Disease detection goes digital

November 04, 1996

I layed down on the picnic table, the doctor applied some anesthesia, and
he yanked my tonsils out."

That's how my grandfather described the procedure he and his classmates shared one day when the doctor came to their school. It happened a long time ago, but the story illustrates a point: The real impact of new medical technology tends to fall in the category of "what you don't do that you used to do."

Remember "exploratory surgery?" It's a procedure hardly used since the 1970s, thanks to the widespread application of computers in the clinical setting. Not only have Computerized Tomography (CT) and Magnetic Resonance Imaging (MRI) shifted the emphasis from traditional surgical techniques to non-invasive imaging procedures, they have allowed the early detection and more accurate treatment of a host of diseases.

And the trend continues. New technologies, now under development at General Electric Corporate Research and Development, Schenectady, NY, promise dramatic improvements in diagnostic health care for the 21st century.

Virtual endoscopy. Imaginea tiny camera, able to "fly through" the colon, aorta, or even the smallest of blood vessels in the brain. Continuously transmitting real-time images of the organ's or vessel's interior wall, such a camera could help locate aneurysms, determine the extent of cholesterol buildup, or calculate the size of a polyp. That's the goal of virtual endoscopy--an imaging technique designed to substitute CT and MRI for conventional endoscopes.

The idea, explains graphics engineer Bill Lorensen, involves a three-step process:

1) Construct a 3-D computer model of the body part in question. 2) Generate a camera path through the model. 3) Present the camera views as though taken from the inside of the organ or vessel, simulating what a doctor would see through a real endoscope.

Lorensen and his colleagues at Brigham and Women's Hospital, Boston, MA, built the 3-D model from a series of high-resolution, 2-D slices using CT or MRI data. Depending on the body part being modeled, they use one of two methods to generate the camera path. The first employs a computer animation technique called key framing.

Good for large cavities like those of the colon or abdomen, key framing makes use of the camera controls on a graphical user interface. With these, the operator moves the virtual camera to "key" locations in and/or around the organ being ex-plored. Once the keyframes are established, the system automatically calculates a smooth "flight path" through the points.

The second approach applies robot planning algorithms and is more appropriate for navigating the restricted space, for example, inside a blood vessel. With the path-planner method, a user specifies the camera's final goal, and the system labels all regions in the 3-D volume with its distance to the goal. Any locations in space occupied by "obstacles" are bypassed. Then, from a user-selected starting point, the system automatically generates a path to the goal.

Once calculations are complete, the user can start, stop, and step along the path; refer to an overview of the entire model with a pointer to indicate endoscope location; or call up the original CT or MRI slice with the endoscope position marked. The viewer can also operate the camera manually to pass through the organ wall and look at neighboring anatomy.

"Compared with real endoscopic views," says Lorensen, "the virtual endoscope offers a non-invasive procedure that requires no anesthesia or sedation." He adds that since models are constructed from CT or MRI images, it is possible to measure distances and size. "There may be times, for instance, when the surgeon will view a tumor or polyp, determine its size, and decide it's not necessary to take action."

Lorensen predicts the technique will see clinical use over the next five years. Further into the future, he believes virtual and real endoscopy will merge, giving doctors the benefits of both systems.

Digital mammography. Early detection remains the besthope in the war against breast cancer, a leading killer of women between the ages of 35 and 50. Very early detection, however, remains difficult since doctors are ultimately looking for the 50-micron trace of microcalcification a cancer cell leaves behind when it dies--a challenge for conventional film-based mammography. Adding to the challenge is the fact that film-based mammography offers poor contrast between glandular, or radiographically dense, breast tissue and lesions.

Full-field digital mammography, under development with funding from the National Cancer Institute, should soon help doctors overcome these barriers. Based on the thin-film technology of GE's flat-panel display research, the design features a seven-inch by nine-inch matrix of four million individually addressed pixels. The first system was delivered to Massachusetts General Hospital, Boston, in April; the first clinical images were taken in July.

Compared to traditional screen/film systems, digital mammography offers better contrast resolution, lower noise, and a linear, as well as larger, dynamic response. These advantages, Beale Opsahl-Ong, a GE physicist in the company's Detector Technologies Program, notes, "should allow for the earlier detection of breast cancer." In the not so distant future, he adds, the digitization of mammograms may permit use of knowledge-based, computer-aided diagnostic programs in applying artificial intelligence to the detection of suspicious lesions.

Digital ultrasound. While it's easier to detect a breast lesion with X-rays than with ultrasound, particularly because of the telltale calcifications, X-rays provide little information about the lesion. Ultrasound, since it interacts more with soft tissue than X-rays, offers the possibility of characterizing breast lesions.

The recent development of digital ultrasound machines, thanks to lower-cost ASICs and A/D converters, makes that possibility all the greater, says Tom Thomas, manager of the GE R&D Center's ultrasound programs. "Once you go digital," he points out, "all your signal processing is under your control in a way it never was with analog machines."

(For a description of one digital ultrasound system, see "Ultrasound tags breast tumors," Design News, June 10, 1996, page 74.)

Results, Thomas says, in-clude dramatically improved flow measurement and bet-ter overall image quality. For the physician, this means faster, more confident diagnosis. It also means more ultrasound applications.

"Presently, a woman with a breast lesion is generally subject to a biopsy procedure," Thomas explains. "This is not only invasive, but expensive. With digital ultrasound, it may be possible to determine if a lesion is cancerous or benign before submitting to a standard biopsy."

In addition to "ultrasound biopsies," Thomas predicts other advances as a result of digital ultrasound technology. These include measuring blood flow in the heart muscle, visualizing perfusion in the kidneys, and viewing lesions in the liver. Real-time, 3-D imaging, which will require advances in processing power and new algorithms for scanning ultrasound beams, he says, will help realize these goals.

Telemammography. In any decade, there is a technology that can be harvested to significantly impact the medical field. In the 1970s, it was the advent of computerized image processing. The '80s saw a dramatic expansion and com-mercial implementation of that capability. For the 1990s and into the next century, the emphasis will be on networking--taking medical information and presenting it broadly and inexpensively beyond a local work area.

The ability for anyone to look at anything anywhere presents a real opportunity for improving the quality and productivity of diagnosis. One example is teleradiology. With the development of digital mammography, it is now possible to take mammograms at remote locations and electronically transmit the information to a radiologist far away.

Today, mobile vans are beginning to make mammogram exams available to women in underserved and remote areas. The current practice uses film-based screening procedures, with the film images sent by courier to a hospital at the day's end for interpretation. Non-compliance, however, is a problem, especially in rural areas. Of all the women asked to return for a sec-ond exam, up to 30% don't, according to some published studies.

Results from telemammography, conversely, can be available within the half hour. With funding from the National Cancer Institute, GE Corporate Research and Development, in collaboration with Massachusetts General Hospital, has set up a telemammography testbed. The system is based on the Digital Imaging and Communications in Medicine (DICOM) standard, created by the American College of Radiology and the National Electrical Manufacturers Association.

Two Sun workstations--one located at GE Corporate R&D and one situated at Massachusetts General Hospital--make up the testbed. At each site, the workstation is cabled to a satellite modem which, in turn, is cabled to a roof antenna. Each satellite link runs at T-1 speed (1.544 Mbits/sec).

As of August this year, more than 58,000 digitized mam-mograms have been transmitted between the two sites, says GE computer scientist Kathryn Eike Dudding. Present analy-sis focuses on issues that can affect system throughput or image integrity, such as buffer size and weather.

Dudding says digital mammography promises greater accessibility to quality healthcare for the population as a whole. Additionally, it offers:

- Significant improvement in patient-physician follow-up.

- Higher image interpretation productivity.

- Better film storage and distribution for lower cost per exam.

As for the long-term future, she envisions a time when today's mobile vans will provide ultrasound and even traditional biopsy capability.

Telesurgery. Networking also implies telerobotics, and the possibility that someday doctors may perform remote surgical procedures. Perhaps they will use systems similar to one graphics scientists Ricardo Avila and Lisa Marie Sobierajski are developing.

The two GE researchers are creating a haptic interaction method for volume visualization applications. Used in conjunction with a force-feedback device and Silicon Graphics Indigo2 Extreme workstation, the software lets users feel, as well as see, virtual objects.

The PHANToMTM force-feedback device connects directly to the workstation through an EISA bus. Designed and marketed by SensAble Technologies Inc., Cambridge, MA, it incorporates motors, motion sensors, and mechanical linkage to convey the sense of touch.

Avila's and Sobierajski's software processes signals from the motion sensors on PHANToM's arm. The program then calculates the ap-propriate feedback force and instructs the system's three motors as to how much resistance should be applied to the stylus or thimble, which forms the operator interface.

Point contact forces are computed directly from volume data and are consistent with the isosurface and volume rendering methods. The result is a strong correspondence be-tween the image on the screen and the haptic device.

"Use of a force-feedback device during visualization," Avila says, "interactively conveys complex information to the operator. This is particularly useful when attempting to precisely locate a feature with-in a volume, or to understand the spatial arrangement of 3-D structures."

A second goal of the haptic interaction method is to usethe PHANToM as input tosimulate a virtual tool. With this type of tool, an operator could then feel and modify a volume-rendered object.

Example? Pre-surgical planning on a CT data set. Using the PHANToM stylus, a doctor can cut away skin or skull to reveal interior regions, "feel" the difference between healthy tissue and a tumor, or "poke" through an organ's surface to explore its inside wall.

In all instances, Sobierajski says, the sense of touch augments the visual input to supply additional information about the structure, location, and material properties of an object. The two scientists also argue that if presurgical planning is possible, why not telesurgery?

Such an operation would require a surgeon and hap-tic device at one end of the system, and a robot at the other. "This would be a good way to treat wounded soldiers on the battlefield, or perform emergency surgery on an astronaut," Avila reasons.

Implementation may be far in the future, but like the other systems described in this story, it is certainly within the realm of possibility, and not farfetched.


Focused ultrasound, guided by MR imaging, shows potential for non-invasive tumor ablation. Studies conducted by Harvey Cline, a physicist in General Electric Corporate Research and Development's MRI and Image-Guided Therapy Program, indicate fruition of the decades-old dream.

Despite initially promising experimental and clinical results, focused ultrasound has not seen development as a surgical tool for two reasons: 1) difficulty in controlling dosimetry of the focused beam, and 2) inability to accurately measure temperature rise within the targeted tissue.

"Cell death is very temperature sensitive," Cline explains. "The amount of heat delivered can be affected by tissue type, blood flow, and other physiological factors. Because MRI provides high-resolution images that are sensitive to the rise in temperature generated by focused ultrasound, the process is uniquely capable of helping guide and control ultrasound therapy."

A prototype system, designed for clinical use, integrates a computer-controlled hydraulic transducer positioner into the patient table of an MR scanner. Three pistons, equipped with optical encoders, move a spherical shell transducer in the X, Y, and Z planes. Filled with degassed water, the system connects to a workstation and pulse generator.

"Our goal," says Cline, "is to combine surgical planning with ultrasound therapy so we can use a computer to control surgery as one does with numerical machining." Clinical systems, he predicts, are still several years in the future.


- Rechargeable batteries to last 40 years

- Biocompatible force and position sensors

- Arrayed micro-electrodes

- New algorithms for scan beams

- New-generation A/D converters


The following are from an October 1945 article in National Geographic entitled "Your New World of Tomorrow," by F. Barrows Colton:

- "Blood plasma, and eventually blood substitutes, will reduce the toll of shock in accidents and injuries."

- "Sterilizing lamps, whose rays kill germs in the air, may be installed in hotel lobbies, railroad stations, and large auditoriums."

- "Air entering large buildings will be cleaned with the 'Precipitron,' developed by Westinghouse scientists. The device charges airborne particles of dust, smoke, and dirt with electricity, and then attracts them out of the air stream."


- Virtual endoscopy. Digital mammography. Digital ultrasound. Telemammography. Teleradiology. Telesurgery.

- "Virtual endoscopes will allow physicians to look inside the body without anesthesia or sedation."

--Bill Lorensen, graphics engineer,GE Corporate R&D

- "Telemammography will provide early warning of breast cancer to more women in underserved and remote areas."

--Kathryn Eike Dudding, GE computer scientist

- "Haptic devices will become a routine part of pre-surgical planning."

--Ricardo Avila, GE research scientist

- "Ultrasound therapy will allow non-invasive tumor treatment."

--Harvey Cline, physicist,GE Corporate R&D

- "Electrical stimulation will augment neural regeneration for functional restoration after spinal-cord injury."

--Hunter Peckham, director,Cleveland FES Center


Emerging technologies may bring early warning of illness and disease, but they can't prevent accidents. Spinal injury, for instance, paralyzes some 12,000 people in the U.S. every year.

Restoring movement to paralyzed limbs by means of electrical stimulation has been a research goal for more than 30 years, involving input from surgeons; therapists; electrical, mechanical, and biomechanical engineers; as well as computer programmers. Despite this multidisciplinary attention, there have been no technology breakthroughs on the scale of CT or MRI. Progress, instead, is incremental, relying on the slow build-up of a knowledge base.

The challenge can be better appreciated when considering an implantable "grasp-and-release" hand system, developed by the Cleveland FES (Functional Electrical Stimulation) Center with primary support from the Department of Veterans Affairs Rehabilitation Research and Development Service and the National Institute of Health Neural Prosthesis Program.

The neuroprosthesis, now in its second generation, is intended to aid people who have had a spinal-cord injury resulting in C5- or C6-level impairment. This means they have use of their shoulder, upper arm, and elbow, but may not have use of their wrist, and have no use of their hands.

How it works. Chief components include an implantable stimulator/telemeter, an implantable joint-angle transducer, and external controller. The 10-channel stimulator generates the constant current pulses used to excite the muscle via electrodes; the stimulator's ASIC regulates pulse width, amplitude, and frequency of each signal. Telemetry allows feedback from the joint-angle transducer.

How much stimulus needs to be delivered? "An occupational therapist profiles each patient's muscles to determine input/output transfer functions," says Senior Engineer Jim Buckett, noting that surgical reconstruction is often required prior to electrode implantation. "We input these parameters, and vary the pulse width, amplitude, and frequency to determine the reaction. This is done for each muscle individually; we then coordinate the muscles groups for programming the ASIC."

Crosstalk and sensitivity problems, Buckett adds, make the above task time-consuming and tedious. He speculates that "perhaps in the future we could take a person's hand, stick it into an instrumented, gel-filled bag, and program a computer to iteratively determine these parameters." Before then, however, there are more immediate needs that Buckett addresses:

- Better batteries. Power is currently supplied from outside the body, which is not efficient. "We need a rechargeable battery that will last 40 years."

- Improved force and position sensors. The technical challenges here are size and bio-compatibility. "The body is a very harsh environment to put something into, but at the same time, it's an environment that is very sensitive to foreign objects."

- Miniature multiple-lead configurations. "We can build the electronics with 30 or 40 leads, but to go in and surgically install a system of that complexity is very difficult."

- Arrayed micro-electrodes. Fabricated with integrated circuit technology to contain their own pulse-generating capability, these would "conserve space and weight considerations."

- Implantable intelligence. "The controller needs to grow functionally, but shrink physically."

Converging technologies. As engineers solve the challenges above, they must look beyond single-function systems, says Hunter Peckham, director of Cleveland's FES Center, adding that "hardware complexity within the body will increase substantially with each additional implant, presenting engineers with new design problems."

Peckham also envisions the ability to extract command control information from the intact cortex. Telemeterizing this information to a grasp-and-release stimulator, for example, would eliminate the shoulder movement necessary to trigger the present-generation hand-grasp system. "It may be that 50 years into the future, neural regeneration will be used to help restore some function after spinal-cord injury," Peckham reasons. "Electrical stimulation would then be used to augment the full restoration of functionality--just as we treat diseases today with multiple approaches."


"Increased clinical use of Functional Electrical Stimulation will require engineers to design miniature; implantable force and position sensors, arrayed micro-electrodes with on-board intelligence; and longer-life rechargeable batteries."

--Jim Buckett, senior engineer, FES Center

Comments (0)

Please log in or to post comments.
  • Oldest First
  • Newest First
Loading Comments...