Semiautomatic and Automatic Brain Tumor Segmentation Methods: Performance Comparison

Brain tumor segmentation plays a vital role in surgical, treatment planning and follow-up studies. The noninvasive nature, better soft tissue contrast, and no ionizing radiation makes the magnetic resonance imaging (MRI) a very useful medical imaging modality in visualizing the brain lesions. However, huge amount of data produced by MRI scanners makes it difficult for the radiologist to manually segment the tumor region. It becomes a tedious and time-consuming task for them. Hence, there is essential to develop consistent brain tumor segmentation algorithm. This work focuses on identifying the most accurate brain tumor segmentation method whose performance is very close to that of an expert radiologist. Various brain tumor segmentation methods that fall under semi-automatic and automatic category are evaluated both qualitatively and quantitatively using state-of the art image segmentation metrics.