An Efficient Detection Model for Brain Stroke Based on Transfer Learning

Brain stroke is one of the most common causes of death, ranking as the second leading cause worldwide. With the advancements of deep learning, the detection of brain strokes from CT images becomes possible. In this study, we propose a method for classifying brain stroke images and predicting the presence of a stroke using convolutional neural networks (CNNs), which are particularly effective in image classification tasks. These techniques involve training different CNN models (EfficientNetB0-B7, InceptionResNetV2, ResNet50, ResNet152, and VGG19) on a dataset of labeled brain strokes and normal brain images. Through this process, the model learns to identify patterns and features that are indicative of a stroke. The results show that the different CNN models were able to classify stroke images with high accuracy and predict the presence of stroke with an overall accuracy of 99.95%, which is a significant improvement compared to similar studies. This demonstrates the potential of deep learning-based approaches using CNNs in assisting the diagnosis of strokes and aiding in early-stage treatment. Finally, we provide an analysis of model performance, highlighting each CNN model used, and providing insights for future research.