As Demby cut through the defense to score, wiped sweat from his brow, got "five" from a teammate, the viewer saw a man not only restored to health but revitalized as an athlete.

What you did not see on television was the full-time therapist who traveled with Demby---who became a national spokesman for the disabled---to adjust and readjust the fit of his artificial legs, says biomedical engineer Joan Sanders.

"The size and shape of an amputee's limb change during the course of a day," says Sanders, Ph.D., an associate professor at the University of Washington in Seattle. "So the fit of a prosthesis also changes, affecting its comfort and performance."

A poor design that improperly distributes stress may cause pain or discomfort or even serious complications. Until recently prosthetics was more of an art form than a medical science. "A protractor and a plumb bob were the only quantitative tools they had before biomedical engineers came along," Sanders says.

As a result, Sanders and other biomedical engineers are for the first time measuring the pressures and shear stresses that occur at the point where an artificial leg meets the human limb. These measurements are expected to provide a basis for estimating the optimum size and shape of a prosthesis for a made-to-order fit.

Currently, interface stresses are assessed indirectly while fitting the prosthesis to the limb. A temporary socket is made of clear plastic so a clinician can view skin color while the patient leans his or her weight into the device. Skin blanching indicates excessive pressure.

Then as the full prosthesis is being fitted, the patient takes a walk while the clinician watches for irregularities in the patient's stride. Afterward, the clinician looks for skin irritation at the end of the limb. These visual examinations require extensive clinical experience.

"An experienced clinician notices slight gait modifications that occur because of improper interface mechanics, but a less experienced clinician might not pick up the subtle indicators," Sanders says.

On the other hand, objective measurements of interface stresses would lead to more accurate assessments and would allow the comparison of different socket designs, alignments or other fitting considerations. "Mathematical models could be used to estimate interface stress distributions or to design socket shapes that achieve specified interface stress distributions even before a prosthesis is first worn," Sanders says. There are currently no models that clinicians can use in this way.

Sanders and her colleagues are using finite element modeling to study the geometry of the residual limb and the prosthetic socket into which it will fit. The shape of the limb is divided into small, regularly shaped polygons, or elements. Properties such as mechanical stiffness are assigned to each element. Then external forces are added to the model boundaries to simulate standing or walking.

Using this approach, Sanders' group has solved three of the four major problems that must be resolved before the model can be used to predict the best shape and design for an artificial limb.

The first problem was to describe the point of contact between the artificial and natural limbs. The two surfaces can slip back and forth by 1 centimeter, making it difficult to write a mathematical description of the point of contact. Secondly, the size of the socket is smaller than that of the limb, which is necessary to keep the socket from falling off. The size difference presented a series of modeling problems.

The third problem was finding a way to accurately measure the shape of the natural limb. The group developed an optical scanner that records 17 silhouette images of the limb in rapid succession. The images are combined to make a three-dimensional picture.

The remaining problem is to develop a model description of the material properties of the skin and muscle. Sanders' group is borrowing from extensive published descriptions of skin and muscle that have been compiled by biomedical engineers over the past 30 years. "Once this remaining modeling problem is solved, and we're close to solving it, the next thing is to do sensitivity analysis," Sanders says. "Do some generic testing. If you change this, what happens? If you change that, what happens?"

The result of this research will be a mathematical model that is expected to suggest better ways of designing prostheses to reduce stress, shear forces, and the resulting irritation and ill effects to the patient's skin. This may include different ways of designing and shaping the pockets into which the limbs fit and different choices of materials for the interface between the limb and the prosthesis. There is a question of whether to try for standard sizes or to make each one a custom device. If customization is the answer, then Sanders' model could be integrated with standard computer-aided design software now used for manufacturing prosthetics.

Sanders' research will also apply to a broader class of patients, all of whom are troubled by skin contact with a surface. These include patients who require custom shoes, those with diabetes and foot problems, and wheelchair patients who are confined to a seat for long periods of time.