MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Anonymization of DICOM Files in HTML5 Based Web Browser for Radiation Therapy

P Rana, W Sleeman*, M Poblacion, J Palta, P Ghosh, R Kapoor, Virginia Commonwealth University, Richmond, VA

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: There is a growing need to share DICOM-RT data amongst institutions for rapidly assembling datasets to address practice quality improvements and Big data research. Before sharing the data, the files must be anonymized by removing protected health information (PHI) to ensure patient privacy. Several desktop installed applications, such as DICOMCleaner, provide anonymization features but lack a zero-footprint browser-based functionality.
We propose a prototype, zero-footprint, single-click HTML5 browser tool that can anonymize and upload multiple DICOM files at the client-end where the anonymization task is performed in the local browser memory stack ensuring no PHI leaves the facility at any time instant.


Methods: Our DICOM anonymization tool is built on the popular dicomParser JavaScript library. Using this library, we parse out the sensitive DICOM tags and overwrite them with the new anonymized value of same length or emptied the tag values completely. The software pipeline consists of DICOM upload, DICOM parsing, unique identifier (UID) validation (involving file correctness checks), anonymization and sending anonymized files to the backend.


Results: Our prototype system has the ability to anonymize 40 different tags covering 19 standard PHI item groups. It also creates fresh UIDs and consistently replaces them in the datasets preserving the identified relationships between the files automatically. Our tool supports multiple file-uploads and batch-processing so that a complete dataset of DICOM files can be anonymized and uploaded in a single click. It takes about 20 ms per file-upload and total runtime grows linearly with number of files due to the recursive dependency checking stored in a tree-based data-structure.


Conclusion: Our prototype meets the requirements for anonymizing DICOM datasets at the client-end while taking care of privacy concerns using web browser-based technologies. Performance analysis of the software confirmed both its scalability and data integrity, i.e., preservation of the relationships between DICOM UIDs.

Funding Support, Disclosures, and Conflict of Interest: National Radiation Oncology Program, US Department of Veterans Health Affairs

Keywords

Not Applicable / None Entered.

Taxonomy

IM/TH- Informatics: Data archiving - Therapy

Contact Email