Brain Tumor Segmentation Methods based on MRI images: Review Paper

Abstract

Statistically, incidence rate of brain tumors for women is 26.55 per 100,000 and this rate for men is 22.37 per 100,000 on average. The most dangerous occurring type of these tumors are known as Gliomas. The form of cancerous tumors so-called Glioblastomas are so aggressive that patients between ages 40 to 64 have only a 5.3% chance with a 5-year survival rate. In addition, it mostly depends on treatment course procedures since 331 to 529 is median survival time that shows how this class is commonly severe form of brain cancer. Unfortunately, a mean expenditure of glioblastoma costs 100,000$. Due to high mortality rates, gliomas and glioblastomas should be determined and diagnosed accurately to follow early stages of those cases. However, a method which is suitable to diagnose a course of treatment and screen deterministic features including location, spread and volume is multimodality magnetic resonance imaging for gliomas. The tumor segmentation process is determined through the ability to advance in computer vision. More precisely, CNN (convolutional neural networks) demonstrates stable and effective outcomes similar to other automated methods in terms of tumor segmentation algorithms. However, I will present all methods separately to specify effectiveness and accuracy of segmentation of tumor. Also, most commonly known techniques based on GANs (generative adversarial networks) have an advantage in some domains to analyze nature of manual segmentations..