Classifying neurological diseases from magnetic resonance imaging (MRI) images remains a challenging study using in artificial intelligence-assisted medical diagnosis. An automated multi-class neurological disease classification system is presented to address inter-class visual similarity, multi-planar acquisition variability, and severe class imbalance. The framework is built on Vision Transformer (ViT)-Base/16 for eleven-class neurological MRI classification covering glioma, meningioma, pituitary adenoma, multiple sclerosis (MS; axial and sagittal planes), stroke subtypes, and healthy controls, employing a progressive curriculum training strategy. Training proceeded through nine stages that incrementally scaled per-class samples from 100 to 900, with class-weighted cross-entropy loss applied throughout to mitigate majority-class dominance. Testing was conducted on a set of 1468 MRI samples, achieving 97.07
Biofumigation using cruciferous crop residues provides an eco-friendly and sustainable method for managing Meloidogyne incognita and improving soil health in tomato cultivation. This study evaluated the comparative efficacy of mustard, cabbage, cauliflower, and radish residues applied at 0.5
Wireless sensor networks (WSNs) transmit the data to the base station through specialized sensors to track environmental parameters, including humidity, temperature, and pollution, which are widely used in the detection of forest fires and also in some agriculture fields. WSNs face some issues, like recognizing faulty nodes and the battery life of the node, which affects the reliability and lifespan of the nodes. In this paper, we employed a combination of Federated learning (FL) and energy-aware clustering like LEACH with sleep scheduling to detect the fault node efficiently in wireless sensor networks. LEACH optimizes energy utilization by clustering the nodes and lowering the communication overhead. FL ensures decentralized model training while preserving privacy and energy.
The steno-endemic species Impatiens aquatica, previously known from the Eastern Ghats, is here reported as a range extension to the Nilgiris in the Western Ghats of Tamil Nadu. A detailed description, colour photographs, ecological notes, and conservation status are provided for this potential wild ornamental species.
Power system engineers continually face challenges in coordinating primary-backup relay pairs to protect distributed generation (DGs) in penetrated electric power systems. Accurate directional overcurrent relays (DOCRs) settings provide fault clearance in the minimum operating time, thereby improving system reliability. In this paper, GA, BGA, and Rao 3 algorithms are implemented to optimize DOCRs settings in a DGs-penetrated power system.Under defined constraints, the primary objective is to minimize the operating times of primary relays and optimize coordination between primary-backup (P/B) relay pairs. The penalty method has been used to incorporate these constraints into the fitness function. The employed algorithms are framed to tackle the nonlinear DOCRs coordination optimization problem. The effectiveness of all employed GA, BGA, and Rao-3 optimization methods is evaluated on the IEEE 14-bus test system penetrated with DGs, which is modeled and simulated in ETAP software. The coordination of DOCRs from an optimal perspective is achieved using MATLAB-based optimization algorithms. Nonetheless, Rao-3, a simple, metaphor-free, algorithm-specific, parameter-free optimization algorithm, is an extremely effective and robust solution to the complex coordination problem of DOCRs in DGs-penetrated power systems, outperforming existing genetic and breeder genetic algorithms.
