| 摘要: | 世界衛生組織在2016發表的期刊指出全球糖尿病的人口不斷攀升，且糖尿病在國人死亡率排名中為第五名，由此可知定期的血糖監控是十分重要的，然而傳統血糖機無法連續測量，且採血過程造成使用者的不適感，因此近年來血糖機的發展逐漸朝向非侵入式血糖量測研究，在目前提出的技術中，又以光學量測方式最有潛力。然而在發展上，由於葡萄糖訊號相當微弱，在做光學量測時容易受外在因素干擾產生誤差，且諸多研究採用的光源多為近紅外光因此量測時也發現易受溫度造成的基線飄移所干擾。因此，本研究欲利用雙光源量測技術，發展非侵入式血糖偵測系統，該系統由單晶片Arm-CortexM4以SPI方式作為溝通連接AFE4490晶片並配合本實驗室設計的PPG量測探頭利用H-bridge電路進行雙光源PPG資料讀取，PPG訊號的資料取樣率設定為500Hz，再利用藍芽方式將受測者PPG數據回傳電腦使用並matlab進行演算法運算。
本系統藉由近紅光對血糖較為敏感的特性作為系統所用之光源，並以綠光作為校正基準用以消除交感神經變化所造成的量測誤差。經由實驗量測18名健康正常人的綠光以及近紅外光PPG訊號，並藉服用糖水的方式改變受試者血糖值，以市售血糖機OneTouch-UltraEasy量測血糖，接著以迴歸分析驗證利用本系統所預測的血糖變化量與實際量測的血糖變化量誤差為22.57%，證實了光學血糖偵測技術的可行性。然而在本研究中發現，因個人生理狀況的差異所以經演算法所計算出來的數值在受測者之間會有基線差異的問題，因此本系統所呈現的結果均為血糖變化量對上旋光性比例變化，對於量測血糖值仍然須對個人生理狀態去做系統校正才能夠應用於血糖量測。
;The World Health Organization published a journal in 2016 pointing out that the global population of diabetes continues to rise and that diabetes is ranked fifth in the country′s mortality rate. This shows that regular blood glucose monitoring is very important, but conventional blood glucose machines cannot measure continuously, and blood collection process causes discomfort to the user. Therefore, in recent years, the development of blood glucose machines has gradually been directed towards non-invasive blood glucose measurement. Among the currently proposed technologies, optical measurement methods are most promising. However, in terms of development, because the glucose signal is rather weak, it is easy to be subject to external factors to interfere with the error during optical measurement. Many of the light sources used in the research are near-infrared light. Therefore, baseline drift that is susceptible to temperature is also found during measurement. Therefore, this study intends to develop a non-invasive blood glucose detection system using dual-lamp measurement technology. The system uses a single-chip microcomputer Arm-Cortex M4 to communicate and connect AFE4490 Chip with the SPI method and to use the PPG measurement probe designed by the laboratory with H-bridge circuit for dual light source PPG data read, PPG signal data sampling rate is set to 500Hz, and then use the Bluetooth method to test the PPG data back to the computer and then use matlab to calculate algorithm.
The system uses near-red light sensitivity to blood glucose as the light source used by the system, and uses green light as the calibration reference to eliminate measurement errors caused by sympathetic nerve changes. The green light and near-infrared light PPG signals of 18 normal healthy people were measured, and the blood glucose level of subjects was changed by taking sugar water, and blood glucose was measured with the commercially available blood glucose meter OneTouch-UltraEasy, and then verified by regression analysis. The error of the blood glucose variation predicted by the system and the actual measurement was 22.57%, confirming the feasibility of the optical blood glucose detection technology. However, in this study, it was found that the values calculated by the algorithm due to differences in the individual′s physiological conditions may have a problem of baseline differences among the subjects. Therefore, the results presented in this system are all the changes in the blood glucose level with Optical activity changes,form this reason our system for the measurement of blood glucose values still need to be recalibrated for individual statment and that can be applied to blood glucose measurement. |
