Rajeev Krishnapillai MS, Abstract Co-Author: Nothing to Disclose

Milliam Kataoka, Abstract Co-Author: Nothing to Disclose

Oleg S. Pianykh, Presenter: Nothing to Disclose

Sam C Yam PhD, Abstract Co-Author: Nothing to Disclose

Vassilios D. Raptopoulos MD, Abstract Co-Author: Research grant, Toshiba Corporation, Tustin, CA

To develop web-based software solution for automated creation of oncology tables from measurement annotations made on PACS for tumor measurements and also to assist the radiologists to apply different clinical trial criteria .

Oncology table is a structured representation of  tumor measurements based on well defined existing methods of identification, classification and measurement. The following system is proposed: 1. Our web based software solution interacts with our institutional PACS to fetch DICOM image measurement annotations associated with specific exams upon request. 2. The application interacts with the queried annotation results and parses the queried blob of data as per predefined logic.Tumor annotations not relevant for the tumor metrics are filtered out during the process. 3. The fully parsed and structured lesion measurement and related annotation items are presented to the user for final review based on the user chosen criteria. 4. The final processing presents the user with any prior tumor metrics for the patient for comparision. Further the tumor response graphs and key tumor images are generated to finalize the report. The web application is developed with an interface to query the institutional PACS database.Further the application is implemented using ASP.NET and SQL2008 database backend.   

Currently available oncology table web application in our institution has the limitation of manual data entry and takes an average of 12 min for one case. Further discrepancies in lesion selection between actual report and oncology table appear in 8% of cases. The new application is currently being tested and early testing suggests ease of use and significantly lesser time (5 min or lesser) for creating an oncology table with no data transposing errors and overall reduction in lesion representation error to less than 2% . We are anticipating significant conformance to consistency in lesion measurement reporting in follow up measurements.

The solution significantly assists the radiologists to conform to trial protocols using RECIST, WHO and CHESON criteria as well as improves the consistency and ease of creating oncology tables on a regular clinical workflow.

The tool can enhance the radiology reports with structured oncology measurements and acts as an automated agent to help with consistency in radiology reports of oncology studies .

Krishnapillai, R, Kataoka, M, Pianykh, O, Yam, S, Raptopoulos, V, Web-based Application for PACS Image Annotation Export to Create Oncology Table on the Fly.  Radiological Society of North America 2009 Scientific Assembly and Annual Meeting, November 29 - December 4, 2009 ,Chicago IL. http://archive.rsna.org/2009/8012093.html