Millions of diabetics prick their finger for a drop of blood a few times a day to check glucose levels. Besides being uncomfortable, these tests can miss sudden dips or bursts in blood sugar. Frequent readings are easier with sensors that can be implanted in a patient's skin. But the glucose sensors available today can cause infections after a few days, and they are bulky and expensive.

Researchers led by Gerald Loeb, a biomedical-engineering professor at the University of Southern California, are now working on a glucose-sensor design based on optical technology. The design shows promise for making sensitive, affordable, and less invasive sensors.

The technique involves measuring the change in fluorescent emissions that occurs when glucose binds to certain molecules. The sensor is a tiny optical fiber that could be implanted in a patient's skin. To read glucose concentrations, a portable analyzer will shine ultraviolet light into the free end of the fiber and measure the fluorescence, says Loeb.

Attached to the end of the fiber inside the skin is a polyethylene-glycol polymer matrix interspersed with pairs of tightly bound chemicals, each tagged with a different fluorescent molecule. Under ultraviolet light, the bound molecules shine at one wavelength. When the researchers place the matrix in a glucose solution, glucose molecules knock out and replace one of the chemicals, dextran. As a result, the chemical complex starts emitting at two different wavelengths. The ratio of the fluorescence intensities at the two wavelengths is in proportion to the glucose concentration.

According to Loeb, the sensor should be cheap and disposable. "Essentially, it's a dot of polymerized goop on the end of an optical fiber," he says. "A few-centimeters-long optical fiber is going to be pennies, and the dot of goop would be even less."

It might also be more reliable than existing devices, because the chemistry doesn't consume glucose. Commercially available implantable sensors measure the voltage caused by a chemical reaction that consumes glucose. If the concentration around the sensor goes down, and glucose from the surrounding tissue doesn't flow in quickly enough, one could be measuring a value that is lower than the actual concentration in the body, Loeb says.