Exponential Chicken Swarm Optimization to CNN for Efficient MRI Brain Tumor Classification: A Hybrid approach

In radiology, magnetic resonance imaging (MRI) is typically used to diagnose the body’s morphology and physiological functions. These days, an important part of the MRI brain imaging process is the tumor region classification. Due to variety and complexity of cancers, the categorization and segmentation of tumour are demanding in MRI brain imaging. This research proposes an automatic brain tumor classification method using a Convolutional Neural Network (CNN) based on the Exponential Chicken Swarm Optimization (Exp-CSO) algorithm. Tumor regions are first segmented using fuzzy c-means clustering, and features are extracted to form a feature vector. The Exp-CSO optimizes the CNN classifier’s cluster center, enhancing classification accuracy. The model achieves higher accuracy of 99% and a minimal Mean Square Error (MSE) of 0.0401, outperforming existing methods in classifying MRI brain images, as confirmed by evaluation metrics.