Enhanced Vgg-19 Model for Accurate Brain Tumor Prediction

One of the most serious diseases for both adults and children is a brain tumor. Through the scans, a substantial amount of image data is produced. The radiologist reviews these images to detect the abnormality. Due to the complexity of brain tumors and their features, a manual examination may be mistake-prone. To visualize the brain and spot any abnormal growths, many imaging techniques are used. Magnetic resonance imaging (MRI) is the method that works best to identify brain tumors. Predicting brain tumors from MRI images also contributes to medical research and the development of more sophisticated predictive models. This continuous improvement in technology can lead to better diagnostic tools and more effective treatment options for brain tumor patients. The existing models exhibit significant shortcomings concerning explanation, interpretability, computational complexity, and accuracy. In order to overcome the drawbacks available in the existing techniques, this research work proposes the Enhanced VGG-19 model by improving the convolutional neural network layers, fully connected layers, and weights of the layers and applying effective preprocessing techniques to the datasets to predict brain tumors from MRI images with high accuracy and less computation time. The proposed model produced an accuracy of 99.72% and performed better than the existing techniques.