Research And Application On Signal Of Functional Near-infrared Spectroscopy In Medicine

Functional Near-infrared Spectroscopy(fNIRS)has become an emerging medical monitoring technology,due to its advantagies of non-invasive,low-cost,portability,acceptable time-resolution and spatial-resolution.It has been widely used in functional brain detection and other important fields.However,compared with some mature detection technologies such as electroencephalogram(EEG)and functional magnetic resonance imaging(fMRI),the passed research literatures on fNIRS concentrated mainly on the specific methods of signal processing.The systematic research on the fNIRS signal was insuffient.This article carried out a comprehensive and meticulous research around the fNIRS signal.The reliabitlity of the signal in the self-developed medical systems was verified.A baseline calibration method for time-domain drift was proposed.We applied the improved fNIRS system in the frequency-domain analysis to explore the change of the low frequency oscillation signal(LFOs)for short-term working memory.The main contents and innovations are as follows,(1)A general processing framework for fNIRS signal was proposed.From the perspectives of frequency-domain analysis and time-domain analysis,the general signal processing modes were analyzed for different fNIRS application purpose.We established the basic steps of fNIRS signal preprocess,including artifacts remove,noise reduction,and baseline drift calibration.We converted the obtained signal into the relative concentration or absolute concentration of hemodynamic parameters by the modified Beer-Lambert Law.We could also extract other feature variables from the hemodynamic parameters,which were indicators of test results and disease diagnose.For multiple sets of signal,the commonly used t-test,F-test and other mathematical statistics methods were able to quantify the significance and the processing effect.For more complicated situations,it was also possible to predict trends by establishing models.(2)By using a self-developed thrombus detection system and a self-developed therapeutic evaluation system,the reliability of the fNIRS signal could be verified.A series of experiments were employed,such as the stability experiment,electromagnetic interference experiment,optical blood model experiment,cuff experiment and so on.The results showed that the signal of the two fNIRS systems was robust,stable and sensitive,which could promote the clinical application of fNIRS.(3)From the time-domain perspective,we explored the important effect to fNIRS signal preprocessing by the time-domain drift,and analysised the possible sources of the drift.An online baseline calibration method was proposed for the first time.The timedomain drift originated mainly from the self-regulation of human body,instability of the system component and the interference during data acquisition.The fourth-order polynomial fitting method was the optimal method for baseline drift calibration.The online time-domain calibration was particularly suitable for the monitoring system,which had high demand of online detection.We verified the temperature of the sensor and the light source without any change in the detection process.The results indicated that the components did not cause the time-domain drift.(4)From the frequency-domain perspective,we presented the method of LFOs extratction.The methods was crucial to understand the cerebral regulation mechanism under verbal working memory.We recorded the changes of hemodynamic parameters with the N-back experimental paradigm.The LFOs were caculated from the recorded results to figure out the short-term memory of brain function.It turned out that,with the increasing difficulties of the task load,the activation area of the brain enlarged,and the power spectrum of the LFOs increased.Furthermore,the degree of activation was positively correlated with the difficulties.The activation state of the left language region was stronger.The results provided reliable and quantitative evidences for the theory of short-term memory when the brain was actived.Therefore,this paper focuses on the systematic research of fNIRS signal,starting from the reliability research,involving hardware development and software optimization from the perspective of time-domain and frequency-domain,and finally applying it in brain function research.We conclude that,as an emerging detection technology,a basic processing framework of the fNIRS signal has been developed.The fNIRS signal in medical applications is reliable.We can use it for research and analysis of brain function from both time-domain and frequency-domain perspectives.The fNIRS signal processing method will get more exploration and promotion in the near future.