What question did this study set out to answer?

This research focuses on developing a novel CNN model for effective brain tumor classification from MRI scans.

February 14, 2026

Novel CNN-driven model for brain tumor image categorization

Key Points

This research focuses on developing a novel CNN model for effective brain tumor classification from MRI scans.
Developed a CNN model incorporating multi-scale convolutional blocks for feature extraction.
Employed various optimization algorithms for hyper-parameter tuning and feature selection.
Used performance metrics including accuracy, sensitivity, and specificity to evaluate outcomes.
Achieved 99.2% accuracy, 99.5% sensitivity, and 98.72% specificity in brain tumor classification.
Outperformed traditional pre-trained CNN models in classification tasks.

Abstract

This study presents a new, efficient Convolutional Neural Network (CNN) design for classifying brain tumors through MRI scans. The approach combines multi-scale convolutional blocks with different kernel sizes (1 × 1, 3 × 3, and 5 × 5) to efficiently capture a range of intricate features from medical images. To improve performance, the framework integrates sophisticated optimization algorithms, such as Ant Lion Optimizer (ALO), Spider Monkey Optimization (SMO), Whale Optimization (WO), Shuffled Frog-Leaping Algorithm (SFLA), Gray Wolf Optimization (GWO), and Particle Swarm Optimization (PSO), for feature selection and hyper-parameter tuning. These algorithms serve two purposes: (1) they simplify the feature space by assessing and prioritizing features according to their significance, removing unnecessary or unimportant features, and (2) they adjust essential hyper-parameters like learning rate, batch size, and number of epochs to promote quicker convergence and enhance model performance. The suggested CNN structure includes an input layer, convolutional blocks featuring batch normalization and activation functions, max-pooling layers to reduce dimensionality, fully connected layers for advanced reasoning, and an output layer dedicated to tumor classification. The integration of optimization techniques ensures the selection of an optimal feature set, significantly improving sensitivity and classification accuracy. Experimental results demonstrate that the proposed method achieves an accuracy of 99.2%, specificity of 98.72%, sensitivity of 99.5%, precision of 98.98%, and f score of 99.24% outperforming traditional pre-trained CNN models and competing with state-of-the-art techniques. The framework is computationally efficient, making it suitable for deployment in resource-constrained environments such as medical diagnostic systems. This work highlights the potential of combining lightweight CNN architectures with advanced optimization techniques to address the challenges of brain tumor classification in medical imaging.

Bookmark

Cite This Study

Abdellatef et al. (Sun,) studied this question.

synapsesocial.com/papers/698fd276306598e8538deaaa https://doi.org/https://doi.org/10.1007/s00521-025-11770-w

Bookmark