A non-invasive device for measuring blood sugar levels using light will undergo its latest set of trials this autumn.
The machine scans the tissue underneath the skin using near-infrared light so there is no need for people with diabetes to draw blood to analyse their glucose levels. The molecules in the blood are then analysed using a technique called Raman spectroscopy to determine how much glucose is present.
Scientists at the Massachusetts Institute of Technology (MIT) have been working on the technology for 15 years, led until recently by the late physics professor Michael Feld.
The latest version of the device is an attempt to allow any patient to use the device without having to keep recalibrating it with his or her own blood samples.
‘We had seen that while we can calibrate for one individual for around three to four hours, we were having trouble creating a universal calibrating system that can work across a large population over longer periods of time,’ said MIT graduate student Ishan Barman, who is developing the device with fellow student Chae-Ryon Kong.
One of the main obstacles was the fact that near-infrared light only penetrates about half a millimetre below the skin, so it measures the amount of glucose in the interstitial fluid that surrounds skin cells rather than the blood.
To overcome this problem, the team developed algorithms to predict blood glucose levels based on the interstitial readings, taking into account the time lag between glucose or insulin entering the bloodstream and the corresponding change in the surrounding tissue.
They also adjusted the algorithms to account for the varying optical absorption level in different people, which depends on factors such as skin colour, hydration and fat and cause distortions in the Raman spectroscopy readings.
The scanner works by sending out infrared light with the shortest wavelength, which interacts with the molecular bonds of the different substances in the fluid. The reflected signal is then analysed to determine the quantities of the substances, including glucose, and the rate at which they are changing.
Barman and Kong hope to have a working prototype device ready in two years, but their equipment is currently the size of a shopping trolley, so they are also working on miniaturising it for use by doctors and in patients’ homes.
Barman speculated that it would take around four to five years to produce a clinical model, costing around $15,000 to $20,000 (£9,600 to £12,800) per unit, and up to eight years to produce a laptop-size model for personal use, costing around $2,000 to $3,000 per unit.