Brain tumor is an abnormal cell population that occurs in the brain. Nowadays, medical imaging techniques play an important role in tumor diagnosis. Magnetic resonance imaging (MRI) is a medical imaging technique that uses a magnetic field and computer-generated radio waves to output detailed images of the organs and tissues in your body. In this study, three different threshold segmentation-based approaches have been reviewed and compared to extract the tumor from a set of MRI brain images. These methods are seeded region growing, k-means, and global thresholding. The images used in this study are obtained from Cancer Imaging Archive (TCIA) and kaggle. All images are grayscale and in JPEG format. The images from TCIA dataset are 100 images which contain abnormal (with a tumor) brain MRI images while there are 35 images in kaggle dataset. The kaggle dataset contains 20 normal images and 15 abnormal images. The results show that the k-means segmentation algorithm performed better than the others on TCIA dataset according to the Root Mean Square Error (RMSE), the Peak to Signal Noise Ration (PSNR), and Segmentation Accuracy while global thresholding is the best on kaggle dataset.