Brain tumors are one of the major causes of death in the world. Early diagnosis of the tumor can be a great help to a person's successful treatment and health. Diagnosis by physicians and radiologists is time-consuming compared to automatic diagnosis. In this paper, a new method for tumor detection in brain MRI images is proposed based on convolutional neural networks. This approach classifies brain MRI images into two classes: with tumor and without tumor. In recent years Convolution Neural Networks have had better results compared to other machine learning techniques in the image classification. To solve the problem of data deficits in the field of medical image analysis, an augmentation method was used. The datasets used to train the proposed network are IXI, BRATS2017 and Cheng dataset. This method classifiesthe MRI images into two categories, performed with a desirable accuracy of 99.63%.
