The most prevalent and dangerous illness, brain tumors have a very low life expectancy in their highest degree. Consequently, a crucial step in enhancing patients' quality of life is treatment planning. Broadly speaking, tumors in the brain, lung, liver, breast, prostate, etc. are evaluated using a variety of imaging modalities, including computed tomography (CT), magnetic resonance imaging (MRI), and ultrasound images. In particular, MRI scans are employed in this work to identify brain tumors. Nevertheless, the massive amount of data produced by an MRI scan makes it impossible to manually classify a tumor as opposed to a non-tumor at a certain moment. Nevertheless, it has a restriction in that only a small number of images can get precise quantitative data. This research focuses on developing an automated brain tumor classification method using convolutional neural network in combination with self-attention mechanism. This combination helps in achieving higher accuracy in classifying MRI images as meningioma, pituitary, glioma, and no tumor.

Magnetic Resonance Imaging, brain tumor, classification, convolutional neural network, sel-attention, meningioma, pituitary, glioma.