Brain tumors are considered among the most life-threatening and painful diseases, resulting in a significantly reduced lifespan. Accurate diagnosis of brain tumors is required to develop treatment plans that can prolong the lives of sufferers. The need for a reliable deep learning method is essential to accurately diagnose brain cancers. Therefore, an Enhanced Brain Tumour Detection from MRI Images using Interpretable Generalized Additive Neural Network optimized with Parrot Optimization (BTD-MRI-IGANN-PO) is proposed in this paper. Initially, the image is collected from MRI dataset. The images are pre-processed using Bayesian Boundary Trend Filtering (BBTF) which used to enhance the image quality and resize the image. Then, the pre-processing images are given into the extraction stage using Separable Synchro Extracting Transform (SSET). It extracts the relevant features such asintensity, texture, shape, and frequency-domain features.Then, extracted features are given into theInterpretable Generalized Additive Neural Network (IGANN) for detect the brain Tumour which classifies as Glioma Tumour, Pituitary Tumour, Meningioma Tumour, No Tumour. The Parrot Optimization (PO) algorithm is employed tooptimize the weight parameters of IGANN. The BTD-MRI-IGANN-PO is executed, itsefficacy is evaluated under some performance metrics, like accuracy, computational time, precision, specificity, and F-measure, recall, ROC and confusion matrix is analyzed. Performance of the BTD-MRI-IGANN-POapproach attains 24.39%, 35.71%, and 25.55% higher accuracy; 20.73%, 13.79%, and 16.47%higher precision; 14.24%, 24.18%, and 14.34% better recallwhen compared with existing techniques:MRI-basedBrain Tumour detection using Convolutional Neural Network (MRI-BTD-CNN), Brain Tumour MRI utilizing Convolutional Neural Network (BT-MRI-CNN), Early diagnosis with Brain Tumour segmentation utilizing Deep Neural Network (ED-BT- DNN) respectively.

Traditional disease classification is slow and lab-dependent. Machine learning aids faster image-based detection but faces challenges like lighting variations, complex leaf shapes, background noise, and limited labeled data. This research develops a robust image-based method to automatically classify corn, rice, and wheat leaf diseases under diverse environmental and imaging conditions. A Hybrid Morlet Wavelet Interactive Attention Neural Network optimized by the red-billed blue magpie optimizer (HMWIANN-RBBMO) is proposed in this study for accurate classification corn (Zea mays), rice (Oryza sativa), and wheat (Triticum aestivum) leaf diseases. First, a modified square-root SageHusa adaptive Kalman filter is used to remove noise and improve image quality by image preprocessing. The DeepLabV3+ is used to accurately segment disease-prone areas, and then the Sharpbelly Fish Optimization is used to identify the most discriminative features in the images. The HMWIANN will combine Morlet wavelet transformation with interactive attention to exhaust the capabilities of the classifier to recognize Healthy (No pathogen), Common Rust (Puccinia sorghi), Blight (Xanthomonas oryzae), Gray Leaf Spot (Cercospora zeae-maydis), BrownSpot (Bipolaris oryzae), Hispa (Dicladispa armigera), LeafBlast (Magnaporthe oryzae), Stripe rust (Puccinia striiformis), and septoria (Zymoseptoria tritici). Furthermore, the RBBMO will be used to improve convergence speed, generalization, and classification accuracy. A graph-based hybrid recommendation system is also incorporated to assist disease management decisions. Experimental evaluation on corn, rice, and wheat leaf disease dataset demonstrates superior performance, achieving 99.70% accuracy, 99.80% precision, 99.50% recall, 99.40% F1-score, and a low false positive rate of 0.8%, outperforming existing state-of-the-art methods.

This research discusses how blockchain technology can be utilized to secure EEG medical reports. Because EEG data is very personal health information, efforts should be taken to ensure it does not fall into the wrong hands and that tampering is not done. Our model achieves this through the integration of a number of sophisticated security techniques. The reports are kept on IPFS, a peer-to-peer storage solution, while hashing is employed on the blockchain to ensure that data has not been altered. A time-lock component does not allow the reports to be accessed except for predefined periods, thereby providing more control over who can access the data and when. Through the examination of studies on secure sharing of healthcare information and privacy-oriented blockchain techniques, this research demonstrates how combining these methods provides an assured and secure means of dealing with EEG reports.

With the widespread development of digital healthcare systems, the protection of medical file integrity has become a major issue. Research on cryptographic hashing techniques such as BLAKE3 and SHA-3 for fast and secure file authentication, pydicom based metadata verification for ensuring the consistency of DICOM data, blockchain-based approaches for tamper-proof EHRs, federated learning for privacy-preserving data integrity verification, AI and machine learning-based anomaly detection, and lightweight cryptographic models for securing medical files. The survey shows the benefits of integrating these techniques to create a robust and scalable system for medical file integrity verification, including enhanced security, compliance, and patient trust in digital healthcare systems.

Introduction Early-stage Brain tumor detection is critical for timely diagnosis and effective treatment. We propose a hybrid deep learning method, Convolutional Neural Network (CNN) integrated with YOLO (You Only Look once) and SAM (Segment Anything Model) for diagnosing tumors.Methods A novel hybrid deep learning framework combining a CNN with YOLOv11 for real-time object detection and the SAM for precise segmentation. Enhancing the CNN backbone with deeper convolutional layers to enable robust feature extraction, while YOLOv11 localizes tumor regions, SAM is used to refine the tumor boundaries through detailed mask generation.Results A dataset of 896 MRI brain images is used for training, testing, and validating the model, including images of both tumors and healthy brains. Additionally, CNN-based YOLO+SAM methods were utilized successfully to segment and diagnose brain tumors.Discussion Our suggested model achieves good performance of Precision as 94.2%, Recall as 95.6% and mAP50(B) score as 96.5% demonstrating and highlighting the effectiveness of the proposed approach for early-stage brain tumor diagnosisConclusion The validation is demonstrated through a comprehensive ablation study. The robustness of the system makes it more suitable for clinical deployment.

As medical imaging technology has advanced, the effective privacy, integrity and security of MRI images has emerged as a daunting challenge, largely due to the ongoing rise of cyber threats against healthcare systems. As medical data traverse's networks and is stored on different systems, it is susceptible to unauthorized access, data tampering and breaches. This work suggests an all-encompassing, multi-layer security framework using steganography and three levels of cryptographic encryption AES-128, ASCON, and ECC to safeguard sensitive medical data such as MRIs from these threats. The proposed system will first use Least Significant Bit (LSB) steganography to embed the MRI image into a cover image so that the medical content is hidden in plain view. The cover image or stego image is then encrypted with three levels of encryption; AES-128 takes advantage of fast symmetric encryption that is one of the most secure encryption methods, ASCON offers lightweight authenticated encryption albeit a secure authenticated encryption, ECC provides the secure encryption keys in an asymmetric method. In order to have the data hidden and properly encrypted, the user (physician/patient) must authenticate themselves with the hospital credentials before the stego image is decrypted and patient MRI image displayed. Also, a final verification step on the user side will take place to identify additional objects to strengthen identity confirmation. After you authenticate, the data is backtracked through all three levels of decryption, and you should have the original MRI image with no loss of quality. This is a general strategy to enhance patient privacy over medical records and creates real security control for healthcare data management systems. We significantly reduce check point data leak potential any time (even while at rest). We meet data privacy standards and could provide a strong and scalable methodology to secure the sending and storing of medical images.

Wireless Sensor Networks play a critical role in modern Internet of Things applications, supporting real-time monitoring in domains such as healthcare, environmental sensing, agriculture, and smart cities. However, Wireless Sensor Networks are inherently constrained by limited energy, computational power, and memory, making them highly susceptible to issues like inefficient data transmission, premature node failure, and data security breaches. Existing solutions often fail to balance optimization and security, especially in large-scale Internet of Things environments. To address these challenges, a novel framework integrating a Modified Paillier Additive Homomorphic Encryption scheme with a Levy Gaussian Coati Optimization algorithm is proposed. The Levy Gaussian Coati Optimization algorithm enhances cluster head selection through a hybrid metaheuristic combining the Coati Optimization Algorithm for efficient Cluster Head selection, the Gaussian Barebone mechanism for refined local search precision, and the Levy Flight strategy for improved global search capability. Simultaneously, the Modified Paillier Additive Homomorphic Encryption mechanism enables secure data aggregation and computation on encrypted data without decryption, reducing the risk of data leakage. Additionally, a lightweight data compression technique is introduced to minimize communication overhead and prolong network lifetime in Wireless Sensor Networks. Experimental results demonstrate that the proposed model achieves a 48

Dental caries is a chronic condition that affects most people during their lifetime. Dentists generally detect caries lesions based only on visual evaluation with panoramic radiographs. However, manual interpretation is labor-intensive, subjective, and dependent on clinician expertise and in many cases, dental caries can be misinterpreted as shadows due to the low image quality. Artificial intelligence (AI) may facilitate the accurate interpretation of radiographs. However, traditional models suffer from data privacy issues when using medical data to train these AI models. To overcome these issues, a novel Federated learning for Automated Detection of Dental pathologies (Fed-ADD) model has been introduced in this paper. The proposed model integrates a Multi-branch dual stream ResNet101-Swin-VIT based pathology detection and a federated transfer learning strategy for efficient detection and secure data sharing. Furthermore, the proposed model integrates a hybrid explainability model for confidence-aware justification suitable for clinical auditing. The efficacy of the model has been evaluated using specific measures such as Accuracy (AC), Precision (PR), Recall (RE), Dice and IoU. The proposed FedADD model achieves the highest accuracy of 95.2%, whereas the existing techniques such as Faster R-CNN, CNN based DL ensemble model and APD-FFNet achieve accuracy of 94.18%, 89.45%, and 94% respectively.

Significant challenges are raised in diagnosing neurodegenerative diseases like Alzheimer's and Parkinson's disease due to their varied symptoms and biomarkers. Multimodal imaging techniques, such as ultrasound imaging, Magnetic Resonance Imaging (MRI), and Positron Emission Tomography (PET), are effective for understanding neurophysiological changes; however, integrating these various imaging modalities is challenging in real-world applications. This paper presents a novel Dense Fusion Framework (DFF) for multimodal image fusion with higher identification accuracy of the disease in the early stage. The proposed method was evaluated using the benchmark ADNI dataset. This novel DFF framework combines deep learning-based feature extraction with a fusion technique that aims to maximize classification performance. The performance metrics analyzed after applying the DFF framework have shown improved performance and achieved 94.6% while comparing the conventional methods. With respect to the fusion metrics the results significantly shows the raise in structural preservation of the imaging data.

Climate change, resource scarcity, and the ever-increasing need for food are three of modern agriculture's most pressing problems. Using historical data, the suggested approach accurately predicts soil moisture using a Seasonal Autoregressive Integrated Moving Average with exogenous variables (SARIMAX) model. This model accounts for both seasonal trends and external factors, making the forecasts more credible. At the same time, a model called MSVM-DAG is used to come up with crop suggestions using all the weather and soil data possible. To further optimize the system and ensure robust model performance, the Grey Wolf Optimizer (GWO) method is used for hyperparameter tweaking. A variety of measures are used to assess the system's performance. For the soil moisture model, these include MSE, MAE, RMSE, R-squared, and MAPE. For the crop recommendation model, they include accuracy, precision, recall, F1-score, confusion matrix, ROC curve, and AUC. Cross-validation is also integrated. By leveraging real-time and historical data, this hybrid system aims to improve irrigation scheduling, water management, and crop selection, thereby optimizing resource allocation and maximizing yields.

Herein, rGO–MnO2 nanocomposite was synthesized via a hydrothermal method. The prepared rGO–MnO2 powders underwent various characterizations to examine their structural, morphological, and electrical properties. XRD was used for preliminary identification of the synthesized powders, while Raman spectroscopy confirmed bond types and functional groups. UV–visible spectroscopy provided key insights into the electronic and optical properties through linear optical studies. SEM and TEM were employed to evaluate the catalyst’s morphological structure. Electrochemical sensing properties were analyzed using cyclic voltammetry, revealing that rGO–MnO2 exhibits good sensitivity toward histamine, making it a promising candidate for real-time histamine sensing in the human body. Additionally, the rGO–MnO2 nanocomposite demonstrated remarkable photocatalytic activity, achieving a 91.7

Excellence in transmission can be assessed in optical transport networks before providing any additional connections or upgrading the connections. Generally, the Physical Layer Model (PLM) is used to assess the transmission quality which has high probability in uncertainty and inaccuracy due to the circumstances of physical layer. The network efficiency is directly proportional to the margins. If the margins getting increases in the PLM, the efficiency of the network decreases. Maintaining the excellence in transmission is the biggest challenge when the margins getting increased. Other significant factors for excellence in transmission is scalable, minimum latency with maximum speed and energy efficient. Photonic switching is a hopeful solution for handling these challenges. Machine learning technique is proposed to assess the excellence of transmission and flow detection. ML-E and Precedence based scheduling algorithms are proposed for excellence of transmission and flow detection respectively. The proposed techniques justify variations, uncertainties in kits like fiber dilution, dispersion and optimizes PSON (packet switched optical network). Simulation results are demonstrated and the proposed work results indicates that it can outperform a benchmark in all aspects.

The increasing prevalence of brain degenerative disorders particularly Alzheimer's and Parkinson's, have incurred a dire need for effective diagnostic methodologies. With advancements in medical imaging techniques and Artificial Intelligence (AI), transformative potential exists to enhance early detection and disease classification. This review examines the recent progress made in the use of diagnostic medical imaging and artificial intelligence (AI) techniques for the early detection of Alzheimer's disease and Parkinson's disease. Despite significant progress, a comprehensive overview is lacking. The review aims to summarize and synthesize existing methods, critically evaluate their strengths and limitations, and identify emerging trends. Multimodal imaging data, combining information from different sources like MRI, PET, and SPECT, has shown progress in improving diagnostic accuracy and providing insights into disease progression. Advanced AI techniques, such as transfer learning and generative adversarial networks, have further enhanced the performance of diagnostic models. Previous studies have predominantly focused on using standalone techniques, often leaving the integration of multi-modal imaging and deep learning methodologies that could lead to more robust diagnostic frameworks. This literature review addresses this gap by synthesizing current research on medical imaging modalities, AI applications, and their implications in diagnosing Alzheimer's and Parkinson's diseases. This review seeks to connect technological innovations with clinical applications to establish more robust and dependable diagnostic methods.

The rapid advancement of medical technology has brought about the impending arrival of the era of big data in the medical field. As a direct and immediate consequence of the analysis and mining of these data, there will be significant effects on the prediction, monitoring, diagnosis, and treatment of tumor-related issues. Due to the fact that it possesses a diverse collection of traits, a low survival rate, and an aggressive nature, brain tumours are widely regarded as the disease that is both the most lethal and the one that causes the greatest disability. It has been proposed that deep learning has the ability to solve the difficulties associated with diagnosing and treating brain cancers. Through the use of FL on brain tumor identification from MRI images, the purpose of this research is to find a solution to the problem of centralized data collecting. Using the Visual Geometry Group (VGG 16) as a tool for detecting brain cancers, implementing a convolutional neural network (CNN) model framework, and determining the parameters to train the model for this problem were the objectives of this research. The MR images might be analyzed with our method to spot brain cancers. The testing results showed that the algorithm worked better than the standard methods that are currently used for detecting brain tumors, with an excellent accuracy of 92%.