Design and Prototype of an Object-Relational Database for Medical Images

Current electronic patient medical records are based on relational databases that have been very successful in managing text-based information. Recently, however, complex data types, such as digitized images, are becoming increasingly important in healthcare. Unfortunately, existing relational databases are not able to manage these complex data types in a satisfactory manner. The goal of this project is to study the feasibility of developing a high-performance image archive that will handle both text and complex data types. In order to achieve this goal, this project designed and prototyped a medical imaging archive. The prototype consists of an Informix Universal Server database engine, an object-oriented database schema, and an application-programming interface (API). The schema and API are implemented in a software module called a DataBlade. This DataBlade extends a generic Informix database with the ability to manage medical information objects as defined in the DICOM (Digital Imaging and Communications in Medicine) standard. The schema incorporates an extensible inheritance hierarchy, while the API includes encapsulation of routines and data structures built specifically for DICOM information. The prototype demonstrates how developers can be shielded from the complexities of the DICOM standard, yet still be able to store and query the complete DICOM information model. Thesis Supervisor: C. Forbes Dewey, Jr. Title: Professor of Mechanical Engineering