Reengineering Workflow for Curation of DICOM Datasets.

Area of Science:

• Medical Imaging
• Bioinformatics
• Open Science

Background:

• The Cancer Imaging Archive (TCIA) is crucial for advancing cancer imaging research through data sharing.
• Increasing data volume and complexity in TCIA submissions have created curation bottlenecks.
• Ensuring data consistency, scientific utility, and patient anonymity is paramount for TCIA.

Purpose of the Study:

• To develop and describe novel tools for efficient curation of DICOM datasets submitted to TCIA.
• To improve the speed, quality, and provenance tracking of data curation processes.
• To facilitate the analysis and management of large-scale cancer imaging data.

Main Methods:

• Developed the open-source "Posda Tools" using Perl, PostgreSQL, and existing DICOM packages.
• Implemented a permissive database schema and automated bulk integrity checks for DICOM datasets.
• Enabled bulk revisions of DICOM data via web interface or command line, with a revision tracker for rollback capabilities.
• Incorporated a UI for verifying edits and methods for identifying duplicate datasets and managing DICOM errors using persistent nicknames.

Main Results:

• The Posda Tools significantly increase the rate and quality of DICOM data curation.
• Novel methods facilitate rapid analysis, automated integrity checks, and bulk data revisions.
• The revision tracker ensures data provenance and allows for rollback of changes.
• Features for identifying duplicate datasets and managing DICOM errors improve data management.

Conclusions:

• The Posda Tools address the curation bottleneck in TCIA, enhancing the efficiency and reliability of cancer imaging data management.
• These open-source tools support the principles of open science by improving data accessibility and reproducibility.
• The novel methods employed by the Posda Tools offer a scalable solution for managing complex DICOM datasets in research repositories.