EBrainNet: An Optimized CNN with Modified Satin Bowerbird Optimization for Accurate Brain Stroke Classification from CT Images

The human brain is a complex and vital organ that controls the nervous system. it plays a crucial role in the way we experience and interact with the world. To enhance the accuracy and performance of brain stroke diagnosis that is critical for timely treatment the Enhanced Brain Stroke Net (EBrainNet) is suggested to classify brain strokes from CT (Computerized Tomography) images. The initial step in the process is gathering the dataset of brain stroke CT images. The images are then cleaned by applying a Gaussian filter to remove noise. One efficient method of finding and retrieving areas of interest in images is image segmentation. From a pre-processed image, it enables the analysis and processing of relevant data a key part of efficient diagnosis and treatment planning. Modified Satin Bowerbird Optimization (MSBO), an algorithm employed to select optimal hyperparameters for boosting classification accuracy and reducing computational complexity, is among the numerous algorithms merged to develop EBrainNet, an enhanced version of the convolutional neural network (CNN) designed specifically for stroke classification. This proposes a new approach to the classification of brain strokes from CT scans, employing EBrainNet improved with MSBO as a potentially valuable resource for medical professionals. The proposed classifier achieves 98.2% accuracy in testing and 99.4% accuracy in training.