Sweet sensor for sugar

German and Austrian medical researchers have developed a portable NIR spectroscopy device for monitoring glucose concentrations in the blood of diabetics.

Glucose monitoring is an important part of managing diabetes, allowing sufferers to alter their diet or insulin injections accordingly. Recently, several research groups have shown that NIR spectroscopy offers a great deal of promise as a quick and reasonably non-invasive technique for glucose monitoring. But questions remain over the best way to conduct such monitoring, as other components in the blood like fat and proteins can interfere with the signal from glucose while bodily tissue such as skin can scatter NIR light.

So the team of medical researchers, led by Lhoucine Ben Mohammadi at the Institut für Mikrotechnik in Mainz, set about developing a portable NIR device for glucose monitoring that would answer these questions. The device consists of a microdialysis probe linked to a microchip illuminated by three light-emitting diodes (LEDs).

The microdialysis probe is a needle-like tube of semi-permeable membrane that allows glucose molecules to pass through it but not the larger protein and fat molecules. When the probe is inserted into a diabetes patient, it collects glucose and then transports it to the external microchip worn on the patient’s arm. Here, the glucose is illuminated be the three LEDs, which emit light at NIR wavelengths known to be absorbed by glucose. In this way, the glucose is separated from the other major components of the blood while scattering of NIR light by bodily tissue is no longer an issue.

As reported in Biosensors and Bioelectronics, when Ben Mohammadi and his colleagues tested out this device on various glucose solutions, they found that it could accurately determine the glucose concentrations in all of them. Next, they tried it out on 10 diabetes patients, finding that it could determine glucose concentrations in their blood with reasonable accuracy. Over all the patients, the device had a mean absolute relative error of just 13.8%.

However, achieving this level of accuracy did require a fairly extensive initial calibration stage, to account for differences in background noise between patients, and continuous corrections for shifts in the sensor signal, caused by variations in flow conditions.

Blog tags: