Mining anatomical, physiological and pathological information from medical images

Abstract

The field of medical imaging has shown substantial growth over the last decade. Even more dramatic increase was observed in the use of machine learning and data mining techniques within this field. In this paper, we discuss three aspects related to information mining in the domain of medical imaging: the target user groups (for whom), the information to mine (what), and technologies to enable mining (how). Specifically, we focus on three types of information: anatomical, physiological and pathological, and present use cases for each one of them. Furthermore, we introduce representative methods and algorithms that are effective for solving these problems. We conclude the paper by discussing some major trends in the related domains for the coming decade.