Multi-grade Brain Tumor Classification Using a Modified Convolutional Neural Network

Brain tumors are one of the deadliest types of cancer, and their severity has made them the leading cause of cancer-related death. The multi-grade classification of brain tumors is essential for the radiologist to provide a better analysis of magnetic resonance imaging (MRI). This paper proposes a modified convolutional neural network (MCNN) for the multi-grade classification of brain tumors. The novelty of this work is classifying brain tumors into multiple graded using an MCNN model inspired by the VGG19 architecture. In the first step, input images are preprocessed using normalization and data augmentation techniques. The next step in the process is to obtain attributes using the MCNN model. Finally, these attributes are used by the fully connected layers tailored for the classification task at hand. The dataset utilized for experimentation contains four types of brain MRI samples: meningioma, glioma, pituitary, and normal brain. The proposed method is compared to a pre-trained VGG19 model. As a result, the proposed method outperforms the existing pre-trained deep-learning techniques.