Abstract: Stroke remains a leading cause of death and disability globally, demanding prompt diagnosis and intervention to enhance recovery outcomes. Leveraging recent advancements in machine learning, this study presents an early stroke detection framework utilizing neuroimage analysis, particularly brain CT scans. A Residual Network (ResNet) model is employed to improve classification performance by extracting critical features from CT images. Cross-validation techniques evaluate the model’s accuracy using precision, recall, F1 score, and ROC-AUC metrics. The proposed system empowers healthcare professionals with a reliable, automated tool for earlier and more accurate stroke detection, potentially reducing patient morbidity and mortality rates.

Keywords: Stroke Detection, Neuroimaging, Machine Learning, Residual Networks (ResNet), Early Diagnosis.

| DOI: 10.17148/IARJSET.2025.12471