Near Infrared Spectroscopy for Continuous Glucose Monitoring
MetadataShow full item record
Millions of people are affected by diabetes. The disease can be managed by measuring the blood glucose levels and administering appropriate insulin doses. However, diabetes management is difficult and time consuming for the patient and the medical community works to achieve a fully automated artificial pancreas (AP) that can manage the disease for the patients so that they can live normal lives and experience better health. Towards the goal of a fully automated AP for diabetes management, there is a need for an accurate long-term continuous glucose monitoring (CGM) device. One possible approach is to measure the glucose by near infrared (NIR) spectroscopy. Although a noninvasive CGM device would be the preferred solution for the patients, it has been difficult to achieve despite many commercialization efforts throughout the past 20 years. Additionally, the glucose concentration measured noninvasively might be delayed for several minutes compared to the blood glucose value, which can diminish the effect of an AP system. As an alternative, measurements in the intraperitoneal (IP) fluid surrounding the inner organs might be an option where the dynamics are faster. The aim of this thesis is to investigate the viability for a CGM sensor based on NIR spectroscopy for measurements in fluids, such as the IP fluid. To avoid the potential risk of placing active devices in the IP space, the thesis focuses on optical fiber-based sensing. Optical fiber components are widely used and developed for the telecommunication range around the 1550nm wavelength, overlapping with the first overtone band of glucose absorption. The thesis contributions are divided into three sections. The first section evaluates existing technology and is comprised of Papers I and II. The first paper reviewed the status of optical technologies for CGM sensing for diabetes. The second paper investigated the effects of interfering substances on calibration models that had not encountered the interfering substances before. The measurements were acquired from a benchtop lab spectrometer. Lactate and ethanol were found to interfere with glucose predictions if they were not part of the calibration set for model building. The paper concluded that the calibration of a near infrared spectroscopy (NIRS) model for CGM must be based on samples with large variance and from different physiological situations. The second section is the main part of the thesis where experimental NIRS setups were built and optimized and span Papers III-V. The third paper developed and characterized lensed fibers to increase the transmission in the interaction region without bulky components. The fourth paper investigated the needed signal to noise ratio (SNR) levels to achieve accurate NIRS glucose sensing. It showed that the detector noise was a limiting factor when applying a broadband lamp as a light source, whereas the relative intensity noise (RIN) of the source was the limiting factor when employing brighter supercontinuum (SC) sources that could overcome the detector noise. The RIN could be reduced by referencing every pulse from one of the SC lasers. Paper V further developed this SC-based system and added a more stable probe and a balanced detection scheme where also the large and varying water absorption could be canceled at well-balanced wavelengths. The system was approaching the fundamental shot noise limit at well-balanced wavelengths, but the noise increased due to instability of the transmission cell. Limiting instrumental factors were investigated and possible solutions to realize a fiber and NIRS-based CGM sensor were discussed. The third section focuses on overcoming the low absorption cross-section in the first overtone band by signal enhancement on an optical fiber and includes Papers VI and VII. This research on enhanced absorption spectroscopy is at an earlier stage than traditional NIR spectroscopy and was conducted stepwise with the ultimate goal of surface enhanced near-infrared absorption spectroscopy (SENIRAS) for glucose sensing. Paper VI realized absorption enhancement on an optical fiber in the visible band on model dyes. Paper VII demonstrated plasmon resonance on optical fibers in the NIR range and around wavelengths where glucose absorbs. In summary, this thesis investigates limitations and challenges for instrumentation, analysis and methods, followed by solutions and strategies to accomplish NIRSbased CGM using optical fibers for the IP space.