The most severe form of cancer sickness is brain tumor. It arises from uncontrollable and strange cell division. Brain tumors can be classified into benign and malignant tumors. The recognition of brain tumors is a complex mission that implied the experience of the classifier. The manual classification of tumor types using data gathered from MRIs is believed to be an exhausting task that may result in human error and false tumor type detection.
In this paper, we compared ML and DL different algorithms for brain tumor classifica-tion such as VGG-16, CNNs, SVM, and KNN to categorize four types of brain tumors (meningioma tumor(originate in the meninges), glioma tumor( improve from different types of glial cells), pituitary tumor (non-threatening tumor), and no tumor).DL achieved high results with accuracy 99% for CNN and 90% for VGG16 (not just accu-racy was used for estimating these models, other evaluation metrics will be calculated as discussed later ) , while ML didn't achieve suitable results for brain tumor classifica-tion ,SVM achieved 91% accuracy .This experimental study was implemented on a real time dataset with different tumor sizes, locations, shapes, and different image intensities.