SOFTWARE SOLUTIONS

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.
 

                



back to main page
back to page on brain imaging