Retinal vessel segmentation plays a crucial role in the early diagnosis and treatment of ophthalmic diseases, yet manual segmentation methods are time-consuming and prone to human error. The increasing demand for automated solutions has led to the exploration of deep learning-based approaches that improve segmentation accuracy and efficiency. However, existing methods often struggle with poor image resolution, a lack of robustness across datasets, and challenges in detecting fine vessel structures, limiting their applicability in large-scale clinical settings. To address these limitations, this study proposes a deep learning-based framework using high-resolution network (HRNet) combined with an advanced preprocessing pipeline, including grayscale conversion, gamma correction, CLAHE, and normalization, to enhance image quality for segmentation. The proposed HRNet-based framework was evaluated on two publicly available datasets, DRIVE and CHASE_DB1. It achieved the highest sensitivity of 0.8193 on DRIVE and the highest specificity of 0.9807 on CHASE_DB1, demonstrating its capability to balance fine vessel detection and false positive reduction. These findings indicate that the proposed framework effectively improves the segmentation accuracy of ophthalmoscope images, enhancing its potential for clinical decision-making in ophthalmology.
Diffusion tensor imaging (DTI) has emerged as a powerful neuroimaging modality for investigating white matter microstructure and its alterations following brain injury. This review presents a comprehensive overview of DTI, encompassing its physical principles, mathematical modeling of diffusion tensors, and known limitations of the technique. We explore key methodological considerations, including acquisition protocols, preprocessing pipelines, vendor-related variability, atlas registration, and the role of diffusion phantoms in calibration. With the rise of big data in medical imaging, we highlight the influence of large-scale, multisite datasets and open-source neuroimaging repositories in advancing DTI research. A central focus is placed on the application of DTI in mild traumatic brain injury, a condition that often eludes detection in conventional imaging settings. We evaluate emerging computational strategies, including Z-score analysis, principal component analysis, random forests, and generative adversarial networks-that improve the sensitivity, specificity, and interpretability of DTI metrics in both clinical and research settings. By bridging methodological rigor with translational insight, this review underscores the evolving potential of DTI as a neuroimaging biomarker for brain injury assessment.
Accurate and effective diagnosis of skin diseases is crucial for clinical decision-making. However, there are still some challenges, including irregular lesion morphologies, class imbalance between rare and common types, and performance degradation in complex scenarios characterized by noise or occlusion. To address these issues, an improved skin disease detection algorithm is proposed based on YOLOv8n, which features three core innovations. First, it integrates deformable large kernel attention (D-LKA) into deep backbone layers to capture global contextual relationships of lesions; then it embeds DCNv3 deformable convolutions in mid-layers to adaptively sample irregular lesion boundaries; finally, it designs an EMA-Slide Loss function to dynamically weight hard-to-classify samples, thereby reducing bias toward common categories. After evaluating on the International Skin Imaging Collaboration (ISIC) dataset (with labels validated by board-certified dermatologists), the algorithm can achieve 96.58% mAP50 and 88.32% mAP50-95, 2.44% and 2.52% higher than the baseline YOLOv8n, respectively. It maintains a real-time inference speed of 31 ms per image, making it suitable for edge devices such as portable dermatoscopes. Supporting nine common types of skin diseases, with extensibility to accommodate rare types and multi-modal data fusion, this work provides a clinically actionable tool for automated skin lesion analysis, bridging the gap between algorithmic performance and real-world clinical demands.
A fractional-order (hepatitis B virus) HBV transmission model is developed using Caputo, Caputo-Fabrizio, and Atangana-Baleanu-Caputo (ABC) operators to capture memory effects and non-local interactions-including asymptomatic carriers and vertical transmission. Existence, uniqueness, positivity, and stability of solutions are established via fixedpoint theorems and Ulam-Hyers criteria. Numerical simulations employ Adams-Bashforth-Moulton and predictor-corrector methods adapted to each fractional derivative. Results show that lowering the fractional order (α < 1) reduces peak viral load, especially with early intervention, but extends infection duration due to slower decay. This fractional modelling framework offers deeper insight into chronic HBV dynamics, vaccination effectiveness, and long-term control strategies.
A multitude of studies have investigated the identification of unique antigens and genetic traits associated with cardiac stem cells (CSCs) since their discovery. This meta-analysis aims to assess the safety and efficacy of cardiac stem cells (CSCs), elucidate existing knowledge regarding their potential applications, and compare them with other novel therapeutic strategies for cardiac repair, including bone marrow cells (BMCs) and mesenchymal stem cells (MSCs). Researchers conducted a search of important indexing databases, including PubMed, Scopus, Cochrane Central, Web of Science (WOS), CINAHL, and Embase, to identify pertinent papers published from 1980 to January 2025. A total of 74 studies involving 5,420 participants were selected from 459 evaluated for the study. A total of 2,931 participants were allocated to the intervention group, whereas 2,489 were assigned to the placebo or control groups. The research encompassed 49 papers on BMC therapy, 17 on MSC therapy, five on CSC therapy, and three on ADRC therapy. A 0.4% enhancement in left ventricular ejection fraction (LVEF) [95% confidence interval (95% CI): 0.23-0.57; I2: 85%, P < 0.00001] was noted after stem cell therapy and in the stem cell therapy cohort, left ventricular end-systolic volume (LVESV) diminished by -0.33 mL (95% CI: -0.47 to -0.19; I2: 73%, P < 0.00001), whereas left ventricular end-diastolic volume (LVEDV) reduced by -0.18 mL (95% CI: -0.31 to -0.05; I2: 68%, P = 0.006). Furthermore, in this cohort, the six-minute walk test (6MWT) exhibited an increase of 0.2% (95% CI: 0.06-0.34; I2: 0%, P = 0.005), whereas the standardized mean difference of infarct size (IS) demonstrated a reduction of -0.36% (95% CI: -0.60 to -0.12; I2: 68%; P = 0.004). These findings augment the existing evidence and underscore the transformative potential of stem cell therapy in cardiac treatment.
The internet of medical things (IoMT) is regarded as a promising framework, which is used to expand and improve telemedicine services. Cloud-based IoMT refers to the integration of medical devices and sensors with cloud computing infrastructure, enabling real-time remote data collection, processing, storage, and analysis. This architecture supports the efficient management of patient health information and facilitates advanced telemedicine services by offering scalable, secure, and accessible healthcare solutions. Ensuring secure access and communication in such systems is critical, as vulnerabilities in the network can expose sensitive patient data to significant risks. Among various security measures, authentication using biomedical signals, particularly electrocardiogram (ECG) signals, is gaining attention due to their unique, individual-specific characteristics. Therefore, this paper develops a new approach called local-global-graph network-based biokey generation (LGGNet-BioKey) for authentication in Cloud-based IoMT. Initially, the Cloud-based IoMT network is simulated, and it includes three entities, like cloud server, gateway, and patient. First, the public key and security parameters are initialized, and then the entities are registered with the cloud server. Next, the key generation is done using LGGNet, and then the BioKey generation is performed using an ECG signal. Next, the lightweight authentication is done and lastly, attribute-based encryption and decryption are performed in the data preservation phase. Furthermore, the LGGNet-BioKey model measured an execution time, memory usage, and key generation time of 3.772 sec, 9.096 MB, and 3.771 sec.
This study aims to compare the efficacy of magnetic nanoparticle hyperthermia for treating cancerous tissues in two distinct scenarios: breast and muscle/prostate tumors. Heat transfer dynamics during magnetic hyperthermia cancer therapy are explored using intravenously administered nanoparticles to a muscle/prostate tumor and intratumoral injection into a breast tumor. Utilizing non-Newtonian blood rheological models, we analyze a complex geometric domain for both tumor types and apply the mixed finite element technique for solving the governing equations. The impact of varying magnetic field frequencies and injected nanoparticle concentrations on heat transfer and nanoparticle transport within muscle/prostate, and breast tissues are examined numerically. Higher magnetic field frequencies and injected nanoparticle concentrations were found to increase localized heating in tumor regions, reduce therapy duration, and maximize thermal damage to cancer cells for both tumor configurations. This research provides valuable insights for optimizing magnetic hyperthermia parameters for different tumor types and also highlights the potential for personalized treatment strategies.
