CTSC:ARRA.100609

Back to CTSC:ARRA supplement

Agenda:

1. Introduction
2. Brief scope of expertise
3. Scope of work
4. Conclusion

Harvard Catalyst Medical Informatics group Meeting Minutes October 6, 2009

In attendance:

• Valerie Humblet
• Mark Anderson
• Randy Gollub
• Yong Gao
• Bill Hanlon
• Ron Kikinis
• Todd Perlstein
• Wendy Plesniak
• Shawn Murphy
• Diane Keogh
• Mike Mendis
• Steve Pieper
• Richard Robertson
• Stephan Voss
• Danny Shaw
• Ramin Khorasani
• Kathy Andriole
• Michael Clyne
• Jesse Wei
• Alexander Zaitsev

1. Introduction

• Randy welcomed all the participants and gave a very brief introduction about the ARRA supplement. The Administrative Supplement award was received on September 30, the official dates of the project are 9/24/09- 9/23/11.
• Lee Nadler, as PI of the Harvard Catalyst is also the PI of this supplement. Bruce Rosen, as Director of the Catalyst Imaging Consortium and a visionary in the development of medical imaging informatics is also providing leadership for the project.
• Shawn Murphy and Randy Gollub are co-leading the operational aspects of the effort. Shawn will be the lead on the technical design. Randy will be in charge of all the administrative tasks. She is currently working with each site to establish their subcontracts. The first thing that Shawn and Randy did when the funding of the supplement was announced was to offer a full time job to a Computational Engineer that has many years of experience in research imaging data management. He was recently interviewed in person by the full Catalyst Imaging Consortium membership who gave him their strong endorsement. He will join the team in a couple of months after completing the transition of his current responsibilities to others and moving to Boston.

2. Scope of expertise

First all the participants introduced themselves, gave a brief update on their background and talked about what they hope to achieve within this project.

• Randy Gollub

Randy is from MGH (Psychiatry and Radiology), she is the co-director of the Catalyst Imaging Consortium and the co-Investigator on this ARRA supplement. Her motivation to complete the aims of this project is to enable some of the wealth of clinical imaging data accessible for secondary use for research. Randy brings expertise in the design, acquisition and analysis of human neuroimaing studies and in the domain of human subject protections, valuable since the issue of patient privacy will require a lot of effort in this project.

• Diane Keogh

Diane is the Director of Research Computing at Partners HealthCare. She is usually not involved in individual projects but, since she finds this particular project so compelling and important, she will provide her direct support to this group.

• Danny Shaw

Danny is the Director of Knowledge Management at Children's Hospital. The clinical research IT team reports to him. He hopes that the tools that will be created will be as self-service as possible.

• Ramin Khorasani

Ramin is Vice Chair of Radiology at BWH, he is also the Director of Medical Imaging Information Technology and the Director of the clinical PACS system. His goal is to support the research effort by making clinical data available easily as long as it does not disrupt clinical practice.

• Shawn Murphy

Shawn is the co-lead investigator on this supplement. He is also one of the Core Leaders for the i2b2 (Informatics for Integrating Biology and the Bedside) project. There are 3 i2b2 instances: BIDMC, CHB and Partners. So far all i2b2 efforts focus on collection of medical procedure, billing lab and other codes from patient care; however, the database completely lacks medical imaging data. He notes that medical imaging provides critically important phenotypic information.
The infrastructure that will be built during the next 2 years must be scalable to very large volumes of data (n = 10,000 +). He notes that there will be a need for flexibility, as the Radiology departments at each institution will have different workflows.

• Todd Perlstein

Todd is a Fellow in Cardiovascular Medicine at BWH, his interest is in cardiac ultrasound (US). He is also a member of the Catalyst Imaging Consortium. He will provide the group with good test cases for clinical research. For example, his group has collected a lot of cardiac US data but so far he has no access to it for secondary research use.

• Alexander Zaitsev

Alex is a member of the Surgical Planning Lab (SPL) lab at BWH. Right now he is working on supplementing the 3D Slicer software and Extensible Neuroimaging Archive Toolkit (XNAT) with a workflow management system. This capability will provide integration and interoperability between 3D Slicer, XNAT Services and 3-rd party image processing applications and high performance computing systems to enable the automatic image processing of very large data sets. The initial implementation of it is within the Structural Planning and Implementation in Neurological Exploration (SPINE) System deployed at BWH under the direction of Charles Guttmann.

• Stephan Voss

Stefan Voss is a radiologist specialized in pediatric oncology at CHB. He is also a member of the Catalyst Imaging Consortium. He defines himself as an end-user of image data, he is involved in several large multi-center clinical trials that are part of the Children's Oncology Group (COG). He has considerable experience with using image analysis and managements tools, many of which have not worked well.

• Richard Robertson

Rick is the Clinical Radiologist-in-Chief and the Medical Director of Radiology Informatics at CHB. He wants to understand the scope of the project to make sure that the right representative from his institution attends the weekly meetings. He will pay particular care to ensure proper oversight for accessing and moving the medical imaging data.

• Mike Mendis

Mike is a software development engineer and a project manager (team lead) for i2b2 who has many years of experience working with Shawn Murphy.

• Ron Kikinis

Ron is the founding Director of the Surgical Planning Laboratory at BWH. He has been doing full time research on image processing, 3D Slicer, multi-centers studies (such as BIRN) for many decades.
Research needs its own PACS, its own workflow. XNAT will fulfill a need for research because it is more flexible than the clinical PACS.

• Yong Gao

Yong is computer scientist working part-time for the Catalyst. He works towards data sharing across sites and is setting up XNAT instances at CHB and MGH.

• Wendy Plesniak.

Wendy is a Research Fellow in the Surgical Planning Laboratory at BWH. She has a lot of expertise on interface design and mapping use cases to the technology.

• Mike Clyne is the Systems Manager, MIIT / Data Storage, he works at BWH where he supports the PACS system.

• Kathy Andriole

Kathy is the Director of Imaging Informatics, CEBI (BWH). She is working on the construction of a research PACS, this system will be web-based and accessible from any Partners workstation. One of her goal is also to expose commercial tools to the research community, her group is doing a lot of testing of these tools.

• Steve Pieper

Steve is a medical image analyst, he is the founder of Isomics. That company provides specialized services to academic and business clients in the fields of software development, image analysis, and technical management with applications in medical research, product development, and manufacturing. He has a lot of experience in multi-institutions software venture. One of his main concerns is to make sure that the research work will not affect clinical practice.

• Jesse Wei

Jesse is a radiologist at BIDMC, he is also the director of informatics for the Radiology department. He is very interested by the possibility to integrate images with i2b2 and sees the benefit of this project both locally for use by BIDMC faculty as well as more broadly.

• Mark Anderson

Mark is Research Programmer in the Surgical Planning Lab (SPL) at BWH. Right now one of his projects is the implementation of a XNAT server for BWH.

• Bill Hanlon

Bill is the project manager for the tumor image metrics core (TIMC). He is bringing a lot of expertise in technical issues linked to data sharing between institutions. Eventually TIMC could be an end-user of this new system. Their special requirement is a very short turn around (48 hours or same day)

3. Scope of work

• Shawn gave a presentation of the scope of the project highlighting key points where decisions need to be made by this group for how to achieve our goals.
  • One goal is to be able to request medical images from within i2b2 with an accession number. A key question is how to implement the request for the medical image data from the PACS. For instance, software could be built to automatically send a request to the clinical PACS directly or their could be intermediary steps. The extracted images are then envisioned to be sent to a Centralized repository (XNAT) likely one at each institution.
  • After de-indentification, the images will be sent to XNAT (see www.XNAT.org). XNAT is an open source software developed at Wash U by Dan Marcus. It is a system designed for clinical research and its has its own specifications (how the data need to be stored, data manipulation etc).
  • The architectures of XNAT and i2b2 are compatible such that the use of plug-ins will allow a PI who requests the images to stay in the i2b2 environment and still use the viewing capabilities of XNAT for his/her research project. Once the data are in the XNAT environment, the PI is free to analyze the data using whatever tools he/she requires. Derived new data can then be entered into the i2b2 database. It is envisioned that over the next 2 years, there will be no image data sharing across institutions, we will we are working on an implementation that will support that in the future.
• Ramin proposed an alternative that he described as is a much simpler solution that does not require the same need for security and privacy. This alternative would be to provide image viewer tools linked to the PACS. For example in RPDR, one can find all the data for a patient. It would be great to add a link to be able to view the images. He proposed that this simpler model could take care of 70 % of the research needs. There was spirited discussion illuminating the need for further clarification of the end goals of this project (see next week's agenda).
  • The discussion logically evolved to the problem of determining the image post-processing needs most often used in research. If the PI is only able to view the images, he will not be able to use all the specialized research tools available.
• Shawn briefly defined 3 use cases for this project:
  • The clinical researcher who tries to confirm a diagnosis. He can used a client based viewing tool.
  • The clinical researcher who wants to view the images and evaluate them but doesn't need post-processing.
  • The clinical researcher that we really want to serve: the one that wishes to apply non-clinical image analysis software to do post-processing automatically and needs the DICOM images. XNAT supports this final use case very well.

4. Some conclusions

• The goal of this project is not to create a mirror PACS.
• Each institution should bring to future meetings (with such a defined agenda) their own institutional requirements for oversights so we can assess similarities and differences.
• There is a need to clarify who can request what data and the process of how access to imaging data will be gained when someone places a request.
• Each institution has to decide who from their team should attend future meetings of this project. Having a defined agenda ahead of time would help making that decision, it is likely at this early stage senior leadership will be invaluable on almost on a weekly basis, but that will hopefully quickly evolve to more strategic work.
• Randy proposed to do a demonstration to teach what SHRINE and i2b2 can offer so far.
• Ron suggested a discussion of what XNAT central has done.

5. Key links

IHE, HIMSS, RSNA XDS