Identification and Segmentation of Tumour in Brain MRI using Deep Learning Techniques

Medical image segmentation is an essential task to dissect the infected portion present in the raw medical images. Gliomas are the most widespread and abrasive form of brain tumour, resulting in a very short expected lifespan in their most severe form. As a matter of fact, diagnosis preparation is a vital step in enhancing the overall eminence of lifespan of cancer patients. Magnetic resonance imaging (MRI) is a popular scanning method for examining these tumours, but the large volume of information generated by MRI precludes manual segmentation in an acceptable time period, restricting the utilization of concise quantitative assessments in medical practice. Gliomas and their intra-tumoral structures must be precisely dissected not just to make clinical judgement as well as for further investigations. Moreover, it is a difficult task because the form, structure, and site of these anomalies vary significantly. As a result, fully automated and reliable dissection methods are desired; moreover, the wide structural and spatial diversity between brain tumours makes automatic delineation a tough problem. This work reviews some recent researches on glioma identification and segmentation and also discussed about the performance metrics.