Aims: Cancerous brain tumors are among the most dangerous diseases that lower the quality of life of people for many years. Their detection in the early stages paves the way for the proper treatment. The present study aimed to present a two-dimensional Convolutional Neural Network (CNN) for detecting brain tumors under Magnetic Resonance Imaging (MRI) using the deep learning method.
Methods & Materials: The proposed method has two stages of feature extraction and classification. A 12-layer CNN was used to extract the features of the MRI images and then the softmax activation function was used to classify these features. The proposed method was applied to a standard database consisting of three brain tumor types of meningioma, glioma, and pituitary.
Findings: The proposed method had better performance compared to previously presented methods. Its accuracy was reported as 98.68%.
Conclusion: Meningioma, glioma, and pituitary tumors are the most common types of brain tumors. Early detection of these tumors can decrease the risk of death. Because of its fully connected structure, the use of proposed deep CNN can help physicians to correctly detect brain tumors with MRI images.

Keywords: Brain tumor, MIR images, Deep learning, Two dimensional convolutional network

Full-Text [PDF 5687 kb]   (3186 Downloads) |   |   Full-Text (HTML)  (17356 Views)