The accurate diagnosis of brain tumors remains a major challenge in medical imaging because tumor structures vary in size, shape, and appearance. Traditional methods such as manual segmentation and classification are time-consuming and may produce inconsistent results due to observer variation. This paper presents an approach that combines the Internet of Medical Things (IoMT) with deep learning models to improve the accuracy and efficiency of brain tumor diagnosis. IoMT enables the collection and transfer of Magnetic Resonance Imaging (MRI) data to cloud platforms for real-time analysis and automated processing. The proposed framework uses Swin UNet for segmentation of tumor regions and EfficientNet-V2 for classification into tumor subtypes. Swin UNet uses transformer-based attention mechanisms to capture multi-scale spatial features, while EfficientNet-V2 supports efficient feature learning for classification. Experiments performed on the BRATS 2020 dataset demonstrate the effectiveness of the framework, achieving a segmentation Intersection over Union (IoU) of 78% and a classification accuracy of 97%. The integration of IoMT supports remote accessibility, faster clinical decision-making, reduced manual effort, and automated workflows. The proposed framework improves reliability and performance compared to conventional CNN-based systems. This study highlights the potential of AI-driven medical diagnosis and provides a scalable solution for practical healthcare applications. Future work will focus on real-time deployment and extension to other medical imaging tasks. The framework also reduces computational complexity and supports patient data, making it suitable for clinical environments where quick and accurate diagnosis is important for effective treatment planning, better patient management, and improved healthcare services today globally.
Despite the progress in Neural Machine Translation (NMT), translating ambiguous and context rich content remains a major challenge, especially in low-resource language pairs like English-Hindi. Traditional NMT systems often fail to solve these challenges due to their reliance on textual data alone. Multimodal approaches, particularly those incorporating visual context, offer promising solutions to the task by resolving linguistic ambiguities. To address this, a novel solution is introduced through a Visual Scene-Aware Hindi Subtitles Dataset (VISA-HIN), designed specifically for English-Hindi Video-Guided Multimodal Machine Translation (VMMT). This dataset aligns English subtitles with corresponding video frames and provides Hindi translations. Alongside the dataset, this study propose a video-guided MMT framework that leverages visual cues to enhance translation quality. The results of the experiments show the potential of scene aware information to improve contextual understanding and fluency in English-to-Hindi translation, paving the way for more robust and accurate multimodal translation systems in low-resource settings.
These days, zinc and its composites have attracted significant interest in the field of biomedical engineering due to their bio absorbability. Micro Electrical Discharge Drilling (& micro;-EDD) was used for making micro perforations in Zn/(Fe+Ag+Mg)-MMC for cardiovascular stent application. The influence of & micro;-EDD process parameters such as peak current, pulse-off time, pulse-on time, tool rotation, tool material, and electrode polarity were studied with respect To Material Removal Rate (MRR), Drill Overcut (DO), and Tool Wear Rate (TWR) during the micro-drilling of the MMC. Further, Taguchi and GRA based optimization were performed. From the experimental study it was found that MRR, DO and TWR increases with peak current, pulse-on-time and rotation of tool electrode. Optimization improved the MRR, DO and TWR by 111.52%, 20.83% and 76.79% respectively. Peak current and pulse-on-time was found significant for MRR DO and TWR whereas peak current and pulse-off-time was found significant for GRA based analysis. SEM examinations of the micro-drilled hole show that high circularity and a good surface finish was achieved.
Corrosion is a rising concern in oil and gas sectors, which causes destruction of the material and results in heavy financial loss accompanied by high risk to human health and safety. This study investigates the corrosion inhibition efficiency of a novel biomaterial based nanostructured coating on carbon steel specimen by dip-coating technique using Ziziphus Spina leaf extract, silica nanoparticles (SiO2NPs), and biopolymer (chitosan) matrix. The anti-corrosion performance of the fabricated composite thin film was investigated by exposing the coated specimen at different environmental conditions and the effectiveness was examined using potentiostatic, weight loss measurement, atmosphere and wet/dry tests. The main characterization tools employed are Atomic force microscopy (AFM), X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), and Field emission scanning electron microscopy (FE-SEM). Tafal plot was used to study the corrosion rate. The results of atomic-level micro structural analysis performed by FE-SEM and AFM suggested the successful development of inhibitor layer at the carbon steel surface to prevent the corrosion. The experimental outcomes underscore the potential of biomaterial based nanostructured coating as an effective corrosion inhibitor for carbon steel in acidic environments. The nanocomposite thin films act as an efficient protective coating and offers superior corrosion resistance with extended lifespan of the carbon steel. The study validates that the green corrosion inhibitors developed from natural resources with minimum layer thickness could be a realistic solution in regulating the corrosion in oil pipelines with improved corrosion resistance, exceptional chemical stability, durability, and improved strength.
Fog Computing (FC) is gradually essential in diminishing communication latency and enhancing resource usage for Internet of Things (IoT) tasks; though, critical experiments like resource overloading, security weaknesses, and excessive energy consumption frequently hinder its operational potential. The range of this study includes the expansion of a trust-based, energy-aware secure load matching and resource provisioning system, definitely calculated for decentralized fog architectures using a Multi-Objective Optimization Algorithm (MOA). The methodology incorporates a Many-to-Few (M2F) balancer for effective task distribution, ordering trust in node-task assignments to certify consistency. Resource management is significantly improved through the hybridization of Improved Salp Swarm Optimization (ISSO) and Modified Whale Optimization Algorithm (MWOA) for dynamic provision. To bolster system integrity, an intrusion detection system with offloading mechanisms is implemented alongside Hierarchical Collaborative Federated Learning (HCFL) to raise secure, privacy-preserving node association. Key results from performance assessments showed in the Matrix Laboratory (MATLAB) establish a high 88% Average Resource Utilization (ARU) and a steady 1300s response time. The model's strength is further supported by a precision of 99.5%, an F-measure of 91.0%, and a recall of 78.6% during oppositional testing. The individuality of this study lies in its concurrent optimization of trust, energy, and load metrics within a single meta-heuristic framework. This system offers a practical value for mission-critical IoT ecosystems, such as smart grids and industrial automation, where real-time handling and data reliability are mandatory for keeping complete system strength and performance.
