Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Clinical performance of a low cost near infrared sensor for continuous glucose monitoring applied with subcutaneous microdialysis

: Ben Mohammadi, Lhoucine; Klotzbücher, Thomas; Sigloch, Susanne; Welzel, Knut; Göddel, Michael; Pieper, T.R.; Schaupp, Lukas


Biomedical microdevices 17 (2015), Art. 73
ISSN: 1387-2176
European Commission EC
Journal Article
Fraunhofer ICT-IMM ()
continuous; glucose monitoring; near infrared; difference spectroscopy; subcutaneous microdialysis; diabetes; interstitial fluid

In this work we present a low cost, minimally invasive, and chip-based near infrared (NIR) sensor, combined with subcutaneous microdialysis, for continuous glucose monitoring (CGM). The sensor principle is based on difference absorption spectroscopy in the 1st overtone band known to be dominated by glucose-specific absorption features. The device comprises a multi-emitter LED and InGaAs-photodiodes, which are located on a single electronic board (non-disposable part), connected to a personal computer via Bluetooth. The disposable part consists of a chip containing the fluidic connections for microdialysis, two fluidic channels acting as optical transmission cells and total internally reflecting mirrors for in- and out-coupling of the light to the chip and to the detectors. The use of the sensor in conjunction with a subcutaneous microdialysis catheter to separate the glucose from the cells and proteins has been demonstrated to be extremely useful and advantageous for obtaining continuous glucose monitoring data and detecting glycemic levels in real time for a long period. Several in vitro and in vivo experiments were conducted to test the reliability of the device. In vitro measurements showed a linear relationship between glucose concentration and the integrated difference signal with a coefficient of determination of 99 % at the physiological concentration range. Clinical trial on 6 subjects with Type 1 diabetes showed that the NIR-CGM sensor data reflects the blood reference values adequately, if a proper calibration and signal drift compensation is applied. The MARD (mean absolute relative difference) value taken on retrospective data over all subjects is 8.5 % (range 6–11.5 %).