Improving Brain Cancer Detection and Classification Using Xception Deep Convolutional Neural Network

The detection and classification of brain cancer needed to do digitized after the rapid growth in the incidence of these kinds of diseases. Various brain tumor classifications are an ongoing research topic made possible by the diversity of cancer features. Local brain images testing in the latest brain images datasets BRAT can be challenging due to their processing time, lower accuracy, and the possibility of overfitting and underfitting. In this paper, Xception Deep Convolution Neural Network (DCNN) model was implemented to test the local images or real images and apply three datasets in this model, analysis the individual result and improve the training speed and stability of neural networks by activations of each layer. The Xception model can significantly reduce the number of parameters and computational complexity without compromising the model’s accuracy. Moreover, this system used popular benchmark datasets such as Kaggle and BRATS, the suggested research Magnetic resonance imaging (MRI) image datasets were assessed, and the local validation dataset was proposed. In this research work, this model was achieved the best performance decisions with training accuracy and loss 99±0.005%, validation accuracy and loss 98±0.2%. The main objectives are improving the accuracy of classification, avoiding overestimation along with underestimating, and decreasing down on processing time.