Conference Paper

iCube: A similarity-based data cube for medical images.

DOI: 10.1109/CBMS.2010.6042663 Conference: IEEE 23rd International Symposium on Computer-Based Medical Systems (CBMS 2010), Perth, Australia, October 12-15, 2010
Source: DBLP

ABSTRACT Issuing analytical (OLAP) similarity queries based on images over data warehouses is a core problem in the medical field, as these queries can allow for the investigation and analysis of health data in decisionmaking processes. In this paper, we propose iCube, a similarity-based data cube for medical images. iCube is an extended data warehouse that encompasses a dimension table specifically designed to store intrinsic features of images, therefore allowing OLAP similarity queries over images. We also show how to build iCube and how to perform OLAP query processing over images. Comparisons of iCube with the current data warehousing technology aided by a metric access method showed that iCube provided an impre ssive performance improvement to process OLAP similarity queries over images. iCube performance gain ranged from 43% up to 76.7%.

  • [Show abstract] [Hide abstract]
    ABSTRACT: Medical images are fundamental in medical processes particularly in the disease surveillance which is a practice that enables to monitor the evolution of patients' states. This practice cannot be understood and described only by a current image but requires the observation of image sequences in order to follow up the evolution of the disease from one human body location to another. This work aims to model a data warehouse where images and their related sequences are gathered and analyzed for decision making purposes such as disease evolution surveillance. The images' features are gathered as intrinsic features representing both the content-based and the description-based descriptors combined to the experts' annotations. We take into account the various modalities of images with the related temporal relationships which describe the sequence, and the conventional dimensions interfering for the target analysis.
    Proceedings of the 2013 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining; 08/2013