According to the National Cancer Clinical Report, brain tumors rank in the top ten in terms of morbidity and mortality. Depending on the location of the tumor and the cause of the lesion, the categories of brain tumors can be subdivided, and the clinical treatment options for brain tumors will also be different, so it is particularly important to correctly classify brain tumors. Currently, the tumors with higher incidence rates are meningioma, glioma, and pituitary tumor. Based on the shape, size, and imaging features of these three brain tumors presented in MRI images, as well as their spatial locations in brain tissues, this chapter proposes a Classification Algorithm for Brain Tumors based on Squeeze-Excitement Module and Spatial Attention Mechanism Module, using the ResNeXt network as the backbone network. In this paper, we propose a brain tumor classification network (ResNeXt-SESA) based on a squeeze excitation module and a spatial attention mechanism module, which incorporates channel information and spatial information into the extracted tumor image features to improve the network's ability to discriminate tumors. The difficulties caused by the different disease locations and imaging features of tumors under different categories, and the diversity of tumor shapes and irregular locations of tumors under the same category on the task of tumor classification are addressed.