive large-scale assessments of water bodies. The Google Earth Engine (GEE) platform provides access to long-term satellite datasets and advanced image processing tools, making it an effective instrument for detecting trends in the trophic state of reservoirs. The use of indices such as NDWI, NDVI, TSI, and Turbidity allows for evaluating the spatiotemporal dynamics of eutrophication without field measurements. The study of Zakarpattia reservoirs using these indicators contributes to the development of an integrated environmental monitoring system aimed at the early detection of degradation processes and the prevention of ecological imbalance in the region’s aquatic ecosystems.

The coronavirus spread rapidly in the world and caused the disease of Covid-19. The proposed research was conducted with the aim of classifying people with Covid-19 from other people. lung ultrasound images were used to diagnose covid-19. The open source Point-Care-of-Ultrasound (POCUS) database was collected, which contained 59 Lung Ultrasound (LUS) images and was collected from several other databases. In this research, KNN classifier with K-fold cross validation was used to classify the feature matrix obtained from Fractional B-spline Wavelet Transform (FBSWT). The proposed method, Block-Matching and 3D filtering (BM3D) filter was used in some methods, which had acceptable results. The proposed method was used to classify healthy people from patients with Covid-19. The results show that when the features based on the wavelet transform (WT) are used, the proposed method can be achieved 90.90% sensitivity, 92.30% specificity, and 91.42% accuracy. While the features extracted from FBSWT show that the proposed method can be achieved 95.45% sensitivity, 93.30% specificity, and 94.28% accuracy. Fractional transforms, especially the FBSWT, can be a useful tool for image processing. They can be used for various purposes such as detection and classification. By using FBSWT, it is possible to accurately diagnose the disease of covid-19.

• DiameterJ shows better accuracy and throughput in fibre analysis than SIMPoly. • ANN exhibits excellent predictive performance with R 2 > 0.97 and errors <4 %. • ANN outperforms RSM in generalizability and reliability for diameter prediction. • Molecular weight and concentration are the dominant factors affecting diameter. • Image processing combined with machine learning aids fibre morphology prediction. Electrospun nanofibrous membranes are widely used in biomedical, filtration, and energy applications, where fibre diameter plays a key role in determining membrane morphology and performance. Conventional manual measurement of fibre diameters from scanning electron microscope (SEM) images is inefficient and subject to human error. This study compared two image-processing tools, DiameterJ and SIMPoly, to improve measurement efficiency and accuracy of fibre diameter. DiameterJ was selected for its higher reliability and batch-processing capability. Using DiameterJ, 144 datasets were collected and used to train an artificial neural network (ANN) model with four electrospinning parameters (molecular weight, solution concentration, flow rate, and tip-to-collector distance) as inputs. The ANN model showed high predictive accuracy, with correlation coefficients exceeding 0.97 and prediction errors below 4%. A response surface methodology (RSM) model was also developed for comparison, but showed limited predictive capability for unseen conditions, with errors up to 28.57%. The ANN exhibited superior reliability and generalizability. Index of relative importance (IRI) and contour analyses revealed molecular weight and concentration as dominant factors. The proposed integration of automated image analysis with ANN provides a data-driven and scalable framework for intelligent design and optimisation of electrospun materials, with potential applicability across diverse fibrous manufacturing systems.

Assessing posture and measuring different angles, such as the Q-angle in the lower limbs, are important findings for clinical examination and the planning of therapeutic interventions. An increased Q-angle has been considered a risk factor for many disorders and injuries. Photogrammetry is a widely used non-invasive technique for measuring aspects of posture. However, the application of this technique depends on the measurement procedure, the training of the assessor, and the image processing tool used. This study aimed to investigate the reliability of the photogrammetry technique in the evaluation of the lower limb for Q-angle using the Leonardo analysis system. Fifty-nine participants (aged 18-35) were recruited from the University of Patras. An assessment of test-retest and inter-rater reliability was conducted. The procedure included photography and static analysis on the Leonardo PL800 postural analysis system and evaluation of the Q angle by photogrammetry. Reliability was tested using the intraclass correlation coefficient (ICC) to examine the degree of correlation of values between the two measurements, the first completion and the repetitive completion, and between the measurements of the two evaluators of the instrument. Excellent test-retest reliability is recorded for the Q angle on the right lower limb and good test-retest reliability for the left limb; moderate inter-rater reliability among technical photogrammetry researchers through the Leonardo posture analysis system is shown for the evaluation of the Q angle of both lower limbs. Further studies are needed to investigate the system's reliability at other body sites.

Introduction: It is well-established that lowering low-density lipoprotein cholesterol (LDL-C) levels is associated with cardiovascular risk reduction, with clinical benefits increasing over time; however, previous analyses may have underestimated the real treatment effect by not accounting for yearly changes in LDL-C levels. Research Question: The aim of this study was to assess the reduction in major cardiovascular event (MCE) risk after each year of statin therapy, considering the actual LDL-C reduction achieved annually. Methods: We included statin randomized controlled trials (RCTs) from the Cholesterol Treatment Trialists' Collaboration meta-analyses that reported the Kaplan-Meyer curve for MCE (defined as a composite of acute myocardial infarction, non-hemorrhagic stroke, cardiac death, and ischemia-driven revascularization) and LDL-C change after each year of follow-up. We used an online image-processing tool ( WebPlotDigitizer 4.5 ) to extract trial-level data from event curves and LDL-C graphs, to estimate hazard ratios (HRs) with 95% confidence intervals (95% CI) for therapy-outcome associations and to track LDL-C changes after each year of follow-up. HRs for MCE and LDL-C measurement after each year of treatment were then scaled to a 1 mmol/L reduction. For each year of follow-up, HRs standardized for 1 mmol/L were pooled through a random-effects meta-analysis. Results: We included a total of 13 RCTs with statins enrolling 99,360 participants. A 1 mmol/L reduction in LDL-C was associated with increasing cardiovascular benefit over time, with a consistent trend across trials lasting up-to 3, 4 and 5 years ( Figure ). In the latter case, the pooled data from 9 RCTs with data available up to 5 years showed a risk reduction per 1 mmol/L ranging from 13% after one year (HR: 0.87; 95% CI: 0.73-1.04) to a 22% reduction after 5 years (HR: 0.78; 95% CI: 0.71-0.86). Conclusions: Our findings reinforce the notion that prolonged exposure to lower LDL-C levels translates into greater cardiovascular risk reduction over time, even for the same reduction in LDL-C levels, thus highlighting the need for timely and persistent management of hypercholesterolemia. These insights should be integrated into clinical guidelines to further optimize long-term cardiovascular prevention strategies.