Thomas Nabhani MD, PhD, Presenter: Nothing to Disclose

Todd R. McNutt PhD, Abstract Co-Author: Stockholder, Koninklijke Philips Electronics NV

Theodore L. DeWeese MD, Abstract Co-Author: Data and Safety Monitoring Board, GenVec, Inc, Gaithersburg, MD

Joseph Michael Herman MD, MSc, Abstract Co-Author: Nothing to Disclose

Fariba Asrari MD, Abstract Co-Author: Nothing to Disclose

Radiation oncology research becomes more difficult as data becomes increasingly complex. Excel spreadsheets and Access databases (DB) are useful for managing small datasets, but are not well suited for multidisciplinary and multi-institutional trials with respect to security, data integrity, data reuse, and scalability. The response to this issue has been for consortia to develop specialized software for creating and managing clinical trials. This provides great benefit for very large groups, but fails to address the needs of the majority of clinicians who have neither the resources nor the skills to implement sophisticated software systems needed for electronic clinical research. Our objective was to provide a virtual community for radiation oncology research to facilitate prospective, online, collaborative clinical research integrating imaging, dosimetry, molecular, pathological and clinical data without requiring significant technical skill on the part of the trial physicians. The software consisted of several relational DB, a data access layer, business logic layer, and a presentation layer. Data was pulled from multiple sources, including PACS, patient records, consults, and tumor boards.

The system design was guided by a few principles: security, data integrity, distributed access, data centralization, and ease of use. In essence, we developed a content management system for cancer research. Our system is engineered using standard tools: MS Visual Studio, C#, SQL Server, T-SQL, ASP.Net, and Ajax. The DB design is based on a framework developed for the SkyServer project.

Although our system makes use of several DB, we focus here on those directly related to clinical research and decision-making. The schema for the clinical data management DB was designed by closely examining data tables, graphs, and charts from >250 published radiation oncology trials. The data types extracted from the trial documents were combined with terms extracted from standard clinical classification scales (i.e. AJCC staging, CTCAE grading, Spetzler-Martin arteriovenous malformation (AVM) grade). This set was semantically clustered to obtain hierarchical data classes. These classes were entered into a vocabulary DB to provide a standardized cancer lexicon. Our study of the cancer trial data terms also facilitated the design of our cancer data warehouse. The terms were projected onto multiple document models (i.e. history and physical, radiology report, radiation therapy plan) to obtain a robust schema to support a wide variety of research activity. Visually programmable web services, user interface controls, and middleware objects were developed to communicate with the DB. These objects were assembled to create a diverse set of data management instruments. The instruments included a decision support system for a multidisciplinary pancreatic cancer clinic, a data capture tool for AVM research, and a set of forms for breast cancer research.

We developed a virtual community for radiation oncology research. The software is secure, scalable, and easy to use. The software was used to construct a variety of tools for radiation oncology research. The software is easy to customize and deploy greatly facilitating complex collaborative research projects.

Nabhani, T, McNutt, T, DeWeese, T, Herman, J, Asrari, F, OncoSpace: An Online Community for Collaborative Radiation Oncology Research.  Radiological Society of North America 2007 Scientific Assembly and Annual Meeting, November 25 - November 30, 2007 ,Chicago IL. http://archive.rsna.org/2007/6000679.html