Internet2/NLM infoRAD
Demos and Tutorials at RSNA

Around 30,000 radiologists, healthcare administrators, technologists and scientists, plus another 30,000 imaging and information system vendors from all over the world, are expected to attend the Radiological Society of North America (RSNA) annual meeting November 28 to December 3, 2004 in Chicago, making it one of the world's largest medical meetings. RSNA supports the exchange of scientific progress in radiology, radiologic education, and in the integration of information and communication systems in radiology practice. The Annual Meeting provides workshops, training, and conference sessions for health professionals. In addition, nearly 500,000 square feet of exhibit space – feature product demonstrations and a broad range of exhibits. The Internet2/NLM tutorials and demos will be part of the infoRAD exhibit space. The infoRAD area is designed to showcase the most innovative technology solutions in an interactive, educational environment. Hands-on demonstrations are encouraged. Computer-aided instruction, digital imaging and communications in medicine, new technologies, computer-assisted diagnosis, and literature searches are just some of the topics featured in infoRAD exhibits.

Featured Demos for infoRAD

The following demos will be featured at the Internet2/NLM tutorials in the infoRAD exhibit area at RSNA.

• Distributed Collaborations Using Neuroinformatics and Information Technology Innovations
• HAVnet: Internet2 Enabled Visualization and Haptics for Anatomical Education
• Immersive Real-Time Tele-Collaboration of Complex Volumetric Medical Imaging for Surgical Planning
• TeleVolView: A Tele-Collaboration Application for Follow-up of Abdominal Aortic Aneurysm Patients

Internet2/NLM infoRAD Tutorial Schedule

Thanks to Our Sponsors!

In addition to the sponsorship and support provided by Internet2 and NLM, we would also like to thank the following organizations for sponsoring these tutorials and demos.

Radiological Society of North America (RSNA)
Metropolitan Pier and Exposition Authority (MPEA)
Metropolitan Research and Education Network (MREN)
SBC

Distributed Collaborations Using Neuroinformatics and Information Technology Innovations

http://www.nbirn.net/index.htm
http://www.nbirn.net/AU/Events/RSNA_2004/index.htm

Developed at:
University California, San Diego

Demonstrators:
Jeffrey Grethe
Jorge Jovicich
James MacFall
Bruce Rosen
Ron Kikinis

Contacts:
Jeffrey Grethe
jgrethe@ncmir.ucsd.edu
(858) 822-0703

Steve Pieper
pieper@bwh.harvard.edu
(617) 525-6222

Partners:
Center for Functional Neuroimaging Technologies, Massachusetts General Hospital
Surgical Planning Laboratory, Brigham & Women's Hospital
Center for Image Science, Johns Hopkins University
fMRI Research Center, University of California, San Diego
Laboratory of Neuro Imaging, University of California, Los Angeles
Neuropsychiatric Imaging Research Laboratory, Duke University
Brain Imaging Center, University of California, Irvine
Howard Hughes Medical Institute
Washington University

Funded by:
NIH

Description:
The Biomedical Informatics Research Network (BIRN, http://www.nbirn.net), is an NIH information technology initiative that fosters distributed collaborations in biomedical science. Currently, this growing consortium involves 22 research sites from 14 universities and hospitals participating in three testbed projects centered on neuroimaging studies of human neuropsychiatric illness and associated animal models. The BIRN is developing a secure, shared infrastructure for collaboration, data sharing, data integration, and analysis in basic and translational research that is available from any Internet capable location. As a part of this open system, software tools and resources are being made available to the biomedical community for their use.

The Brain Morphometry BIRN testbed focuses on correlating structural brain differences to neuropsychiatric disorders, starting with studies of Alzheimer's Disease and depression. The overall goal is to develop the capability to share, analyze, mine and interpret both imaging and clinical data acquired at multiple sites using advanced processing and visualization tools, also developed at multiple sites. Two fundamental information technology requirements needed for this research are the secure integration of distributed heterogeneous data (i.e. both databases and data repositories) and the availability of high performance computing resources for advanced analyses and visualization. Three focused applications will be demonstrated that highlight the use of the BIRN infrastructure to advance science within the Morphometry BIRN:

1) The Alzheimer's Project demonstrates how the BIRN infrastructure can be used for mining multi-site clinical MRI studies with a preliminary study of Alzheimer's disease that integrates legacy data from clinical research studies on going at Brigham & Women's Hospital, Massachusetts General Hospital, and UCSD.  This demonstration utilizes the data mediation infrastructure deployed within BIRN that enables researchers to submit multi-source queries and to navigate freely between distributed databases. Unlike a data warehouse, which copies (and periodically updates) all local data to a central repository and integrates local schemas through the repository's central schema, the mediator approach creates the illusion of a single integrated database while maintaining the original set of distributed databases.

2) The Multi-site Imaging Research in the Analysis of Depression (MIRIAD) project integrates advanced brain morphometry tools from multiple sites (Brigham & Women's Hospital and UCLA) to analyse MRI structural data from one site (Duke) and measure volume changes in cortical and subcortical gray matter, that correlate with various clinical measures in depression and age-matched controls. Utilizing high performance computing through the BIRN infrastructure, the automated segmentation analysis within the MIRIAD project is able to generate many more segmentations (i.e. tissue classes and brain regions) in 1 hour than what was originally completed in a semi-automated fashion in 650 hours. Utilizing this advanced computational infrastructure, the MIRIAD project is the first study to show brain structural change over time in response to treatment in unipolar depression.

3) The Semi-Automated Shape Analysis project (SASHA) is developing a seamless and robust processing pipeline among multiple institutional sites (initially Massachusetts General Hospital, Johns Hopkins University, and Brigham & Women's Hospital) that segments sub cortical structures from structural MRI data, computes the geodesics in the space of infinite dimensional diffeomorphisms, visualizes results and enables statistical analyses of the results. This approach is being used to test the hypothesis that hippocampal shape differs between patients with Alzheimer's disease and demographically matched healthy controls.  Similar to the MIRIAD project, the utilization of the BIRN infrastructure has led to a significant decrease in the amount of time required to complete these analyses.

Role of Internet2:
Internet2 provides the backbone for all the distributed data resources within the BIRN.  Additionally all data transfers to and from computational resources within BIRN and also outside of BIRN (e.g. the TeraGrid used by the SASHA project mentioned above) utilize Internet2. For example, the SASHA project can produce mutliple terabytes of data from a standard analysis that must then be transported to and then further analyzed and visualized by researchers at Johns Hopkins University.

HAVnet: Internet2 Enabled Visualization and Haptics for Anatomical Education

http://havnet.stanford.edu/
http://visu.uwlax.edu/NGI/NGI.html

Developed at:
Stanford University
University of Wisconsin, La Crosse

Demonstrators:
Parvati Dev
Steven Senger
W. LeRoy Heinrichs
Robert Cheng

Contact:
Parvati Dev
parvati@stanford.edu
(650) 723-8087

Funded by:
NLM

Description:
We will demonstrate a group of applications that are being actively developed as part of the HAVnet project funded through a National Library of Medicine SII contract. The applications incorporate end-to-end performance monitoring and leverage the unique quality-of-service attributes provided by Internet2 to provide access to remote computation and data resources. The applications make extensive use of radiological data sets (CT, MRI) and provide unique forms of collaboration. The live demonstration will use Access Grid technology to provide a strong sense of presence between attendees at the RSNA site and HAVnet collaborators at Stanford University and the University of Wisconsin, La Crosse. The demonstration area will include a small projection screen to display the Access Grid video streams from multiple cameras at the remote site. This arrangement will provide attendees with a rich view of the work environment of the remote sites and the activities of the remote collaborators. This will enhance the ability of the attendees to understand both the local and remote sides of the demonstration and the role the network plays in enabling collaborative opportunities. Among the applications which can be demonstrated are the Remote Stereo Viewer, Immersive Segmentation, Nomadic Anatomy Viewer and the Virtual Emergency Room.

The Remote Stereo Viewer application provides users with interactive and collaborative access to large stereoscopic image sets. The Immersive Segmentation application allows users to collaboratively visualize and segment volumetric data sets using a stereoscopic immersive interface. The application produces multiple, high-bandwidth visualization streams and supports haptic feedback to provide users with additional control over the segmentation of anatomical structures. The Nomadic Anatomy application combines the capabilities of handheld computers with multiple remote servers to provide access to large volumetric data sets. The intuitive interface allows remotely separated users to engage in collaborative exploration. The Virtual Emergency Room is a virtual 3D world that supports "face-to-face" interaction by remote users.

Role of Internet2:
The applications proposed for this demonstration cover a spectrum of bandwidth and latency requirements that cannot be supported by the commercial Internet. Each visualization stream produced by the Immersive Segmentation application can produce brief bursts at data rates of 70 Mbps and sustained rates up to 40 Mbps, depending on the mode of operation. Each stream can be transported either by unicast UDP or multicast. In addition to the visualization stream, the server also supplies the client with a stream of data used to produce a haptic feedback response at the client station. The Nomadic Anatomy application has modest bandwidth requirements but does require low latency connections to multiple remote servers to provide a seamless, intuitive interaction model. The potential use of DVTS for videoconferencing would require 30 Mbps per stream.

Immersive Real-Time Tele-Collaboration of Complex Volumetric Medical Imaging for Surgical Planning

http://www.immersivemedical.com/

Developed at:
Immersive Medical Systems
Digital ArtForms, Inc.
University of Kentucky

Demonstrators:
Michael J. Mastrangelo, Jr., M.D.
Michael Sheetz, Ph.D.
Cody Bumgardner
Ivan George
Paul Mylniec
Jeff Bellinghausen

Contact:
Michael J. Mastrangelo, Jr.
m1789@mac.com
(541) 390-3423

Description:
This demonstration uses commercial data manipulation tools to render 3D models from DICOM datasets of actual patient CT scans. Tracked pinch gloves and stereo projection provide an immersive environment that allows collaborators to navigate, manipulate and view data from any perspective. Pointers, measurement tools, markers and devices for examining the data in magnified detail facilitate interaction between the viewer and the dataset. Internet2’s advanced network capabilities allow these interactions with the datasets to occur in real-time between physicians at distant sites using video and audio. Volumetric imaging provides intuitive and flexible visualizations for minimally invasive surgical planning and for education. The system frees physicians and students from having to mentally generate 3D representations from a series of two-dimensional images and offers views of anatomical structures that are not possible with other imaging modalities or by live or cadaveric dissection.

Additionally, high resolution datasets are reviewed in consultation from multiple sites including academic and community institutions.  High speed Internet, Internet2, and the Access Grid are options for providing the data, voice and video bandwith for this telecollaboration.

Role of Internet2:
Internet2 will allow us to distribute high resolution datasets in real-time to participating sites. In the past data was shipped on tapes or CD, which would allow sites to view data but not interactively manipulate it. Using Internet2 high-performance networks and the Access Grid to provide the data, voice and video—multiple sites can participate in the creation of a 3D model. After the model is created users can then load the model data from a networked file system and all view the same 3D information. Internet2 provides a high-speed, feature rich platform to launch emerging 3D technologies from the single lab to research facilities all over the country in a low latency environment conducive to medical collaboration.

TeleVolView: A Tele-Collaboration Application for Follow-up of Abdominal Aortic Aneurysm Patients

http://www.kitware.com
http://www.itk.org
http://www.vtk.org

Developed at:
KITWARE, Inc.
SINTEF Unimed

Demonstrators:
Bill Hoffman
Luis Ibanez

Contact:
Luis Ibanez
luis.ibanez@kitware.com
(518) 371-3971

Funded by:
NLM
NIH

Description:
Abdominal Aortic Aneurysm (AAA) is a common health disorder; in the U.S. it affects 6% of the population over 60 years old. Continuous monitoring of the aneurysm is fundamental for preventing rupture and therefore reducing mortality among AAA patients. KITWARE and SINTEF have jointly developed a research application for facilitating the monitoring of AAA patients from remote locations. This application provides medical image processing and state of the art visualization techniques for evaluating the risk of aneurysm rupture without requiring patients to be travel to distant clinical facilities.

TeleVolView supports collaboration sessions in which two users located in remote sites can establish a point-to-point connection and interact with the same dataset. Operations performed by one user are visible to the other user allowing for collaborative data exploration. This functionality was developed to allow regional hospitals to obtain image analysis support from experts located in central hospitals in remote cities. Patients that otherwise would have to travel hundreds of miles for scanning and evaluation can now remain in a regional hospital and only the resulting dataset travels to the central hospital to be processed with the assistance of an expert radiologist.

TeleVolView is a powerful volume visualization application that has been customized for medical applications. It has minimum hardware requirements and can be run on any desktop or laptop system. TeleVolView can be use to interactively visualize, annotate and measure volume datasets such as those produced by CT and MRI scanners. Methods from the Insight Toolkit (ITK) have been added as plugins enabling users to perform segmentation and registration of medical images in an interactive environment.

We will demonstrate TeleVolView remote collaboration sessions between the following sites: (1) a system at McCormick Place and a system in Norway, (2) between McCormick Place and Albany NY, and (3) between McCormick Place and the University of North Carolina at Chapel Hill.

Role of Internet2:
Internet2 is fundamental in this application for facilitating the transfer of large datasets between the two users involved in a collaboration session. A large bandwidth makes possible to transmit large datasets in a reasonable time. In this context, an expert in a central hospital will optimize the use of her/his time in actually interacting with a remote collaborator instead of having to wait for the time of data transfer over the network.

Using the commodity Internet it is difficult to integrate this type of collaborative session into the clinical workflow due to long data transmission times. The larger bandwidth of Internet2 will make it possible to work on several datasets, one after another, in a single session on a routine basis.