Image De-Identification Service

Protecting the health and safety of Froedtert and the Medical College of Wisconsin patients is a top priority. To facilitate research, the CTSI provides a service for anonymizing images and a centralized data management resource to support investigators and research studies. The service center has implemented a series of systems that consistently enforce HIPAA regulations for image extracts from PACS, thereby protecting patient confidentiality and allowing selective data sharing among collaborators and industry sponsors.

All imaging data will be delivered within our XNAT environment, which allows us to make imaging data openly accessible to collaborators. XNAT is an open-source imaging informatics software platform developed by the Neuroinformatics Research Group at Washington University. By managing data with XNAT, laboratories are prepared to better maintain the long-term integrity of their data, to explore emergent relations across data types, and to share their data with the broader neuroimaging community.

Frequently Asked Questions

Image De-Identification

What is the process of using de-identified images in publications or presentations for education and training?

MCW Corporate Policy (AD.HP.040) Use and Disclosure of PHI for Treatment, Payment & Health Care Operations requires a PI to get separate patient authorization for the use of protected health information. The policy states: An authorization from the patient is required prior to using PHI for external case study presentations or in publications, or at grand rounds or other forums when attendees may include non-MCW employees/residents/students. Use of PHI in these instances is not considered health care operations.

In order to get images de-identified for education and training complete the De-Identification Request Form (image de-identification form) and enter Education & Training as the Study Name. Fill in today’s date as the start and end dates. Once the form has been completed we will contact you.

Permalink 1 0

What is the cost of this service?

Please view the Processes and Fees page for more information.

Permalink 0 0

How quickly will I receive my images?

After receiving approval from OCRICC, standard turnaround time is 5 business days. You may request expedited processing (1 or 2 days), which carries an additional fee per request. No work on de-identification is performed during non-bussiness hours or on weekends and holidays.

Permalink 0 0

Can you de-identify images from Children’s Hospital or the Clement J. Zablocki VA Medical Center?

No, we solely have processes and agreements in place for access to images from Froedtert.

Permalink 0 0

Can I get the images in formats other than DICOM?

No, we only deliver images in DICOM format. You are responsible for conversion to other formats using tools like RadiantViewer.

Permalink 0 0

Can you handle data formats other than those that come from PACS, such as echocardiogram movies?

No, video formats are not available yet. We are currently Piloting video de-identification.

Permalink 1 0

Once de-identified, how can I tell which accession numbers below to which subject, since this data is removed from the images?

If you are already using de-identified Subject IDs, please use them when you submit your list of Accession IDs. We will then use your IDs. If not, your project will get random subject IDs. Example: Subject1, Subject2, etc.

After each set of images has been processed, we will provide you with a spreadsheet that includes Accession IDs you submitted paired with the subject IDs we used for you to keep in your records. For security reasons, these will never be stored on the XNAT.

Permalink 0 0

How long will my images be stored on the XNAT Server for our study?

There is currently no time limit. We are in the process of determining length of storage, but a timeline has not yet been determined. We will not remove your images without notifying you in advance.

Permalink 0 0

Language to Include in IRB Applications

Language to Include in IRB Applications

TEMPLATE LANGUAGE FOR INVESTIGATORS USING MCW CLINICAL RESEARCH DATA WAREHOUSE (CRDW) – FOR USE IN MCW IRB APPLICATIONS

FOR ALL 5 SCENARIOS BELOW: Investigators that MIGHT utilize CRDW data MUST link their eBridge study applications to the CRDW bank (PRO00013874) in Section 26 of the SmartForm.

Please explain your use of the CRDW using the following templated language within the indicated sections of the IRB application SmartForm… otherwise the IRB will not understand what you are doing and the review of your application, as well as the approval of your data request(s), will be delayed.

I. FOR INVESTIGATORS USING THE CRDW TO IDENTIFY AND/OR CONTACT ELIGIBLE PARTICIPANTS

In the Recruitment section (17.1) of SmartForm, the investigator must clearly state whether the CRDW is the ONLY way potential participants will be identified or the CRDW is one of several recruitment strategies.

The investigator should state:

“To identify potential participants and obtain their contact information for recruitment, the investigator will (a) search for eligible participants with the following inclusion/exclusion criteria by querying a cohort discovery tool (i2b2 and/or TriNetX) from the MCW Clinical Research Data Warehouse (PRO00013874) , and then (b) seek IRB approval to extract the following contact information for each identified potential participant using the Honest Broker tool:

  1. <Specify cohort inclusion/exclusion criteria to be queried>
  2. <Specify participant contact information to be extracted>
Using this contact information, the investigator will then…”

II. FOR INVESTIGATORS USING THE CRDW TO ACCESS IDENTIFIED DATA WITH OR WITHOUT PARTICIPANT RECRUITMENT

In the Protocol Summary (19.7) or Procedures (30.1) section of the SmartForm, the investigator must clearly state whether the CRDW is the ONLY way potential participants will be identified or the CRDW is ONE OF SEVERAL participant identification strategies.

The investigator should state:

“Clinical data for this study will be obtained by… <indicate one of the following options>

  1. …searching for eligible participants by querying a cohort discovery tool (i2b2 and/or TriNetX) from the MCW Clinical Research Data Warehouse (PRO00013874) with (a) the following inclusion/exclusion criteria, and (b) seeking IRB approval to extract or access the following clinical information for each identified participant using the Honest Broker tool:
    1. <Specify cohort inclusion/exclusion criteria to be queried>
    2. <Specify clinical information to be extracted/accessed>
  1. …entering a list of medical record numbers for identified participants into the Honest Broker tool and then submitting an approval request to the IRB for the following clinical information for each participant:
    1. <Specify clinical information to be extracted/accessed>
Using this identified clinical information, the investigator will then…”

III. FOR INVESTIGATORS USING THE CRDW TO ACCESS FULLY DE-IDENTIFIED DATA OR A LIMITED DATA SET for purposes other than participant recruitment

**Some investigators may not be required to submit anything to the IRB for this scenario – contact CRDW.mcw.edu for guidance. Other investigators may have “mixed” projects involving CRDW de-identified data AND other activities requiring IRB review.**

In the Protocol Summary (19.7) or Procedures (30.1) section of the SmartForm, the investigator must clearly state whether the CRDW is the ONLY way potential participants will be identified or the CRDW is ONE OF SEVERAL participant identification strategies.

The investigator should state:

“Clinical data for this study will be obtained by searching for eligible anonymous participants using a cohort discovery tool (i2b2 and/or TriNetX) from the MCW Clinical Research Data Warehouse (PRO00013874) with (a) the following inclusion/exclusion criteria, and then (b) extracting the following de-identified clinical information for each identified participant:

  1. <Specify cohort inclusion/exclusion criteria to be queried>
  2. <Specify clinical information to be extracted>
Using this de-identified clinical information, the investigator will then…”

IV. IMAGE DE-IDENTIFICATON (the most typical situation) the investigator has IRB approval to access identified F&MCW images, but wants them de-identified so the images can be sent outside F&MCW in de-identified format

The investigator should “prove up” to the CRDW gateway that the investigator already has IRB approval to access or use these specific images, and IRB approval for access to IDENTIFIED images. When in doubt, the CRDW can and should always check with an IRB staff person.

The investigator wants to send a scan image to an external party (sponsor, collaborator) de-identified because the STUDY PROTOCOL requires de-identification in this situation, and/or the IRB APPLICATION promised de-identification in this situation, and/or the STUDY CONSENT FORM promised de-identification in this situation. Thus — even though scrutiny of the HIPAA authorization language may show that the participant has in general authorized the investigator to share some identified information with the third party — de-identification of the image scan is still REQUIRED.

In the Procedures section (30.1) of SmartForm, the investigator should state in the IRB application:

“Identified scan images to be provided to XXX and YYY (external third parties) will be totally de-identified per study protocol by submitting index images and destinations to the Honest Broker / Image De-Identification Service of the MCW Clinical Research Data Warehouse and having that Honest Broker send the de-identified images.” [Though it may be clear elsewhere in the protocol, the PI should again state here what kinds of images and how many images will be sent this way.]

V. IMAGE DE-IDENTIFICATION (less common situation) where the investigator wants to use a CRDW cohort discovery tool (i2b2 and/or TriNetX) to identify eligible participants and then have the CRDW deliver de-identified images for that cohort, so the activity (on the part of the PI receiving de-identified images from a valid honest broker) does not technically qualify as “human subjects research.”

In the Procedures section (30.1) of SmartForm, the investigator must clearly state whether the CRDW is the ONLY way potential participants will be identified or the CRDW is one of several participant identification strategies.

The investigator should state in the IRB application:

“Clinical data for this study will be obtained by using a cohort discovery tool (i2b2 and/or TriNetX) from the MCW Clinical Research Data Warehouse (PRO00013874), to (a) search for eligible participants with the following inclusion/exclusion criteria, and then (b) seek IRB approval for this project asking the CRDW to provide the following clinical information and SCANS for each identified participant:

  1. <Specify cohort inclusion/exclusion criteria to be queried>
  2. <Specify clinical information to be extracted>

 

Note: this method of accessing de-identified data may ultimately be deemed “not human subjects research” depending on what else the investigator will be doing.

Citing CTRI Image De-Identification Services

Citing CTRI Image De-Identification Services

Please acknowledge the Medical College of Wisconsin CTSA Grant 8UL1TR000055. All publications must comply with NIH Public Access Policy.

Researchers other than K Scholars (KL2) – Please cite Grant 2UL1TR001436, as follows:
“This publication [or project] was supported by the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant Number 2UL1TR001436. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.”
K Scholars (KL2) – Please cite Grant 2KL2TR001438, as follows:
“This publication [or project] was supported by the National Center for Research Resources, the National Center for Advancing Translational Sciences, and the Office of the Director, National Institutes of Health, through Grant Number 2KL2TR001438. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.”