Abstract
Early detection of a brain tumor increases life expectancy and survival chances of a patient. Experts use modalities like magnetic resonance imaging (MRI) and computerized tomography (CT) to locate brain tumors in images. Medical professionals carefully study patterns of the soft tissues, viz. gray matter, white matter, and cerebrospinal fluid within brain images to trace possible abnormality. A manual image analysis task is time consuming, nonreproducible and highly dependent on the individual's skill. On the other hand, computer-assisted analysis of a medical image helps experts in quick decision making, generates reproducible results and electronic patient record, improves diagnosis, and helps in treatment planning. This chapter covers state-of-the-art review for automated brain tumor segmentation and focuses on supervised form of learning. Initially, the chapter covers conventional methods but later shifts focal point to uncover deep neural network for brain tumor segmentation. Deep neural networks have an excellent capability of automatic feature discovery and they also fight against curse of the dimensionality. This chapter covers brain tumor segmentation using MRI images. Any one of the four MRI modalities, namely, T1, T2, T1c, and FLAIR image, is given as an input to a method, which segments out the tumor. The approaches for brain tumor segmentation are analyzed and their comparative study is presented on the publicly available dataset. The chapter also presents various open source tools for brain tumor segmentation and quality metrics to quantify result. Overall purpose of this chapter is to provide comprehensive picture to a reader about the learning approaches in brain tumor segmentation, available tools, and the quality metrics for segmentation.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call