The Role of Artificial Intelligence in Neuro-oncology Imaging

Abstract

AbstractDiagnostic imaging is widely used to assess, characterize, and monitor brain tumors. However, there remain several challenges in each of these categories due to the heterogeneous nature of these tumors. This may include variations in tumor biology that relate to variable degrees of cellular proliferation, invasion, and necrosis that in turn have different imaging manifestations. These variations have created challenges for tumor assessment, including segmentation, surveillance, and molecular characterizations. Although several rule-based approaches have been implemented that relates to tumor size and appearance, these methods inherently distill the rich amount of tumor imaging data into a limited number of variables. Approaches in artificial intelligence, machine learning, and deep learning have been increasingly leveraged to computer vision tasks, including tumor imaging, given their effectiveness for solving image-based challenges. This objective of this chapter is to summarize some of these advances in the field of tumor imaging.