Segmentation of Brain Stroke MRI Images by Bio Inspired Genetic Swarm Clonal Algorithm and Multi-Classification of Stroke using Deep Adversarial Naïve Segnet Architecture
stroke detection, brain image segmentation, multi-classification, MRI brain image, bioinspired deep learning
AuthorsAbstractHuman stroke incidence has been progressively rising recently for a variety of reasons. The patient's management depends critically on the type of brain stroke that has been diagnosed. Although magnetic resonance imaging (MRI) is a commonly utilized diagnostic method for stroke, manual interpretation of MRI results by professionals is laborious and time-consuming. For automatic segmentation as well as categorisation of stroke on brain MRI, computer-aided diagnostic (CAD) models are therefore required. This research proposes novel technique in stroke detection based on brain image segmentation with multi-classification in stroke detection using MRI brain image by bio-inspired deep learning model. Here the input is collected as MRI brain image in which the images have been pre-processed for noise removal and normalization. Then the processed image is segmented using conventional Boltzmann transfer convolution with genetic swarm clonal algorithm (CBTC-GSwClA). The segmented image shows the brain MRI regions with stroke prediction and the image will be classified using deep adversarial naïve SegNet architecture (DANSegNet). This classified output shows various classes of stroke ranging from stage 0 to 6. The experimental analysis has been carried out for various stages of stroke in terms of Training accuracy, Validation accuracy, Recall, Precision, F1-score, Kappa coefficient, ROC. Proposed technique achieved 96% Precision, 99% Validation accuracy, 98% Training accuracy, and a 97% F1-score, recall of 98%, kappa co-efficient of 94%.
•••••••••••••••••••••••••••••••• Ejprd.org - Published by Riset Publication Services LLC
EJPRD
Copyright ©2026 by Riset Publication Services LLC