Deep Learning Architecture Research Articles

Deep Learning-Driven Single-Lead ECG Classification: A Rapid Approach for Comprehensive Cardiac Diagnostics

Background/Objectives: This study aims to address the critical need for accessible, early, and accurate cardiac di-agnostics, especially in resource-limited or remote settings. By shifting focus from traditional multi-lead ECG analysis to single-lead ECG data, this research explores the potential of advanced deep learning models for classifying cardiac conditions, including Nor-mal, Abnormal, Previous Myocardial Infarction (PMI), and Myocardial Infarction (MI). Methods: Five state-of-the-art deep learning architectures—Inception, DenseNet201, MobileNetV2, NASNetLarge, and VGG16—were systematically evaluated on individual ECG leads. Key performance metrics, such as model accuracy, inference time, and size, were analyzed to determine the optimal configurations for practical applications. Results: VGG16 emerged as the most accurate model, achieving an F1-score of 98.11% on lead V4 with a prediction time of 4.2 ms and a size of 528 MB, making it suitable for high-precision clinical settings. MobileNetV2, with a compact size of 13.4 MB, offered a balanced performance, achieving a 97.24% F1-score with a faster inference time of 3.2 ms, positioning it as an ideal candidate for real-time monitoring and telehealth applications. Conclusions: This study bridges a critical gap in cardiac diagnostics by demonstrating the feasibility of lightweight, scalable, single-lead ECG analysis using advanced deep learning models. The findings pave the way for deploying portable diagnostic tools across diverse settings, enhancing the accessibility and efficiency of cardiac care globally.

A GraphKAN-Based Intelligent Fault Diagnosis Method of Rolling Bearing Under Variable Working Conditions

Unsupervised domain adaptation (UDA) can effectively address the two main drawbacks of transfer learning: the requirement of a large number of samples collected from different working conditions, and the inherent defects of convolutional neural networks (CNNs). In the realm of UDA, it is essential to leverage three types of information: class labels, domain specifications, and data organization. These components play a vital role in linking the source domain with the target domain. A technique aimed at identifying issues in rolling bearings is presented, employing an integration of CNN-KAN and GraphKAN structures to support the UDA methodology. A cohesive deep learning architecture is employed to represent the three types of information involved in UDA. The initial two types of information are represented through the roles of classifier and domain discriminator. To begin with, an architecture leveraging CNN-KAN is employed to extract features from the incoming signals. Following this, the features obtained from the CNN-KAN architecture are input into a specially developed graph creation layer that constructs instance graphs by analyzing the relationships among the structural characteristics found within the samples. In the following step, an innovative GraphKAN model is applied to illustrate the instance graphs, concurrently employing CORrelation ALignment (CORAL) loss to assess the structural discrepancies among instance graphs from different domains. Results from experiments conducted on two separate datasets demonstrate that the proposed framework surpasses alternative approaches and successfully recognizes transferable characteristics that are advantageous for domain adaptation.

Non-iterative generation of optimized meshes for finite element simulations with deep learning

The finite element method is one of the most widely used computational methods in engineering and science. It provides approximate solutions to boundary value problems. The quality of these solutions critically depends on the underlying discretization, the so-called mesh. To optimize the mesh, adaptive refinement methods have been proposed over the last years that can improve mesh quality over a series of iteration steps. Herein, we propose a novel deep learning architecture that can cut short the process of mesh optimization. This architecture exploits fundamental invariance and equivariance properties to keep the amount of training data modest. It can generate high-quality meshes for a given boundary value problem and a desired target approximation error in a direct, non-iterative way. We demonstrate the performance of our method by the application to standard two-dimensional linear-elastic elasticity problems. There, our method generates meshes that reduce the solution error by 22.6% (median) compared to uniform meshes with the same computational demand.

Automated Fungal Identification with Deep Learning on Time-Lapse Images

The identification of species within filamentous fungi is crucial in various fields such as agriculture, environmental monitoring, and medical mycology. Traditional identification methods based on morphology have a low demand for advanced equipment usage and heavily depend on manual observation and expertise. However, this approach may struggle to differentiate between species in a genus due to their potential visual similarities, making the process time-consuming and subjective. In this study, we present an AI-based fungal species recognition model that utilizes deep learning techniques applied to time-lapse images. The training dataset, derived from fungi strains in the IBT Culture Collection, comprised 26,451 high-resolution images representing 110 species from 35 genera. The dataset was divided into a training set and validation subsets. We implemented three advanced deep learning architectures—ResNet50, DenseNet-121, and Vision Transformer (ViT)—to assess their effectiveness in accurately classifying fungal species. By utilizing images from early growth stages (days 2–3.5) for training and testing and later stages (days 4–7) for validation, our approach shortens the fungal identification process by 2–3 days, significantly reducing the associated workload and costs. Among the models, the Vision Transformer achieved the highest accuracy of 92.6%, demonstrating the effectiveness of our method. This work contributes to the automation of fungal identification, providing a reliable and efficient solution for monitoring fungal growth and diversity over time, which would be useful for culture collections or other institutions that handle a large number of new isolates in their daily work.

Multi-Classification of Skin Lesion Images Including Mpox Disease Using Transformer-Based Deep Learning Architectures

Background/Objectives: As reported by the World Health Organization, Mpox (monkeypox) is an important disease present in 110 countries, mostly in South Asia and Africa. The number of Mpox cases has increased rapidly, and the medical world is worried about the emergence of a new pandemic. Detection of Mpox by traditional methods (using test kits) is a costly and slow process. For this reason, there is a need for methods that have high success rates and can diagnose Mpox disease from skin images with a deep-learning-based autonomous method. Methods: In this work, we propose a multi-class, fast, and reliable autonomous disease diagnosis model using transformer-based deep learning architectures and skin lesion images, including for Mpox disease. Our other aim is to investigate the effects of self-supervised learning, self-distillation, and shifted window techniques on classification success when multi-class skin lesion images are trained with transformer-based deep learning architectures. The Mpox Skin Lesion Dataset, Version 2.0, which was publicly released in 2024, was used in the training, validation, and testing processes of the study. Results: The SwinTransformer architecture we proposed in our study achieved about 8% higher accuracy evaluation metric classification success compared to its closest competitor in the literature. ViT, MAE, DINO, and SwinTransformer architectures achieved 93.10%, 84.60%, 90.40%, and 93.71% accuracy classification success, respectively. Conclusions: The results obtained in the study showed that Mpox disease and other skin lesion images can be diagnosed with high success and can support doctors in decision-making. In addition, the study provides important results that can be used in other medical fields where the number of images is low in terms of transformer-based architecture and technique to use.

Exploring Applications of Artificial Intelligence in Critical Care Nursing: A Systematic Review

Background: Artificial intelligence (AI) has been increasingly employed in healthcare across diverse domains, including medical imaging, personalized diagnostics, therapeutic interventions, and predictive analytics using electronic health records. Its integration is particularly impactful in critical care, where AI has demonstrated the potential to enhance patient outcomes. This systematic review critically evaluates the current applications of AI within the domain of critical care nursing. Methods: This systematic review is registered with PROSPERO (CRD42024545955) and was conducted in accordance with PRISMA guidelines. Comprehensive searches were performed across MEDLINE/PubMed, SCOPUS, CINAHL, and Web of Science. Results: The initial review identified 1364 articles, of which 24 studies met the inclusion criteria. These studies employed diverse AI techniques, including classical models (e.g., logistic regression), machine learning approaches (e.g., support vector machines, random forests), deep learning architectures (e.g., neural networks), and generative AI tools (e.g., ChatGPT). The analyzed health outcomes encompassed postoperative complications, ICU admissions and discharges, triage assessments, pressure injuries, sepsis, delirium, and predictions of adverse events or critical vital signs. Most studies relied on structured data from electronic medical records, such as vital signs and laboratory results, supplemented by unstructured data, including nursing notes and patient histories; two studies also integrated audio data. Conclusion: AI demonstrates significant potential in nursing, facilitating the use of clinical practice data for research and decision-making. The choice of AI techniques varies based on the specific objectives and requirements of the model. However, the heterogeneity of the studies included in this review limits the ability to draw definitive conclusions about the effectiveness of AI applications in critical care nursing. Future research should focus on more robust, interventional studies to assess the impact of AI on nursing-sensitive outcomes. Additionally, exploring a broader range of health outcomes and AI applications in critical care will be crucial for advancing AI integration in nursing practices.

UNET-FLIM: A Deep Learning-Based Lifetime Determination Method Facilitating Real-Time Monitoring of Rapid Lysosomal pH Variations in Living Cells.

Lifetime determination plays a crucial role in fluorescence lifetime imaging microscopy (FLIM). We introduce UNET-FLIM, a deep learning architecture based on a one-dimensional U-net, specifically designed for lifetime determination. UNET-FLIM focuses on handling low photon count data with high background noise levels, which are commonly encountered in fast FLIM applications. The proposed network can be effectively trained using simulated decay curves, making it adaptable to various time-domain FLIM systems. Our evaluations of simulated data demonstrate that UNET-FLIM robustly estimates lifetimes and proportions, even when the signal photon count is extremely low and background noise levels are high. Remarkably, UNET-FLIM's insensitivity to noise and minimal photon count requirements facilitate fast FLIM imaging and real-time lifetime analysis. We demonstrate its potential by applying it to monitor rapid lysosomal pH variations in living cells during in situ acetic acid treatment, all without necessitating any modifications to existing FLIM systems.

Voice Recordings from Short Mobile Sessions versus Clinical Interviews for Mental Illness Screening: A Comparative Study with Deep Transfer Learning

The prevalence of mental illness symptoms is increasing. While questionnaires are traditionally used, prior research applied machine learning to clinical interview recordings conducted by a virtual agent for depression screening. In this research, we compare their screening ability with brief mobile voice recordings. The latter reduces completion time while increasing access to screening. We collected 546 mobile voice recordings in response to four clinical interview questions from over 200 crowd-sourced adults reporting English fluency. We also collected responses to traditional screening questionnaires to use as labels in the classifiers. We extract textual transcripts from these voice recordings, facilitating both unimodal and multimodal classification. We compose acoustic and language representation models into deep learning architectures fine-tuned for mental illness screening. In our study, we compare the mental illness screening ability of four questions from mobile voice recordings and complete clinical interviews. When screening for depression, the brief mobile voice responses to individual questions proved as good or better for half of the interview questions. Overall, we achieved AUC of 0.76 for depression, 0.76 for anxiety, and 0.81 for suicidal ideation screening using our mobile voice recordings. Our findings can help guide the development of efficient and effective mental illness screening technologies.

Enhancing the Prediction of Flow Characteristics in an Inventive Plate Heat Exchanger Using Deep Learning Techniques

Abstract In modern energy research, minimizing fuel usage and harmful gas emissions are critical priorities. The application of advanced deep learning (DL) models to predict thermohydraulic characteristics of an innovative plate heat exchanger (IPHE) is investigated in this study. Building upon our prior work utilizing machine learning (ML) models, the focus is placed on predicting the Nusselt Number (Nu), friction factor (f), and performance (P) within a Reynolds number range of 500 to 5000. Advanced DL architectures—GRU, LSTM, and CNN—are utilized, resulting in substantial improvements in prediction accuracy and robustness. The LSTM model demonstrates superior performance, achieving R² scores of 0.9986, 0.9985, and 0.9968 for Nu, f, and P, respectively, significantly surpassing prior ML model results of 0.98, 0.979, and 0.9628. The findings highlight the capacity of DL models to capture complex, nonlinear relationships in thermohydraulic data, offering an enhanced approach to optimizing plate heat exchanger (PHE) performance. This work contributes to energy-efficient technological advancements, supporting global efforts to reduce environmental impacts while addressing rising energy demands.

Aerobic Stress Detection in Aquatic Environments with Water Quality Data Using Hybrid Deep Learning Based ConvRec Model

Depletion of dissolved oxygen in the water is a serious threat to fish and other aquatic organisms, it causes aerobic stress disease in fish. Detection of aerobic stress is crucial to maintain better growth and spawning in the fishes. Recently many studies proposed deep learning-based water quality analysis techniques, but these techniques inadequate in handling the complex water quality data. Because water quality has both spatial and temporal characteristics, this makes most of the deep learning models inadequate. To handle such complex and multifaceted data we proposed ConvRec, a deep learning architecture that incorporates CNN (Convolution neural network) and LSTM (Long-short term network) structures. CNN component extracts feature in the spatial domain from the water quality data from different locations while LSTM captures temporal features hence the model can learn both spatial and temporal correlations between the movement of water quality parameters to classify the aerobic stress in aqua ponds. In this work we use the two dataset both are unlabelled collected using IoT (Internet of things) devices. To handle this data using ConvRec model, usus the fine-grained annotation of data points that have the effect of empowering the model to detect relevant traits associated with oxygen stress in fish. It can be therefore ascertained that ConvRec yields high degrees of accuracy of 99.2% and 99.65%, on the “ponds” and “waterx” datasets respectively while the past models can only 98.2% and 98.1% respectively on the same datasets. These results demonstrate that ConvRec is not only promising for estimating the health of fish during oxygen deficiency but also it can take part in reducing the negative impact of low oxygen levels in the water on fish.

A Benchmark Dataset for Tornado Detection and Prediction using Full-Resolution Polarimetric Weather Radar Data

Abstract Weather radar is the primary tool used by forecasters to detect and warn for tornadoes in near-real time. In order to assist forecasters in warning the public, several algorithms have been developed to automatically detect tornadic signatures in weather radar observations. Recently, Machine Learning (ML) algorithms, which learn directly from large amounts of labeled data, have been shown to be highly effective for this purpose. Since tornadoes are extremely rare events within the corpus of all available radar observations, the selection and design of training datasets for ML applications is critical for the performance, robustness, and ultimate acceptance of ML algorithms. This study introduces a new benchmark dataset, TorNet, to support development of ML algorithms in tornado detection and prediction. TorNet contains full-resolution, polarimetric, Level-II WSR-88D data sampled from 9 years of reported storm events. A number of ML baselines for tornado detection are developed and compared, including a deep learning (DL) architecture capable of processing raw radar imagery without the need for manual feature extraction required for traditional ML algorithms. Despite not benefiting from manual feature engineering or other preprocessing, the DL model shows increased detection performance compared to non-DL and operational baselines. The TorNet dataset, as well as source code and model weights of the DL baseline trained in this work, are made freely available.

Hydrological Dynamics and Climate Variability in the Sava River Basin: Streamflow Reconstructions Using Tree-Ring-Based Paleo Proxies

This study reconstructs historical streamflow in the Sava River Basin (SRB), focusing on hydrological variability over extended timescales. Using a combination of Machine Learning (ML) and Deep Learning (DL) models, streamflow patterns were reconstructed from self-calibrated Palmer Drought Severity Index (scPDSI) proxies. The analysis included nine ML models and two DL architectures, with a post-prediction bias correction applied uniformly using the RQUANT method. Results indicate that ensemble methods, such as Random Forest and Gradient Boosted Tree, along with a six-layer DL model, effectively captured streamflow dynamics. Bias correction improved predictive consistency, particularly for models exhibiting greater initial variability, aligning predictions more closely with observed data. The findings reveal that the 2000–2022 period ranks as the lowest 23-year flow interval in the observed record and one of the driest over the past ~500 years, offering historical context for prolonged low-flow events in the region. This study demonstrates the value of integrating advanced computational methods with bias correction techniques to extend hydrological records and enhance the reliability of reconstructions. By addressing data limitations, this approach provides a foundation for supporting evidence-based water resource management in Southeastern Europe under changing climatic conditions.

Enabling Multi-Step Forecasting with Structured State Space Learning Module

Data-driven soft sensor incorporated with the model predictive control (MPC) algorithms facilitating product quality and cost control are of imperative importance in industrial processes. However, the widely used one-step forecasting method can not incorporate with MPC and therefore restricts the practical usage of soft sensor. Multi-step forecasting introduces long-term dependencies problems yet has not been effectively resolved within traditional model structure. To address this problem, this paper proposes the deep learning network architecture named Extended State Space Learning Module (ESSLM). ESSLM extends the nonlinear mapping architecture of deep learning based on state space and retains state transfer matrices to characterize the dynamics of the system. ESSLM distinguishes itself from explicit network architectures such as gated RNNs by addressing the long-term dependencies problems through an implicit initialization method, and the MLP and RNN algorithms can be regarded as the manifestation of ESSLM in special cases. ESSLM characterizes the latent space as the coefficients of the orthogonal basis functions so that the input data can be encoded into a high-dimensional feature space with minimal information loss which efficiently achieves multi-step forecasting and give greater utility and practical significance.

Deep learning architectures for location and identification in storage systems

Deep learning architectures for location and identification in storage systems

Discovery of anticancer peptides from natural and generated sequences using deep learning.

Anticancer peptides (ACPs) demonstrate significant potential in clinical cancer treatment due to their ability to selectively target and kill cancer cells. In recent years, numerous artificial intelligence (AI) algorithms have been developed. However, many predictive methods lack sufficient wet lab validation, thereby constraining the progress of models and impeding the discovery of novel ACPs. This study proposes a comprehensive research strategy by introducing CNBT-ACPred, an ACP prediction model based on a three-channel deep learning architecture, supported by extensive in vitro and in vivo experiments. CNBT-ACPred achieved an accuracy of 0.9554 and a Matthews Correlation Coefficient (MCC) of 0.8602. Compared to existing excellent models, CNBT-ACPred increased accuracy by at least 5% and improved MCC by 15%. Predictions were conducted on over 3.8 million sequences from Uniprot, along with 100,000 sequences generated by a deep generative model, ultimately identifying 37 out of 41 candidate peptides from >30 species that exhibited effective in vitro tumor inhibitory activity. Among these, tPep14 demonstrated significant anticancer effects in two mouse xenograft models without detectable toxicity. Finally, the study revealed correlations between the amino acid composition, structure, and function of the identified ACP candidates.

STCNet: Spatio-Temporal Cross Network with subject-aware contrastive learning for hand gesture recognition in surface EMG.

This paper introduces the Spatio-Temporal Cross Network (STCNet), a novel deep learning architecture tailored for robust hand gesture recognition in surface electromyography (sEMG) across multiple subjects. We address the challenges associated with the inter-subject variability and environmental factors such as electrode shift and muscle fatigue, which traditionally undermine the robustness of gesture recognition systems. STCNet integrates a convolutional-recurrent architecture with a spatio-temporal block that extracts features over segmented time intervals, enhancing both spatial and temporal analysis. Additionally, a rolling convolution technique designed to reflect the circular band structure of the sEMG measurement device is incorporated, thus capturing the inherent spatial relationships more effectively. We further propose a subject-aware contrastive learning framework that utilizes both subject and gesture label information to align the representation of vector space. Our comprehensive experimental evaluations demonstrate the superiority of STCNet under aggregated conditions, achieving state-of-the-art performance on benchmark datasets and effectively managing the variability among different subjects. The implemented code can be found at https://github.com/KNU-BrainAI/STCNet.

Lung and Colon Cancer Classification Using Multiscale Deep Features Integration of Compact Convolutional Neural Networks and Feature Selection

The automated and precise classification of lung and colon cancer from histopathological photos continues to pose a significant challenge in medical diagnosis, as current computer-aided diagnosis (CAD) systems are frequently constrained by their dependence on singular deep learning architectures, elevated computational complexity, and their ineffectiveness in utilising multiscale features. To this end, the present research introduces a CAD system that integrates several lightweight convolutional neural networks (CNNs) with dual-layer feature extraction and feature selection to overcome the aforementioned constraints. Initially, it extracts deep attributes from two separate layers (pooling and fully connected) of three pre-trained CNNs (MobileNet, ResNet-18, and EfficientNetB0). Second, the system uses the benefits of canonical correlation analysis for dimensionality reduction in pooling layer attributes to reduce complexity. In addition, it integrates the dual-layer features to encapsulate both high- and low-level representations. Finally, to benefit from multiple deep network architectures while reducing classification complexity, the proposed CAD merges dual deep layer variables of the three CNNs and then applies the analysis of variance (ANOVA) and Chi-Squared for the selection of the most discriminative features from the integrated CNN architectures. The CAD is assessed on the LC25000 dataset leveraging eight distinct classifiers, encompassing various Support Vector Machine (SVM) variants, Decision Trees, Linear Discriminant Analysis, and k-nearest neighbours. The experimental results exhibited outstanding performance, attaining 99.8% classification accuracy with cubic SVM classifiers employing merely 50 ANOVA-selected features, exceeding the performance of individual CNNs while markedly diminishing computational complexity. The framework’s capacity to sustain exceptional accuracy with a limited feature set renders it especially advantageous for clinical applications where diagnostic precision and efficiency are critical. These findings confirm the efficacy of the multi-CNN, multi-layer methodology in enhancing cancer classification precision while mitigating the computational constraints of current systems.

Automating container damage detection with the YOLO-NAS deep learning model.

Ensuring the integrity of shipping containers is crucial for maintaining product quality, logistics efficiency, and safety in the global supply chain. Damaged containers can lead to significant economic losses, delays, and safety hazards. Traditionally, container inspections have been manual, which are labor-intensive, time-consuming, and error-prone, especially in busy port environments. This study introduces an automated solution using the YOLO-NAS model, a cutting-edge deep learning architecture known for its adaptability, computational efficiency, and high accuracy in object detection tasks. Our research is among the first to apply YOLO-NAS to container damage detection, addressing the complex conditions of seaports and optimizing for high-speed, high-accuracy performance essential for port logistics. Our method showcases YOLO-NAS's superior efficacy in detecting container damage, achieving a mean average precision (mAP) of 91.2%, a precision rate of 92.4%, and a recall of 84.1%. Comparative analyses indicate that YOLO-NAS consistently outperforms other leading models like YOLOv8 and Roboflow 3.0, which showed lower mAP, precision, and recall values under similar conditions. Additionally, while models such as Fmask-RCNN and MobileNetV2 exhibit high training accuracy, they lack the real-time assessment capabilities critical for port applications, making YOLO-NAS a more suitable choice. The successful integration of YOLO-NAS for automated container damage detection has significant implications for the logistics industry, enhancing port operations with reliable, real-time inspection solutions that can seamlessly integrate into predictive maintenance and monitoring systems. This approach reduces operational costs, improves safety, and lessens the reliance on manual inspections, contributing to the development of "smart ports" with higher efficiency and sustainability in container management.

Deep Learning-Driven IoT Defence: Comparative Analysis of CNN and LSTM for DDoS Detection and Mitigation

The extensive utilization of Internet of Things (IoT) devices has revolutionized multiple sectors, ranging from smart homes to industrial automation, while concurrently broadening the attack surface for cyber threats, including Distributed Denial of Service (DDoS) attacks. This study examines the efficacy of Convolutional Neural Networks (CNNs) and Long Short-Term Memory Networks (LSTMs) in detecting DDoS attacks, focusing on the distinct security concerns presented by IoT networks. Employing the extensive CICDDoS2019 dataset, these algorithms scrutinize individual IP flow records to attain real-time anomaly identification with elevated precision. The evaluation results reveal that both CNN and LSTM models exhibit strong performance, with CNNs showing enhanced precision (99.42%) and F1-score (99.26%) due to their capacity to extract spatial patterns from multidimensional traffic data. Although LSTMs are proficient in capturing temporal dependencies, their elevated computing demands render them less appropriate for real-time applications in resource-limited IoT settings. This paper emphasizes CNNs as a scalable and efficient option for IoT network defence and advocates for more research into hybrid deep learning architectures to improve anomaly detection.

Universal conditional networks (UniCoN) for multi-age embryonic cartilage segmentation with Sparsely annotated data

Osteochondrodysplasia, affecting 2–3% of newborns globally, is a group of bone and cartilage disorders that often result in head malformations, contributing to childhood morbidity and reduced quality of life. Current research on this disease using mouse models faces challenges since it involves accurately segmenting (precisely delineating) the developing cartilage in 3D micro-CT images of embryonic mice. Tackling this segmentation task with deep learning (DL) methods is laborious due to the big burden of manual image annotation, expensive due to the high acquisition costs of 3D micro-CT images, and difficult due to embryonic cartilage’s complex and rapidly changing shapes. While DL approaches have been proposed to automate cartilage segmentation, most such models have limited accuracy and generalizability, especially across data from different embryonic age groups. To address these limitations, we propose novel DL methods that can be adopted by any DL architectures—including Convolutional Neural Networks (CNNs), Transformers, or hybrid models—which effectively leverage age and spatial information to enhance model performance. Specifically, we propose two new mechanisms, one conditioned on discrete age categories and the other on continuous image crop locations, to enable an accurate representation of cartilage shape changes across ages and local shape details throughout the cranial region. Extensive experiments on multi-age cartilage segmentation datasets show significant and consistent performance improvements when integrating our conditional modules into popular DL segmentation architectures. On average, we achieve a 1.7% Dice score increase with minimal computational overhead and a 7.5% improvement on unseen data. These results highlight the potential of our approach for developing robust, universal models capable of handling diverse datasets with limited annotated data, a key challenge in DL-based medical image analysis.