Medical Text and Image Processing

Abstract

Image and text data have a vast volume in electronic medical records (EMR) of healthcare systems. Their processing leads to interesting applications related to the diagnosis of different diseases, suggesting optimal treatment plans in terms of costs and efficiencies for (1) disease challenges like its risk or evolution prediction, patient or disease clustering, and population and care patterns; (2) drug development and reaction prediction; (3) supporting hospital operations like enhancing management efficiency and decision support for physicians and managers, and quality improvement; (4) mobile health and personalized healthcare using network treatment; and (5) social support, gathering online questions, and quality assessment based on social media. Unstructured data, like image and text processing, is challenging due to many factors, including noises in medical data, variability in medical imaging devices, a complex structure of human organs anatomy, ungrammatical structure of clinical notes, abbreviations, and many others. Interestingly, both image and text processing have a similar methodology in many cases like classification and prediction, including pre-processing, feature extraction, model generation, and learning, and using the model phases. This chapter describes different applications of medical text/image processing. The main issues and related challenges to these applications are highlighted and some general methodologies and tools of EMR processing will be described.