An experiment on the Advanced Communication Technology Satellite (ACTS) which seeks funding from the Advanced Research Projects Agency (ARPA) under the Communication Networking Program has the Ohio Supercomputer Center (OSC), Columbus, Ohio, partnering with the University of Hawaii (UH) and Georgetown University Medical Center (GUMC).
This research grant of $2.7M for two years plus one optional year is sought by Principal Investigator Dr. David Y. Y. Yun, Professor of Electrical and Computer Engineering at the UH, under ARPA's BAA 92-36. Participating partners include GUMC, OSC, Tripler Army Medical Center (TAMC, funded under Medical Diagnostic Information Systems), and Maui High Performance Computing Center (MHPCC, funded by the Air Force/DoD).
The project is a feasibility experiment on the delivery of medical expertise and services to remote, disadvantaged regions by combining the revolutionary technologies of ACTS communication and supercomputing with existing medical imaging capabilities. This application will create new medical services and a delivery mechanism that transcend both time and distance barriers.
When remote access of supercomputing is available via high speed portable connections, teleradiology and other medical imagery related services will be changed profoundly. A study of this impact is important, in that it may point to new innovative means of providing computer support to more aspects of medical care and alter the ways of delivering medical care without needing to transport patients, doctors, or equipment.
The use of volumetric imaging obtained by non-experts from the field for generating real-time, user-steerable 3D reconstructions to be used by remotely located experts in cooperative medical triage support applications will demonstrate:
- The feasibility of using NASA's ACTS for remote medical visualization. Because the resolution of imaging (sensor) technology and the number of sites entering the system can be varied, this application is inherently scalable. Initial tests can be performed on the Internet and final implementations should fully utilize the bandwidth of ACTS.
- The ability to provide remote expert consultation around the clock in critical medical situations and directly improve medical delivery to remote or disaster-stricken populations. This system will provide the mechanism to place needed expertise into devastated regions or in extreme remote locations, e.g. the Arctic and space stations.
- Develop new techniques for real-time, user-steered segmentation for image-based modeling, which provide the capacity for remote experts to more precisely investigate structural damage in disaster/battlefield casualties.
- Developments in rapid acquisition techniques of computed radiography (CR), CT, and mobile MRI images electronically, without needing films or chemical processing, suggest the use of supercomputed slice and volumetric models for remote consultation using high resolution images. Algorithms and simulation models experimented via ACTS may significantly impact future delivery of health/trauma care.
The overall project is composed of three phases requiring increasing progression of communication and computing sophistications:
- Teleradiology and Computer Aided Diagnosis: Plans call for 200-300 chest images to be transmitted daily from Hawaii to Georgetown for diagnosis. Image quality will be studied by comparing the accuracy of diagnosis in teleradiology environment and film environment. Images of various compression ratios will be compared. Physicians using this technology and service will be interviewed to determine how teleradiology service is affecting professional relationship, work habits and loads. Testing can be done on workstations in the field. This effort w ill be jointly carried out by GUMC and UH, remotely using supercomputers at OSC or MHPCC (Maui High Performance Computing Center) and workstations locally.
- Treatment Planning and Optimization: The primary aim of radiation therapy is to deliver a tumorcidal dose of radiation to a tumor volume while minimizing doses to surrounding, normal tissues. Normal tissue complications are the limiting factor in prescribing doses as high as required to achieve local control and to lower the chance of recurrence and metastasis. Conformal 3D dose distributions are best achieved using multiple beams of radiation pointed toward the target volume from several non-coplanar directions. To achieve this goal 3D dose calculations need to be performed and superimposed on CT/MRI anatomy model displays - 3D visualizations which require supercomputer performance.
When supported by supercomputer power and HDR communication, 3D radiation treatment and surgery planning should be reduced to less than one minute (i.e. less than 10 seconds per beam) while dose distribution and optimization are expected to be completed in a twenty-minute interactive planning session, by new algorithms for Monte Carlo simulation of radiation transport on a supercomputer.
- Image-based Medical Service The system will consist of medical imaging devices, volumetric image models, 3D image processing and visualization workstations, as well as initially one supercomputer at OSC and later augmented by Maui's massively parallel machine. Medical imaging data (possibly from multiple scanning devices or sites) will be transmitted by satellite to a central processing computer. Reconstructive modeling and predictive computations will be carried out by the supercomputer(s) and distributed to the operating sites (where the expertise is or the care is delivered) by satellite. Visualization of volumetric data will take place on the workstations. The user will be able to interactively manipulate the rendering of the visualization by changing parameters determining viewing, shading, lighting, etc. In addition, the interface will provide the capability for steering segmentation which is remotely computed on the supercomputer.
- Initial work would consist of the development of a real-time interactive volumetric renderer and interface developed at The Ohio State University Advanced Computing Center for Arts & Design and OSC running on a high performance workstation. The system provides the capability for manipulating reconstructions of volumetric data and an interface for real-time segmentation.
Eventually, volumetric data sets would be captured from TAMC in Hawaii and transferred via the ACTS to the OSC in Columbus. These data would be segmented on the supercomputer and resulting volumetric data sets would be sent via the ACTS to be rendered by local workstation at GUMC and simultaneously at Tripler in Hawaii.
The proposed testbed will utilize three geographically separate sites. GUMC in Washington, D.C., is located in the East Sector of the ACTS satellite. ACCAD/OSC, Columbus, Ohio, is located in the West Sector of the satellite. The third site, Tripler Army Hospital and the University of Hawaii, will require the use of steerable high gain antennas and high bandwidths found in ACTS.