I. Introduction
The diagnosis, prognosis, and therapy of brain tumors all benefit from the automatic segmentation and categorization of medical images. Early detection of brain tumors indicates a speedier response to therapy, thereby increasing the patient survival rate. It requires a significant amount of time and effort to manually identify and categorize brain malignancies in large medical image collections obtained during routine clinical duties. It is desired and useful to implement an automated detection, localization, and categorization technique [1]. Figure 1 shows examples of these three types of tumors.
(a) Meningioma tumor, (b) Glioma tumor and (c) Pituitary tumor