Image analysis techniques for the quantification of brain tumors on MR Images

Abstract

Advances in neuro-imaging methods over the last few decades have enabled collection of detailed anatomical and functional information about the brain. Although functional imaging provides rich information for diagnosis and treatment planning, practical considerations such as cost and availability currently limit its clinical utility. As a result, structural imaging methods that provide detailed information about the anatomical structures of the brain are routinely used to manage brain tumors in the clinical setting. Typically, radiological images are visually inspected and interpreted by trained health professionals to detect gross anatomical abnormalities, which are associated with various types of brain tumors. This approach entails generally qualitative interpretations that do not fully realize the potential of modern imaging technologies. Furthermore, several types of brain tumors manifest with gross anatomical changes that are visually similar, which limits the use of MRI in differentiating between them. Computer-Aided image analysis enables a quantitative description of brain anatomy and detection of subtle, but important, anatomical changes that may be difficult to detect by visual inspection. Therefore, it's imperative to develop sophisticated image analysis tools that can handle the highly complex and varied organization of the brain across individuals. Such tools will form the foundation for decision support systems (DSSs) to aid health professionals in more precise and personalized management of brain tumors.