An Automatic Brain Tumors Detection and Classification Using Deep Convolutional Neural Network with VGG-19
Venmathi A.R, S. Sumam David, Ennam Govinda, K. Ganapriya, R. Dhanapal, A. Manikandan
Abstract
Early detection of brain tumours (BT) is essential for increasing patient survival and potential. Magnetic resonance imaging (MRI) of BTs was physically studied in this research. Therefore, algorithmic methods are required to diagnose tumours more accurately. However, determining structure, volume, limits, tumour detection, size, segmentation, and grading are still challenging. We propose a hybrid enhanced deep convolutional neural network (DCNN) classifier based on the VGG-19 classifier algorithm in this research work. The primary goal of this study was to determine the tumour's exact location and classify it as malignant or benign. We began by splitting the dataset with the enhanced DCNN classifier. Preprocessing is accomplished using a Laplacian Gaussian (LOG) filter. In general, the suggested method seeks to enhance the efficiency of classifiers using deep learning. All CNN models are analyzed using performance measures such as precision, accuracy, specificity, and recall. The proposed technique achieves greater than 99.65% accuracy. The proposed technique significantly outperformed the previous algorithms.