Advancing Medical Imaging with IDL
by Jon Snyder
(Issue 10 · posted June 13, 1997)
Introduction
Non- and minimally invasive diagnostic
techniques are giving physicians improved
methods for assessing patient illness and
determining optimal treatment plans.
Computers powerful enough to process
these large data sets are now
cost-effective for radiologists, cardiologists,
general practitioners, and others using
ultrasonic, computed tomography (CT), or
magnetic resonance imagery (MRI).
Software application developers now have
powerful tools such as Research Systems'
Interactive Data Language (IDL) to
produce easy-to-use programs tailored for
specific medical imaging conditions. The
results are greater safety and accuracy in
objective and subjective clinical
observations as well as increased surgical
success.
Ultrasound for Assessing Arterial
Damage
Ultrasound imaging has always been the
safest way to investigate the human body.
The cost of ultrasonic imaging is relatively
low compared to CT and MR. In
combination with other testing, a full
assessment of a patient's progress and
health can be made reasonably accurately,
and without exposing the patient to
radiation inherent with X rays and
conventional CT scans.
Paritosh
Dhawale of
Technology
Solutions
Group, Ltd., has
adapted
ultrasonic
imaging to
provide 3-D
visualization and quantification of
atherosclerosis, a major cause of heart
attacks. Dhawale's IDL application, called
Ultrasound Quantitative Analysis Software,
delivers more accurate estimates than
X-ray imaging of the extent of
atherosclerotic damage.
To analyze the atherosclerotic buildup,
ultrasonic data is collected by a 1.2
mm-diameter transducer that is introduced
into the artery and maneuvered to the
point of blockage. Initially, the transducer
provides a 2-D cross section of the arterial
wall. 3-D data is produced by pulling the
sensor through the atherosclerotic
narrowing. The data is then processed and
visualized in IDL, allowing physicians to
assess the extent of arterial blockage and
help determine whether angioplasty is
required. To provide better insight,
Dhawale's application simultaneously
displays the image in multiple windows for
side-by-side comparison.
One reason that intracoronary ultrasound
analysis surpasses the accuracy of X-ray
analysis is that it doesn't rely on a dye
injected into the artery. Depending on the
angle of observation, physicians sometimes
have difficulty distinguishing arterial
plaque from dye in X-ray images. With
intracoronary analysis there is no dye to
cloud the picture, and data is collected at
the point of blockage, resulting in
high-resolution, highly reliable images.
Although inserting a catheter into the
artery is an invasive procedure, the
associated risk is low and the data is
gathered as quickly as possible.
Cardiologists read collected data, initiate
processing, and adjust image
characteristics through a graphical user
interface (GUI). Most importantly, the GUI
makes the system available and useful to
users with absolutely no programming
experience.
Initially Dhawale was skeptical about using
non-workstation platforms for data
analysis and graphics. However, after
running time tests of the application,
Dhawale learned that a "P5 166's
benchmarks were better than a standalone
Sparc-10, which costs three times as
much. Also, I find IDL code in Windows 95
to be stable, and the graphics performance
is excellent."
Dhawale's package runs on Power
Macintosh, Windows, and Unix computers,
and is available in five levels of
functionality to meet varying performance
needs.
Better Solution for Common Problems
In many life-threatening situations, CT and
MR imagery can give doctors a life-saving
edge. Abdominal aortic aneurysms are
distensions in the aortic artery that usually
form near the point where it meets the
iliac arteries. The bulge is painful, puts
pressure on any nearby organs, and may
rupture, which would eventually result in
death. For treating such aneurysms,
medical physicist Anne Martel of the
Queen's Medical Centre in Nottingham,
United Kingdom, developed an analysis
tool for radiologists and surgeons that
makes CT images useful for accurate
aneurysm measurements. From such
measurements a suitable graft is built to
replace or patch the distended aorta.
The old method of
analyzing abdominal
aortic aneurysms
involved selecting
oblique slices through
CT data and measuring
the distance between
two or three points on
those slices. According
to Martel, "This was very time-consuming.
Inadequate, if the vessels were tortuous."
In her application, the CT data is processed
and enhanced with IDL. A general-purpose
segmentation tool, implemented through
the point-and-click GUI, allows the
radiologist and surgeon to isolate the
abdominal aorta within the 3-D images.
Software-controlled functions like
thresholding, region growing, and
morphological operations are also
performed through the GUI. The bone is
masked and maximum intensity projection
(MIP) images are created. This allows the
doctor to see the aneurysm itself as well as
selected bones, blood vessels, and
surrounding organs. The physician can
visually assess the extent of the
aneurysm's effect and devise a plan for
effectively reaching and repairing it.
The radiologist defines curved slices
through the MIPs to display the entire
aorta. Then a line is defined down the
calculated center of the aorta and iliac
arteries. After this line is established, it is
possible to determine the distance between
any two points and obtain precise
measurements for building the graft. Since
the graft is made to fit perfectly, high-risk
patients spend less time under anesthesia
because the surgeon has an effective
strategy for repair. Surgeons benefit by
knowing before the patient is on the
operating table that the graft will solve the
problem.
Assessing Cancer Progress with MRIs
When Dr. Justin Smith of the First Hill
Diagnostic Center in Seattle, Washington,
wanted a better, easier way to look at MRI
data to determine how far cancer had
spread in a body, he found that none of
the available methods were sufficient. He
turned to Applied Research Technologies of
Portland, Oregon, for help. Ken Gindroz,
president of Applied Research
Technologies; Dr. Smith; and a consultant
from Research Systems planned and
developed a software application called
PRO-MRI. PRO-MRI is a GUI-controlled
application built in IDL that applies pattern
recognition methods to the high volume of
multivariate data generated by MR
imaging.
A crucial part of the system
was the user interface
design. "Most radiologists or
bio-imaging scientists don't
have the time or
background to use currently
available statistical
application software,"
Gindroz said. Therefore,
access to IDL's large data-set processing
power needed to be exceptionally intuitive.
Working with the physicians who were
going to use the application, analysis
options such as the ability to draw regions
of interest by hand were incorporated and
made easy to implement. To measure the
progress of a cancer's metastasis,
radiologists extract data for classification
from a series of aligned MR images and
compare the MR signature for each pixel or
a group of pixels to the MR signature of
known healthy and cancerous tissues.
Results are presented as a color overlay to
an image, or numerically for statistical
validation. Images can be displayed with
specific colors highlighting all the pixels or
groups of pixels with similar values. By
comparing the MR signature of a patient's
primary tumor with suspicious lesions,
cancer can be identified in other areas of
the body.
Summary
Ultrasound, computed tomography, and
magnetic resonance imaging each have
important roles in diagnostic imaging.
Many benefits are offered by more
insightful, convenient, and cost-effective
methods for understanding a patient's
condition without further jeopardizing
health with unnecessary surgery,
injections, or exposure to harmful
radiation. And with the data in a form that
can be transmitted around the world in
moments, physicians can share results and
insights with ease.
IDL (Interactive Data Language) from
Research Systems, Inc., is a
fourth-generation programming language
(4GL) used by scientists, researchers,
engineers, and software developers to
build data analysis and data visualization
applications. IDL includes a rich suite of
mathematics, statistics, graphics, image
processing, mapping, and general data
manipulation features in an integrated,
high-performance package. Because IDL's
4GL simplifies programming and does not
require the tedious edit-compile-link-debug
cycle required by traditional languages,
IDL lets users develop fully portable
applications much faster than with C/C++
or Fortran.
Purchasing information
IDL version 5.0 is available for the popular
operating systems: Windows 3.1x,
Windows 95, Windows NT, Macintosh,
Power Macintosh, Unix (Sun, HP, SGI,
Digital, and IBM), Linux, and Digital's
OpenVMS. Windows, Macintosh, and Linux
pricing starts at $1,500. Unix and
OpenVMS pricing starts at $3,495.
Research Systems, Inc., can be contacted
by e-mail at info@rsinc.com, by phone at
(303) 786-9900, or by mail at 2995
Wilderness Place, Boulder, CO, 80301.
Jon Snyder is a freelance
technology writer based in Boulder,
Colorado. He has published articles
in Recording magazine and is a
staff writer for Colorado Golf
magazine. He currently works in
the Marketing Department of
Research Systems, Inc.