Abstract:
Brain tumour is a serious malignancy that can lead to death. Early diagnosis is therefore essential in the therapy procedure. Deep learning advances have made a significant contribution to medical diagnostics in the healthcare business. CNNs have been widely employed as a deep learning strategy for detecting brain cancers using MRI images. Deep learning techniques like CNNs should be upgraded to be more efficient because of the restricted dataset. As a result, Data Augmentation is one of the most well-known methods for improving model performance. This article details the implementation of multiple VGG-19 architectures as a foundation layer for specific models. Pre-processing, cropping, augmentation, VGG-19 as a base layer with transfer learning-based brain tumour binary classification and extra layers of normalisation, dense, and activation layers are all part of the proposed system. On brain tumour kaggle MRI datasets, the suggested technique obtained Cohen Kappa Score, f1-score, recall, accuracy, Precision, and ROC AUC score are
.9900, .9949, .9950, .9950, .9950 and 1.000 respectively. The experiments demonstrated that the proposed methodology is efficient and effective, and that it outperformed comparable recent research in the literature on kaggle MRI datasets.
