Physiologic Imaging
NLM Teleradiology Project Goals
Provide education on volumetric imaging and analysis
through multimedia tutorials to rural physicians and
technologists
The first step in implementation of this project is the need to educate
both primary care and regional center personnel on the aspects and
benefits of state-of-the-art multimodality image processing and analysis.
We will use HTML and the World Wide Web. Optimal utilization of the
strengths of volumetric imaging requires an informed rural medical
personnel. As the rural physicians and technologists become more familiar
with the utility of existing volumetric imaging methods in their practice
via the multimedia system, they will be able to contribute to the breadth
of examples given so as to appropriately serve the more common needs. The
multimedia presentations will include:
Basics of volumetric image
formation, display and analysis
- Information on the types of
analysis applicable to particular clinical conditions
- Scanning protocols for various
scanning modalities to assess particular clinical conditions
- Means of identifying the nearest
and most applicable scanner facility capable of a desired evaluation
- In all of these objectives, although the long range goal of this
project is to provide all of the above services for the full breadth of
clinical questions, in this project period we will concentrate on
cardiac, pulmonary, and abdominal imaging protocols with images gathered
via CT, ultrasound, and conventional X-ray projection imaging. The latter
will serve as a means for comparison with the full volumetric analysis.
We will involve rural physicians in prioritizing the order with which we
bring volumetric analysis services (on-line interactive consultations)
on-line in a user friendly environment, and we will provide information
to other institutions, through publications in peer reviewed journals,
regarding the above activities. The information will be accessible over
the statewide, fiber-optic Iowa Communications Network. The initial
three remote sites will be:
- Davenport
- Ottumwa
- Keosauqua
Install hardware and software in rural imaging centers for
improved and reproducible data acquisition: specifically for volumetric
CT scanning and image display and analysis.
Because systems capable of acquiring volumetric image data sets in a
single breath hold are only available at a very limited number of sites,
we propose to make available inexpensive
hardware solutions which allow for high quality volume acquisitions
on more commonly available CT scanners in multiple breath holds. Such
methods will significantly improve the image quality and thus utility of
volumetric (multiple contiguous slices) data involving both the thoracic
cavity as well as the abdomen. Particular areas of importance are
pulmonary structure and function, mediastinal structures, and
organ-to-tumor volumes.
Provide real time interactive image analysis consultation
for volumetric data sets acquired from rural centers.
Rural centers that do not have the capability to analyze their image data
can obtain expert consultation on their images with the University of
Iowa via the Iowa Fiber optic network. Trained technologists, with a team
of physicians and members of Dr. Hoffman's laboratory, will be made
available to provide consultation. Common equipment at all sites will
make possible the initial goal of on-line interactive consultation.
Furthermore, inexpensive solutions will also allow for the eventual
participation by the regional referral sites.
Develop and/or modify a clinically useful workstation and software
that integrates image diagnostic display capabilities with sophisticated
image analysis tools featuring distributed network-based real-time
interactive consultation.
The majority of currently used clinical display systems only provide
straightforward image display and primitive image processing. In
addition, these teleradiology systems are based on a point to point
architecture where only one way ("fax-like") communication is possible.
Our proposed workstation and network will provide both high performance
diagnostic image display functions already developed by Dr. Chang and
sophisticated image analysis tools. Software modules developed by our
laboratories are being extensively used by basic science and clinical
researchers throughout the US and abroad. However, there is a strong need
to develop friendlier and more intuitive user interfaces to facilitate
these modules being used by non-research oriented clinicians and
technologists. Such software also needs to be supported on a broad range
of computer platforms to increase the software's accessibility. Providing
software support only is not sufficient for successful clinical use of
our system; easily accessible interactive consultation with experts is
critical. Therefore, our system will also provide real-time interactive
consultation via the network amongst multiple users.
Workstations designed for the primary care health provider will
concentrate on providing high performance, flexible diagnostic image
display with interactive distributed network-based consultation
capability. Cost effectiveness will be a major consideration. The images
will be displayed on a high resolution IMLOGIX monitor with 2K x 3K
memory (enabling the storage and manipulation of 2K x 2K pixel images)
and a 1024 x 1024 pixel, 12 bits/pixel display and fed to the IMLOGIX
system via a Sparc station using a consultation module built for VIDA
which allows for down loading images from the VIDA shared memory to the
IMLOGIX box.. The IMLOGIX monitors feature an extremely stable phosphor
and offer a very simple user interface with basic but powerful functions
such as window/level, real-time bilinear interpolated zoom, and pan. A
separate high resolution 19 inch color monitor will be used for the
display of a "user friendly" graphic user interface runing in the VIDA
environment.
The image analysis software integration into the teleradiology work
station will be based upon our current implementation of VIDA. VIDA is a
multi-process system written in C and developed on Sun Microsystems
workstations. It is built on top of the X Window System, Version 11, and
currently uses the XView toolkit to provide an Open Look Graphical User
Interface. (this will be converted to the emerging new standard which
incorporates unique features of the OpenLook environment, such as push
pins and drag and drop capabilities, into Motif.) VIDA's system
architecture uses a model similar to hub and spokes of a wheel. A central
kernel comprising shared data storage and interprocess communication acts
as data manager and mediates communication among the separate VIDA
processes.The shared memory structure is designed to allow for multiple
multidimensional data sets (objects) to be manipulated simultaneously. It
consists of a single master segment and a number of data segments. The
master segment contains information about each process attached to the
VIDA kernel as well as an area for token-passing among processes. The
master segment also contains information about available colormaps and
their associated color cells. Key and size information is maintained
about each of the subsidiary data segments. The master segment also
stores the total number of items in shared memory, the object names and
types, a list of process numbers to be notified when the shared memory
changes its status, the segment number, offset and length of each object,
a semaphore to provide exclusive access for writing, and a time stamp of
last modification.
Multiple data sets can reside in a single shared memory data segment.
Data sets are organized by five mutually orthogonal dimensions. Three are
used for spatial information (x,y,z) and two for temporal information
(time, phase). Each shared memory object begins with a header which
specifies the data dimensions, voxel size, imaging plane, image data
location, and size within the data segment.
Functions are provided to user programs (VIDA modules) for data
insertion, access, and removal. Interprocess communication is achieved
using software signals. Interested processes are notified when a data
object changes status (written, moved, deleted) and when shared memory
contents change. A token can be exchanged among modules via an area in
the master segment.VIDA is configurable by the end-user and can
incorporate user-developed applications.
Provide technical assistance to rural centers that wish to analyze
images on site
We will advise centers on computer hardware and software requirements
needed for on-site image visualization and analysis. Also, we will make
our software available in a cost effective manner. Through this funding
mechanism, we will provide two of the regional referral test sites
(Ottumwa and Davenport) with appropriate hardware and software. We will
train and support local physicians and technologists through on-site
visits and multimedia tutorials (objective 1).
Assure that the final system will easily allow for incorporation of
advances in imaging and image analysis capabilities. In particular, we
are targeting two evolving automated image analysis methods to test this
capability. These tasks are expected to significantly aid faster,
objective image visualization and analysis.
We do not want the deliverables of this proposal to represent merely a
feasibility study. We wish to lay the ground work whereby our research
activities as well as the research activities in similar laboratories
around the country can more readily find avenues of directly impacting
health care in America in a more wide spread, homogeneous fashion.
Evaluation and Validation.
- "To call in the statistician after the experiment is done may be no
more than asking him (or her) to
perform a postmortem examination: he (she) may be able to say what the
experiment died of."
- ---Sir R.A. Fisher
As we develop new medical technology, we must evaluate its performance
to learn from the enterprise. Technology may be evaluated at several
levels depending in part upon its level of development. In early stages
of development, the most important question is whether and how well the
technology meets its most fundamental requirements. For our project on
the remote use of advanced physiological and three-dimensional image
processing, we will attempt to verify that it is technically feasible.
Images must be acquired at the remote site and transmitted to the
University of Iowa for processing. Important clinical information must be
returned back to the referring site in a timely way. Furthermore, there
needs to be communication between both the primary site and the rural
referral center as well as between the tertiary site and the other two
locations. Can each of these technical requirements be met? Next, a more
detailed understanding of the monetary and social cost versus benefits of
this process would help to determine whether further studies of efficacy
should be pursued. While ROC and paired comparisons tests along with
outcome analysis will become an important part of our evaluation in out
years of the project, initially, our test instruments will be more
descriptive in nature.
Sources for Topic Overviews:
Hoffman, E.A. and J.D. Hoford: Tool box-based cardiac volumes:
visualization and quantitation by computed tomography. American Journal
of Cardiac Imaging 7(3): 164-178, 1993.
Chang, P.J., "MR multiband viewing workstation prototype," Radiology 185:
416, 1992.
Hoffman, E.A., D. Gnanaprakasam, K. B. Gupta, J. D. Hoford, S. D.
Kugelmass, and R. S. Kulawiec. "VIDA: An environment for multidimensional
image display and analysis," Proceedings SPIE 1660: 694-711, 1992.
Hoffman, E.A. W B. Gefter and J. Venegas: Frontier Pulmonary Imaging. In:
A.P. Fishman: Update: Pulmonary Diseases and Disorders, Second Edition
McGraw-Hill, Inc., 323-340, 1992.
Kalender, W.A., R. Reinmuller, H. Fichte, J. Behr, T. Beinert, W.
Seissler, et al., "Spirometrically gated CT measurement of lung density
and structure," Radiology, 185: 354, 1992.
Hoffman, E.A., "A historical perspective of heart and lung 3D imaging,"
in 3D Imaging In Medicine, Eds: J.K. Udupa and G.T. Herman, pp. 285-311,
CRC Press, Boco Raton, 1991.
©1994-99 Division of Physiologic Imaging, Dept.
of Radiology, Univ. of Iowa
NLM |
DPI Homepage |
VIDA |
Contact Us |
Search
References for Topic Overviews