Mining Brain Tumors and Tracking their Growth Rates

Abstract

Mining brain tumors and tracking their growth trends in the course of magnetic resonance imaging is an important task that assists medical professionals to describe the appropriate treatment. Nevertheless, applying conventional techniques to carry out this process manually is time-consuming and often unreliable and insufficiently accurate. Automating this process is a challenging task due to the fact of the fractal shape of tumor and its biological structure, which is often, has a high degree of intensity and textural similarity between normal areas and tumor tissues. Moreover, tumor uptake measurements are not easy given the small size of many tumors, the limitations of spatial resolution, and the change of tumor location from slice to slice across the brain. Furthermore, the arbitrary shape of tumors makes it extremely hard, if not impossible, to adopt traditional geometric rules for tumor measurements. In this paper, we present a computational approach for modeling and mining a large number of MRI data for patients with brain tumors. In this approach, we adopt a spatial data mining technique to extract useful information from MRI data in order to identify the size of tumors and growth trend, as well as classifying tumors of patients upon specific similarity measures.