Deep and statistical learning in biomedical imaging: State of the art in 3D MRI brain tumor segmentation

Clinical diagnostic and treatment decisions rely upon the integration of patient-specific data with clinical reasoning. Cancer presents a unique context that influence decisions, given its diverse forms disease evolution. Biomedical imaging allows noninvasive assessment based on visual evaluations leading to better outcome prediction therapeutic planning. Early methods brain cancer characterization predominantly relied statistical modeling neuroimaging data. Driven by breakthroughs in computer vision, deep learning became de facto standard domain medical imaging. Integrated have recently emerged as new direction automation practice unifying multi-disciplinary knowledge medicine, statistics, artificial intelligence. In this study, we critically review major models their applications research focus MRI-based tumor segmentation. The results do highlight model-driven classical statistics data-driven is potent combination for developing automated systems oncology.