Enterprise Imaging: Data Quality, Migration and Management
Implementing a central image repository in the form of a Vendor Neutral Archive (VNA) provides an organization the unique opportunity to standardize data acquisition and apply data quality standards in the process. Whoever administers the VNA will also need to ensure data quality is maintained. With high data quality, the system can leverage data mining for research, clinical, and business purposes.
All data will ultimately age out. Therefore, information lifecycle management (ILM) needs to be part of overall data management. Different regulatory requirements dictate the minimum time of diagnostic image retention. The system needs to intelligently identify and track each image type and recognize when the data is ready to retire.
Early on, the governance committee should establish what will be done with aged-out images. Should they be deleted or just anonymized and retained for research? Should they be moved to a cheaper storage location? The organization’s health information management (HIM) and risk management groups should be involved in the decision-making process and testing of data retention/destruction algorithms with a special focus on litigation issues. Any data that is locked in litigation should be reviewed for unlocking once the case has been resolved. Deleting old data has the added benefits of reducing overhead costs and helping speed up database search activities.
VNAs now come with many tools to assist in making the data easy to access and manipulate. Some key features include leveraging Imaging Object Change Management (IOCM), which has the ability to update all the databases that store and manipulate DICOM images. This feature is especially important because specialty PACS systems often have images annotated or new views created and added to a study. These updates need to be transmitted back to the VNA. Another feature the VNA should offer is the ability to anonymize data. This means all information identifying a patient is removed, but the image integrity is maintained for research like deep learning for artificial intelligence (AI).
Items to investigate include:
- Are presentation states preserved?
- Are these presentation states properly displayed in other systems?
- Is data being fully synchronized and additions and deletions properly reflected in a bidirectional manner?
- Does the VNA properly respond to patient merges using HL7 protocols?
- How do the systems respond to patient un-merge? What are the tools and workflows?
- How does the system manage DICOM tag morphing?
- What tools does the vendor offer to anonymize images with burned-in PHI data (identifying information that appears inside the image, e.g., ultrasound images)?
When a VNA is introduced, data migrations from other image storage will become one of the major chores in setting up the historic database. The data management team and the clinical owners of the historic data need to review and discuss what data needs to be migrated. The following steps need review.
- What is the volume of data that should be migrated to the VNA?
- How good is data quality in the source system?
- Does it contain all the needed meta data to be properly identified within the VNA?
- What will be done with data that does not meet the ingestion criteria?
- Are data sets being checked for duplicates?
- Does tag morphing need to be addressed?
- How will non-migrated data be maintained until it exceeds its useful/regulatory lifetime?
- What date is associated with the oldest data?
- Will you migrate the youngest data first?
- How will the pilot migration be conducted?
- Who will validate the data for the pilot?
- Who will conduct data validation once the bulk data starts migrating to the VNA?
If the organization is adding legacy data from mergers, acquisitions, and partnerships, multiple patient identity domains exist and an enterprise master patient index (EMPI) will be needed. The EMPI will be very closely linked to the electronic medical record (EMR) system(s) used by the organization.
Engaging the imaging and data governance committees involved with and establishing policies, workflows, and rules around the EI will provide a good foundation for implementing the program.