An Intelligent Framework for Early Detection of Brain Strokes Using VGG16 and SVM

Brain strokes, encompassing ischemic and hemorrhagic varieties, are predominant contributors to disability and mortality globally. Prompt identification and immediate intervention are essential for reducing long-term consequences. This research presents an intelligent framework for the early diagnosis of cerebral strokes, integrating VGG16 for extracting features with Support Vector Machines (SVM) for categorization. The framework employs pre-trained VGG16 to autonomously extract pertinent features from medical images, including CT and MRI scans, therefore obviating the necessity for human feature engineering. The SVM classifier processes these features, trained to distinguish between stroke and non-stroke instances. The model, assessed on a dataset of annotated brain scans, attains an accuracy of 96.50 %. The findings indicate significant sensitivity and specificity, establishing the suggested framework as a dependable, scalable approach for early stroke diagnosis in clinical environments, potentially enhancing patient outcomes and facilitating prompt medical decisions.