Engineers Strive for Partnership between Robots and People

Medical Device & Diagnostic Industry Magazine | MDDI Article Index

An MD&DI April 1997 Column

R&D HORIZONS

Robotics is advancing faster than the ability of health-care personnel to accept it. For the time being, commercialization will require adapting the technology to this reality.

Mechanical assistants as exact and precise as they are indefatigable stir the imagination of surgeons and technologists alike, conjuring visions of superhuman capabilities. But they also send a chill down the collective spine of the health-care profession. Their raison d'être--the precision and persistence with which they perform tasks--targets the inherent failings of humanity. Robots can be programmed to filter out the tremor in a surgeon's hand or to hold an instrument indefinitely and exactly in one spot. They can aliquot samples with greater speed and precision than the best-trained technologist or even stand in for a human technologist, allowing staff at a central laboratory to conduct bedside analyses in an ICU.

Robots can be programmed to shuttle specimens around in a clinical laboratory. Photo courtesy of California Computer Research (Lake Arrowhead, CA).

In fiction, robots have been portrayed as both willing servants and as destroyers run amok. Medical robots have the potential to be both. The benefit to be derived from their use must be balanced against their potential for harm. And that potential for harm is greatest in the surgical suite, where a robot has the most direct and immediate impact on the patient. Typically, fail-safes are developed to prevent surgical robots from poking into dangerous regions of the body, unless expressly directed to do so. "Robots can be assigned to do the noncritical parts of a procedure, but these devices will still insist that the surgeon have his hands on when safety-critical margins are involved," says Russell H. Taylor, professor of computer science at Baltimore's Johns Hopkins University.

The comfort factor involved in using robots and the accompanying risk assessment built into their designs have led engineers to develop robots that complement human strengths and compensate for human limitations. "There are things that robots can do that people cannot do well, and you are trying to make a partnership between the robot and the person--to use their complementary skills to do something that neither can do very well alone," Taylor says. These robotic assistants are being designed to provide cost-efficient, stable control over surgical tools.

ROBOTIC PIONEERS

A pioneer in medical robotics, Robodoc, made history on November 7, 1992, when it became the first robotic device to perform an invasive surgical procedure based directly on input from a computerized preoperative planning system. Robodoc leverages the extraordinary geometric accuracy possible with robots to precisely mill joints in preparation for cementless total hip replacement. An exact fit is critical because cementless prostheses succeed when bone grows into the porous coating of the implant. "People have done cementless hip replacements manually for many years," says Taylor, who helped develop Robodoc while at the IBM T. J. Watson Research Center (Yorktown Heights, NY). "The advantage of the robot is that it is much more accurate."

The precision possible with robots led to their early adoption in the clinical laboratory. The robots now available, such as the one built into the Gilson M215, an automated pipetting device, are relatively simple and designed to perform mundane tasks. "From our standpoint, a robot is simply a device that moves a probe in the x, y, z planes," says Gordy Hunter, director of marketing at Gilson, Inc. (Middleton, WI). "If you associate a pump with that arm, you can either suck up a liquid or dispense a liquid."

The laparoscopic-assisted robot system, under research at Johns Hopkins University, can be used as a steady-hand manipulator.

The near-term opportunity is to daisy chain technologies to perform a series of tasks with minimal effort by technologists and nurses. Robotic interfaces appear to be the key to such daisy chaining. The utility of the Gilson M215 has been expanded to help perform other tasks within the clinical laboratory, by interfacing the system to a gripper robot arm that picks up individual tubes or whole racks of tubes and places them in other instruments. With only minimal changes, the M215 has been adapted to help prepare urine specimens for analysis on a Hitachi 717 analyzer, perform preanalytical aliquoting of specimens into prelabeled secondary daughter tubes or specimen cups, and conduct postanalytical aliquoting of serum from Vacutainers into specimen-storage containers.

The work was done by researchers at the Medical Automation Research Center (MARC), established in 1995 at the University of Virginia Health Sciences Center (Charlottesville). The staff, led by Robin A. Felder, PhD, director of MARC and a professor of pathology, are now trying to adapt a robot arm made by CRS Robotics Corp. (Burlington, Ontario) for use on an Abbott Laboratories immunoassay analyzer. The arm picks samples off a conveyor belt, loads the analyzer and then unloads it, placing samples back on the belt. "Most laboratory instruments are not designed to work hand in hand with another piece of automation," says Denis Ferkany, CRS Robotics' business development manager for clinical and pharmaceutical markets. OEMs are redesigning their analyzers for the robot world, Felder notes, "but that takes time."

IMMINENT ADVANCES

Interfaces for existing equipment will provide the interim solutions the medical community will see in the near term. In June, Computer Motion (Santa Barbara, CA) plans to unveil a system for automating the surgical suite. Part of its Hermes project, OR 2000 will provide voice control over virtually all of the peripheral equipment in an operating theater. "Imagine a surgeon commanding the lights to go on or the patient table to go to the Trendelenburg position," says Yulan Wang, chief technical officer and executive vp at Computer Motion who helped found the company in 1989. "We're taking our computer, voice, and feedback technology and making the operating room intelligent so the surgeon can interact with basically all the equipment in the room."

Computer Motion achieved these core competencies through the development of surgical robots. The AESOP (Automated Endoscopic System for Optimal Positioning) 1000, which entered the American marketplace in October 1994, is designed to hold and move a lapa-roscopic camera and light source dur-ing minimally invasive procedures. The robot arm moves in accordance with joystick or foot pedal commands. The system has assisted in more than 18,000 operations at more than 150 surgical centers and hospitals around the world. In May 1996, the company released a new version of the surgical robot, the AESOP 2000, which is distinguished primarily by its ability to respond to voice commands. "Instead of controlling AESOP using a hand or foot control, the surgeon merely speaks to the robot and the robot instantly obeys," says Gene Wang, CEO of Computer Motion. The robot goes into command mode when the surgeon says "AESOP." Directional commands, such as "left" or "right" and "move in" or "move out," instruct the robot arm.

Coming after AESOP is Zeus, which Computer Motion designed specifically to assist in microsurgical procedures. "We believe Zeus will do for surgeons' hands what the microscope did for their eyes," says Yulan Wang. "As you start working on smaller and more delicate structures, pretty soon the human hand with conventional instruments is not sufficient and the surgical instrument has to be enhanced--and the way to do that is to computerize it."

A surgeon using Zeus grasps two control arms that translate movement to miniature end effectors inside the patient's body. The surgeon watches the robotic hands via a television monitor. Sweeping motions by the surgeon are transformed into submillimeter actions by the robot arm. This translational control allows the system to damp out hand tremors, which research indicates occur naturally at a frequency of about 7 to 10 Hz. "When working with extremely small blood vessels, every surgeon's hand shakes somewhat," says Michel Gagner, a surgeon in the Minimally Invasive Surgery Center of the Cleveland Clinic, one of three sites in the United States where Zeus is now being clinically tested. Zeus promises to improve dexterity so much that cardiac surgeons may be able to complete endoscopic multiple vessel coronary artery bypass grafting through an incision less than an inch long. "As we work with smaller incisions and smaller vessels, we test the limits of human physical capabilities and require the assistance of new technology," says Joseph F. Hahn, MD, chairman of the Cleveland Clinic Division of Surgery.

RESISTANCE TO CHANGE

But not all new technologies are embraced. The engineering lab at Computer Motion is developing leading-edge robotic controls. One tracks eye movements in much the same way targeting systems designed for use by fighter pilots target their weapons on air and ground targets. Another registers head nods forward, back, and sideways to direct the robot. But the time is not yet right for such technologies, says Yulan Wang. "In selling a product, you need the economic drivers and you need to find the reason for people to pay for it," Yulan Wang says. And there are other obstacles to the acceptance of medical robots.

Robodoc, arguably one of the most advanced surgical technologies in the world, is available only in Europe. Integrated Surgical Systems, Inc. (Sacramento, CA), has not yet submitted a request to FDA for marketing approval in this country, even though some five years have passed since the system's first clinical use in hip replacement. The reason has to do with FDA. The landmark surgery in 1992 was the first in a 10-patient pilot study authorized by FDA to determine device safety. The agency then requested a larger multicenter trial, which began in September 1993. "We had hoped FDA would ask for a short follow-up, like six weeks," says Brent Mittelstadt, director of surgical applications. Instead FDA required a two-year follow-up. The data from that trial have now been gathered and should be submitted to FDA by the end of this year, Mittelstadt says.

Computer Motion's Zeus allows surgeons to work with incisions and vessels much smaller than the human hand can manipulate.

These early efforts to develop robotic assistants, such as Robodoc, are only the first steps en route to grander concepts now taking shape. Some of the most futuristic work is going on in academia. Taylor, who assisted in the early development of Robodoc while working for IBM, has since joined the faculty of Johns Hopkins University. "I came here so I could be more closely coupled with end-users, who in this case are the surgeons," he says. "So, in a sense, I bet my career that these technologies will have a very high clinical impact."

Taylor is currently developing a robot that maneuvers a laparoscopic camera under joystick control. The robot, called LARS (laparoscopic-assisted robot system), also has the ability to position the camera--or surgical instrument--autonomously on a target designated by the surgeon in video images. "The advantage of LARS is that you can point at something on the screen by putting crosshairs on it, and the computer will command the robot to point the scope there," Taylor says.

Taylor is also using a robot to assist in placing radioactive pellets into abdominal organs, such as the liver and kidneys. To be most effective, these pellets must be implanted as close to a tumor as possible. The problem is that soft tissue, especially in the abdomen, is constantly in motion due to respiration. Keying on a three-dimensional model of the liver created with CT scans, the robot can place landmarks near the tumor. Then, under x-ray fluoroscopic guidance, the robot can very precisely implant the radioactive pellets. "In the time of a single breath hold, the robot corrects the alignment and injects a row of radioactive seeds," Taylor explains. "So, in real time, we're analyzing the fluoroscopic images to adjust the aim of the robot."

LEADING-EDGE DEVELOPMENTS

Similarly, efforts to develop robotic laboratory instruments are aiming much higher than the relatively simple systems now in commercial use. The University of Virginia has created a hybrid of corporate and academic R&D environments as part of MARC. With funding from industry, the university provides a test bed for the development of some of the best and brightest ideas in medical robotics. Technologies that evolve in the center are put to the test in the Virginia Health Sciences Center.

One of the premier projects at MARC is the Remote Automated Laboratory System (RALS), which allows technologists in a central lab to operate clinical analyzers near the patient bedside. There, robots run analyses of blood gases, pH, electrolytes, glucose, and hemoglobin after receiving blood drawn from the patient. "It's often more cost-effective to have the analytical process near where you need it, for instance near the patient bedside," says MARC director Felder. The remote lab is situated for easy access by the support staff on a hospital ward, such as an ICU. "RALS is very much like telemedicine, but this is called a telelaboratory," he says.

Point-of-care testing promises rapid turnaround of specimens. Additionally, the precision possible with a robotic system offers a potential reduction in laboratory errors and repeats. The medical device industry has already begun to embrace the concept. Several companies, including Medical Automation Systems (Charlottesville, VA), Chiron Diagnostics (Norwood, MA), Mallinckrodt Group (St. Louis), and Nova Biomedical (Waltham, MA), have released commercial versions of remote laboratory systems.

Only Felder and his group have taken the system to the ultimate level of sophistication, integrating a robot arm that prepared the sample for input to the instrument. Tied into the analysis system, the arm, which was provided by CRS Robotics, allowed rapid entry of multiple specimens as well as unattended operation. It could perform simple pick-and-place operations on syringes containing whole blood and was trained to remove and then replace the cap of the syringe, mix the specimen, and even remove air bubbles. Yet Felder and his colleagues dismantled the arm. "The robot was too far ahead of its time," Felder says. "People didn't want all that functionality. It was scary. It was just too new." Samples are now prepared and placed in RALS by a nurse, he says.

There are other uses for robots, some of which have characteristics strikingly similar to those of people. Among them is RoboCart, a 4-ft, 85-lb mobile robot developed by California Computer Research (Lake Arrowhead, CA) and tested in Felder's Virginia proving ground. Up to 20% of technologists' time is spent transporting medical specimens around the lab. RoboCarts equipped with collision avoidance sensors can be programmed to shuttle specimens from station to station in a clinical laboratory, following special tape pasted to the floor.

These mobile robots, which are priced around $28,000 each, perform the mundane tasks that technologists hate, completing tirelessly and without complaint more than 90 circuits around the University of Virginia clinical lab in a single workday. RoboCarts have evolved to perform such higher functions as speech. "It's the friendliest person in the lab," says Don Nagy, CEO at California Computer Research. "It keeps saying thank you." Despite such abilities and good manners, these robots have not been widely adopted. They have found homes in only about a half-dozen medical centers across the United States.

To be more widely accepted, the robots will need increased functionality, such as the ability to load and unload deliveries without human assistance. Additionally, they may need to be more responsive. Engineers at the company are currently looking into building a telephone into RoboCart so that lab staff could call for assistance. Another possibility is to remotely control RoboCart.

Nagy has already proven the concept with a satellite link, used to command a RoboCart in Tokyo from the Lake Arrowhead laboratory. Such satellite connections are too expensive, but using the Internet would get around the cost. Nagy envisions a time when laboratories are operated 24 hours a day by technologists working either from home or from workstations halfway around the world. "People perform best from 8 a.m. to 5 p.m.," Nagy says. "From midnight to 7 a.m., their performance is degraded. But 8 to 5 occurs some place in the world at all times. Rather than driving to the lab, they'll sit in front of their computers. It's nothing magical. We're just about doing it now."

Greg Freiherr is a contributing editor to MD&DI.

Copyright © 1997 Medical Device & Diagnostic Industry