Big Data Integration Case Study for Radiology Data Sources

Abstract

Today's digitized world urgently needs Big Data integration and analysis. Healthcare records are responsible for generating petabytes of data in a single day. Such data is heterogeneous in nature, captured in different files and formats, and varies from hospital to hospital. By integrating data from different sources and extracting meaningful information for the medical community, we can improve the overall quality of patient care. Our research targets the problem of integration for health records. To start, we already developed the Integrated Radiology Image search (IRIS) engine, which could represent a data integration framework for the healthcare domain. IRIS provided support for multiple public data sources and incorporated medical ontologies which would help radiologists and improve search interpretation by considering the meaning of the search query terms. In this paper, we describe a case study of data integration for radiology data sources. While the need for data integration is self-evident, we learned that rather than being a single step, data integration is an iterative process that requires continuous integration of metadata and additional supporting data sources. Our results show that an each step of data integration further improved IRIS engine results.