A method for acquiring brain response information under taste stimulation based on fNIRS

Abstract

This invention discloses a method for acquiring brain response information under taste stimulation based on fNIRS, belonging to the food field. It acquires the raw optical response signals of a multi-channel fNIRS system during changes in hemoglobin concentration in the brain of subjects under different taste stimuli. After downsampling, abnormal signal removal, truncation, and splicing of effective signal segments, spliced ​​data is obtained. Motion artifact signals in the spliced ​​data are further detected and corrected to obtain artifact-corrected data. After filtering and correcting the Beer-Lambert transform, the relative change in hemoglobin concentration is obtained. The slope features of the data are extracted and statistically tested. Based on the statistical test results, the brain response intensity and specific response regions detected by the multi-channel fNIRS system are obtained. This invention, tailored to the characteristics of fNIRS taste experiments, can detect effective response information of neural activity under different taste stimuli, and has broad application prospects in the fNIRS field.

Description

Technical Field

[0001] This invention relates to the food industry, and in particular to a method for acquiring brain response information under taste stimulation based on fNIRS. Technical Background

[0002] As one of the five ancient senses, taste is crucial for human survival. Typically, the signal generated by the binding of taste substances to taste receptors is transduced and transmitted hierarchically through afferent nerves to higher processing centers. After integration and encoding, it is projected onto the gustatory cortex of the cerebral cortex, ultimately producing the sensation of taste. Neural integration and the cognitive aspects of the brain have long been the focus of human research. With the development of neuroimaging technology, functional near-infrared spectroscopy (fNIRS) systems, due to their portability, low cost, safety, the ability for participants to undergo scanning under normal conditions, and low sensitivity to motion artifacts, have been increasingly applied to the study of taste in neuroscience in recent years.

[0003] Researchers have been continuously striving to elucidate the neurophysiological mechanisms of taste. fNIRS, as an optical neuroimaging tool, can provide quantitative information on cerebral hemodynamics and plays a crucial role in studying taste cognition in the cerebral cortex.

[0004] Compared to EEG-related taste experiments, fNIRS taste experiments typically have a longer experimental period due to slower hemodynamic response and a significant time commitment for rinsing. Furthermore, the raw signals acquired often contain a large amount of invalid signals, making conventional processing methods prone to signal contamination and alteration of valid signals. Current data analysis methods are still imperfect and lack a rigorous and complete paradigm, requiring targeted improvements and development. Summary of the Invention

[0005] This invention proposes a method for acquiring brain response information under taste stimulation based on fNIRS, targeting the characteristics of fNIRS taste experiments. By removing abnormalities, truncating, reconnecting, and then correcting and quantifying the original optical response signal, it can detect the effective response information (response intensity and response area) of neural activity under different taste stimuli, and has broad application prospects in the field of fNIRS.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A method for acquiring brain response information to taste stimulation based on fNIRS includes the following steps:

[0008] (1) Obtain the raw optical response signal of the multi-channel fNIRS system during the change of brain hemoglobin concentration in subjects under different taste stimuli;

[0009] (2) The original optical response signal is downsampled to obtain data x1(t). Abnormal channels and trials in x1(t) are removed to obtain the overall time series data x2(t) after removing abnormal signals.

[0010] (3) Extract the effective signal segments related to taste stimulation from the data x2(t), perform de-drift processing on the extracted effective signal segments, and then reassemble them to obtain the spliced ​​data x3(t);

[0011] (4) Detect motion artifact signals in the spliced ​​data x3(t) and correct the window segment containing motion artifact signals to obtain the artifact-corrected data x4(t);

[0012] (5) The artifact-corrected data x4(t) is filtered, and the filtered optical signal is converted into the relative change data of hemoglobin concentration x5(t) by modifying the Beer-Lamber law;

[0013] (6) The data of each channel in data x5(t) is averaged in segments according to the number of different taste stimuli, and the slope feature f(s) of the data from the 2nd to the 7th second after segment averaging is extracted.

[0014] (7) The slope feature was used as a quantitative indicator for statistical testing. Based on the statistical test results, the brain response intensity and specific response area detected by the multi-channel fNIRS system were obtained.

[0015] Furthermore, step (2) specifically involves:

[0016] (2.1) Remove channels and trials with abnormal overall trends;

[0017] (2.2) Calculate the coefficient of variation for the remaining channels and trials. The specific calculation formula is as follows:

[0018]

[0019] Wherein, CV represents the coefficient of variation, and σ and μ represent the standard deviation and mean of the channel or trial;

[0020] (2.3) Channel variation coefficient CV chah >15%, coefficient of variation (CV) of trials trial Data exceeding 5% is considered outlier and will be removed.

[0021] Furthermore, step (3) specifically involves:

[0022] (3.1) Locate and mark the starting point of the taste stimulus in the overall time series data x2(t) after removing abnormal signals;

[0023] (3.2) Extract the signal 5 seconds before and 30 seconds after the marker point. Subtract the signal 5 seconds before the marker point from the extracted 35 seconds of data. S Data mean;

[0024] (3.3) The extracted data segments are reassembled to obtain the spliced ​​data x3(t).

[0025] Furthermore, step (4) specifically involves:

[0026] (4.1) Divide the dataset x3(t) into windows with a length of W = 2k + 1, where k is a natural number;

[0027] (4.2) Use the quartile method to find outliers. The criteria for judgment are:

[0028] x(t i ) < Q1-1.5IQR or x(t) i >Q3+1.5IQR

[0029] Where Q1, Q3, and IQR are the first quartile, third quartile, and fourth quartile, respectively; x(t) i ) represents t in the data x3(t) i The signal corresponding to the time; will satisfy the above equation x(t) i Mark the signal as a motion artifact signal and obtain the window segment containing the motion artifact signal;

[0030] (4.3) Calculate the standard deviation of the window movement, set a threshold T, and if s(t0) < T, mark the window segment as a window segment containing motion artifact signals;

[0031] (4.4) Store the window segments containing motion artifact signals detected in steps (4.2) and (4.3) into set x. MA In (t), a polynomial least squares fit is performed on the data within a window segment of length W = 2k + 1. All time points of the window segment containing motion artifact signals are substituted into the fitting formula to obtain the dataset x′ composed of the corrected window segment. MA (t);

[0032] (4.5) Convert the corrected window segment dataset x′ MA (t) is concatenated with the window segment dataset without motion artifacts in a time series manner to reconstruct the signal;

[0033] (4.6) Calculate the mean of each window segment in the reconstructed signal, and use the difference between the mean and the mean of the previous window segment as the vertical displacement for parallel displacement. After correction, the complete artifact-corrected data x4(t) is obtained.

[0034] Furthermore, the formula for calculating the moving standard deviation is as follows:

[0035]

[0036] Where s(t0) represents the moving standard deviation of a window segment, x(t0+j) represents the signal at time t+j within the window segment, and t0 is the center time within the window segment.

[0037] Compared with the prior art, the advantages of the present invention are:

[0038] (1) This invention proposes a method for obtaining information on brain response intensity and response area under taste stimulation based on a near-infrared brain imaging system. Considering the characteristics of fNIRS taste stimulation experiments, such as the need for rinsing the mouth between stimulations and the long interval between single stimulations, a data processing method is proposed that involves removing and then extracting the original data, reconnecting it, correcting it, and finally performing quantitative analysis, which effectively reduces the complexity of experimental data.

[0039] (2) The entire process from data acquisition to the establishment of evaluation criteria established by this invention can analyze and utilize the near-infrared brain imaging system to detect the brain’s response to different taste stimuli and the specific response areas.

[0040] (3) The data processing method proposed in this invention can provide supplementary reference information and tools for us to explore the neural mechanism of taste stimulation using fNIRS, and should have broad application prospects in the field of fNIRS. Attached Figure Description

[0041] Figure 1 This is a flowchart of a method for acquiring brain response information under taste stimulation based on fNIRS;

[0042] Figure 2 It is a distribution map of the laser source and optical probe of the functional near-infrared brain imaging system and their corresponding locations on the brain;

[0043] Figure 3 This is a flowchart of motion artifact detection and correction in this invention;

[0044] Figure 4 The results show the differences between different channels in the 0.15M sucrose solution and the control group.

[0045] Figure 5 The results show the differences between different channels in the 0.3M sucrose solution and the control group.

[0046] Figure 6 The results show the differences between different channels in the 0.6M sucrose solution and the control group.

[0047] Figure 7This is the single taste stimulation paradigm used when obtaining experimental samples in this embodiment. Detailed Implementation

[0048] The present invention will be further described below with reference to the accompanying drawings.

[0049] Example 1

[0050] Taking the stimulation of sweet solutions of different concentrations as an example, such as Figure 1-2 A method for acquiring brain response information under taste stimulation based on a near-infrared brain imaging system mainly includes the following steps:

[0051] (1) Under stimulation by different concentrations of taste solutions, the raw optical response signals of the multi-channel fNIRS system were obtained during the change of brain hemoglobin concentration in R subjects;

[0052] (2) The original optical response signals of R subjects were downsampled to obtain data x1(t). Based on the experimental process records, the overall trend of the data and the coefficient of variation, abnormal channels and trials in x1(t) were removed to obtain the overall time series data x2(t) after removing abnormal signals.

[0053] (3) Extract the effective signal segments related to taste stimulation from the data x2(t), perform de-drift processing on the extracted effective signal segments, and then reassemble them to obtain the spliced ​​data x3(t);

[0054] (4) The motion artifact signal in the spliced ​​data x3(t) was detected by the quartile method and the sliding standard deviation method, and the window segment containing the motion artifact signal was corrected to obtain the artifact-corrected data x4(t).

[0055] (5) The artifact-corrected data x4(t) is bandpass filtered from 0.01 Hz to 0.1 Hz, and the filtered optical signal is further converted into the relative change data of hemoglobin concentration x5(t) by modifying the Beer-Lamber law.

[0056] (6) The data of each channel of data x5(t) were averaged in segments according to the number of stimulations by different concentrations of taste solution, and the slope feature f(s) of the data from the 2nd to the 7th second of the segmented average was extracted.

[0057] (7) Use slope characteristics as a quantitative indicator for statistical testing, and obtain the brain response intensity and specific response area detected by the multi-channel fNIRS system based on the statistical test results.

[0058] In this embodiment, the abnormal data removal in step (2) specifically involves:

[0059] (2.1) Based on the experimental process records and the overall trend of the data, manually mark and remove abnormal channels and trials;

[0060] (2.2) Calculate the coefficient of variation for each channel and trial. The specific calculation formula is as follows:

[0061]

[0062] Wherein, CV represents the coefficient of variation, and σ and μ represent the standard deviation and mean of the channel or trial;

[0063] (2.3) Channel variation coefficient CV chan >15%, coefficient of variation (CV) of trials trial Data exceeding 5% is considered outlier and will be removed.

[0064] In this embodiment, step (3) specifically includes:

[0065] (3.1) Locate and mark the starting point of the taste stimulus in the overall time series data x2(t) after removing abnormal signals;

[0066] (3.2) Extract the signal 5 seconds before and 30 seconds after the marker point, and subtract the average value of the data 5 seconds before the marker point from the extracted 35-second data to achieve the purpose of removing signal drift;

[0067] (3.3) After removing the drifting data, the data is reassembled to obtain the reassembled data x3(t).

[0068] In this embodiment, as Figure 3 As shown, the motion artifact correction in step (4) specifically involves:

[0069] (4.1) Divide the dataset x3(t) into windows with a length of W = 2k + 1, where k is a natural number.

[0070] (4.2) Use the quartile method to find outliers.

[0071] Arrange all values ​​in the data x3(t) from smallest to largest and divide them into four equal parts. The values ​​at the three dividing points are called quartiles. Q1 is the first quartile, equal to the 25th percentile of all values ​​in the sample arranged from smallest to largest; Q2 is the second quartile, equal to the 50th percentile of all values ​​in the sample arranged from smallest to largest; Q3 is the third quartile, equal to the 75th percentile of all values ​​in the sample arranged from smallest to largest. The difference between the third quartile and the first quartile is called the fourth quartile (IQR). Set the outlier detection limit to [Q1 - 1.5IQR, Q3 + 1.5IQR]. If x(t) i ) < Q1-1.5IQR or x(t) i )>Q3+1.5IQR, where x(t)i ) represents t in the data x3(t) i The signal corresponding to time t will satisfy the above equation x(t) i The image is marked as a motion artifact signal, and a window segment containing the motion artifact signal is obtained.

[0072] (4.3) Calculate the standard deviation of the window movement, set a threshold T, and if s(t0) < T, then mark the window segment as a window segment containing motion artifact signals. The specific calculation formula is as follows:

[0073]

[0074] Where s(t0) represents the moving standard deviation of a window segment, x(t0+j) represents the signal at time t+j within the window segment, and t0 is the center time within the window segment.

[0075] (4.4) Store the window segments containing motion artifact signals detected in steps (4.2) and (4.3) into set x. MA In (t), the data mainly includes high-frequency peak signals and baseline offset signals. Polynomial least squares fitting is performed on the data within a window segment of length W = 2k + 1, fitting the equidistant data within the window into an m-th degree polynomial. The fitting formula is:

[0076] x′ MA (t)=a0+a1t+…a m t m

[0077] Substituting all time points of the window segment containing motion artifact signals into the fitting formula, we obtain the dataset x′ composed of the corrected window segment. MA (t), a0, a1...a m These are coefficients to be determined.

[0078] (4.5) Convert the corrected window segment dataset x′ MA (t) and the window segment dataset x without motion artifacts OK (t) The signal is reconstructed by connecting them according to the time sequence.

[0079] Specifically, the window segment dataset x without motion artifacts OK (t) and the corrected window segment dataset x′ MA (t) are respectively: x OK (t)={x OK,k1 (t)}, k1=1,2,...L1;x′ MA (t)={x′ MA，k2 (t)}, k2=1,2,......L2。where L1 is stored in xOK The number of window segments without motion artifacts in (t), where L2 is the number of segments stored in x′. MA The number of corrected window segments in (t).

[0080] The reconstructed data x′(t) is as follows:

[0081] x′(t)={x OK,1 (t), x′ MA，1 (t), x OK ,2(t),x′ MA，2 (t), ...{x OK , L1 (t)},x′ MA，L2 (t)}

[0082] (4.6) Calculate the mean of each window segment in x′(t), and use the difference between the mean and the mean of the previous window segment as the vertical displacement for parallel displacement. After correction, the complete artifact-corrected data x4(t) is obtained. The specific formula is as follows:

[0083] x4(t)={x OK,1 (t), x′ MA，1 (t)+d1,...x OK,i (t)+d i ...+d L1+L2-1}

[0084] Where, d i Let be the vertical displacement of the i-th window segment.

[0085] In this embodiment, step (6) specifically involves: averaging the data under the same taste stimulus in data x5(t) to obtain the average data HbO_Avg. Calculating the slope f(s) 2-7 seconds after the marker point under different taste stimuli, where s = 1, 2...S, and S is the number of experimental settings, i.e., the number of taste stimuli. The specific calculation formula is:

[0086]

[0087] Wherein, HbO_Avg(s2) is the average data of the second second after the marked point under experimental condition s.

[0088] In this embodiment, the sample grouping for statistical detection and comparison in step (7) is specifically as follows: Under each experimental condition, the slope characteristics of R subjects in the same channel are grouped into a statistical detection sample, that is, a sample contains R values. A paired-samples t-test is then performed between each experimental group and the control group. The specific calculation formula is as follows:

[0089]

[0090] in, denoted as the mean of the differences between two paired samples, μ1 and μ2 are the mean slope characteristics of the experimental group and the control group, respectively, s′ is the standard deviation of the differences between paired samples, and R is the number of samples.

[0091] The channels with significant t-test results are the taste stimulation activation channels. The response regions can be determined based on all taste stimulation activation channels, and the magnitude of the response signal corresponding to the taste stimulation activation channel is used as the brain response intensity.

[0092] Example 2

[0093] See Figure 1-2 To better illustrate the method of this invention, using sweet solution stimulation as an example, 29 subjects were recruited for the experiment, which included the following steps:

[0094] Step (1): Collect behavioral experimental data of the subjects using brain fNIRS. The experimental setup is as follows:

[0095] In this study, 29 volunteers were recruited. All participants were right-handed and in good physical and mental condition. Participants were not allowed to eat or smoke one hour before the experiment. The experiment was conducted in a safe, quiet, and odorless room at a suitable temperature of 26±1℃. Sucrose solution was used as the sweetener in this study, with concentrations of 0.15M, 0.3M, and 0.6M (denoted as SU1, SU2, and SU3, respectively), and purified water served as the control group.

[0096] In this embodiment of the invention, each stimulation group in the experiment included 30 seconds of resting state, 5 seconds of automatic sample injection, 30 seconds of oral administration, and 30 seconds of swallowing the aftertaste, followed by rinsing the mouth and repeating twice. The experiment consisted of three experimental groups and one control group. The paradigm of a single taste stimulation is as follows: Figure 7 As shown, this experiment utilized a self-made automated oral sample introduction system for near-infrared brain imaging taste experiments. During the preparation phase, the subject held the sample introduction module connected to an external delivery tube in their mouth, and the sample introduction time was set to 5 seconds and the interval to 90 seconds according to the experimental paradigm. The brain response signals induced by taste stimulation were recorded during the experiment.

[0097] In this embodiment, the fNIRS imaging system (Brite24, Artinis Medical Systems) was used to record data. This system is a dual-band portable 24-channel system, including an optical cap that can completely cover the frontal lobe of the human body. Ten laser sources and eight optical probes are evenly distributed in a staggered pattern at 30mm intervals within the optical cap. See details... Figure 2The data sampling rate was 50 Hz. In the experiment, a three-dimensional magnetic spatial digital instrument (Polhemus Inc, http: / / www.polhemus.com) was used to measure the three-dimensional spatial position of each photoelectrode on the scalp of each subject.

[0098] Steps (2)-(7) are the same as in Example 1.

[0099] Step (8): Under each experimental condition, the slope characteristics of R subjects in the same channel are grouped into a statistical test sample, i.e., a sample contains R values. Paired-samples t-tests are then performed on each experimental group and the control group. Based on the test results, it is analyzed whether the near-infrared brain imaging system can detect the brain's response to different taste stimuli and the specific response areas, as follows:

[0100] Using the obtained slope index, paired t-tests were performed on 0.15M sucrose solution and control group, 0.3M sucrose solution and control group, and 0.6M sucrose solution and control group. Based on the test results, channels of significant differences were analyzed to identify brain regions with different neural activity under stimulation by different concentrations of sweeteners. See details... Figures 4-6 .

[0101] Figure 4 The diagram shows the activation status of different channels obtained after operation according to the method proposed in this invention, under stimulation with a 0.15M sucrose solution and a control group. The horizontal axis represents the corresponding... Figure 2 The figure shows 24 channels distributed across different brain regions. The vertical axis represents the t-value obtained from a paired-samples t-test for the two stimulus conditions. * indicates regions that showed significant brain activation compared to the control group (p<0.05). As shown in the figure, under stimulation with 0.15M sucrose solution, the effectively activated brain channels detected by fNIRS were Ch8, Ch10, Ch12, Ch13, Ch14, Ch15, and Ch17.

[0102] Figure 5 The diagram shows the activation status of different channels obtained after operation according to the method proposed in this invention, under stimulation with a 0.3M sucrose solution and a control group. The horizontal axis represents the corresponding... Figure 2 The figure shows 24 channels distributed across different brain regions. The vertical axis represents the t-value obtained from a paired-samples t-test for the two stimulus conditions. * indicates regions that showed significant brain activation compared to the control group (p<0.05). As shown in the figure, under 0.3M sucrose solution stimulation, the effectively activated brain channels detected by fNIRS were Ch5, Ch8, Ch10, Ch13, Ch14, Ch15, Ch16, and Ch17.

[0103] Figure 6The diagram shows the activation status of different channels obtained after operation according to the method proposed in this invention, under stimulation with a 0.6M sucrose solution and a control group. The horizontal axis represents the corresponding... Figure 2 The figure shows 24 channels distributed across different brain regions. The vertical axis represents the t-value obtained from a paired-samples t-test for the two stimulus conditions. * indicates regions that showed significant brain activation compared to the control group (p<0.05). As shown in the figure, under 0.6M sucrose solution stimulation, the effectively activated brain channels detected by fNIRS were Ch5, Ch8, Ch10, Ch11, Ch12, Ch13, Ch14, Ch15, Ch16, and Ch17.

[0104] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for acquiring brain response information under taste stimulation based on fNIRS, characterized in that, Includes the following steps: (1) Obtain the raw optical response signal of the multi-channel fNIRS system during the change of brain hemoglobin concentration in subjects under different taste stimuli; (2) Data is obtained by downsampling the original optical response signal. ,right Channels and trials exhibiting abnormalities were removed to obtain the overall time series data after removing abnormal signals. ; (3) In the data Effective signal segments related to taste stimuli are extracted from the data. These segments undergo drift removal processing and are then reassembled to obtain the final data. ; Step (3) specifically involves: (3.1) Overall time series data after removing outliers Locate and mark the starting point of taste stimulation in the middle; (3.2) Extract the signal 5 seconds before and 30 seconds after the marker point, and subtract the average value of the data 5 seconds before the marker point from the extracted 35-second data; (3.3) The extracted data fragments are reassembled to obtain the spliced ​​data. ; (4) Detect the spliced ​​data The motion artifact signal in the image is analyzed, and the window segment containing the motion artifact signal is corrected to obtain the artifact-corrected data. ; Step (4) specifically involves: (4.1) For the dataset Divide the window into sections, with a window length of [value missing]. , It is a natural number; (4.2) Use the quartile method to find outliers. The criteria for judgment are: or ; in, These are the first quartile, the third quartile, and the fourth quartile, respectively. Representing data middle The signal corresponding to the time will satisfy the above equation. Mark the signal as a motion artifact and obtain the window segment containing the motion artifact signal; (4.3) Calculate the moving standard deviation of the window, set the threshold T, if If so, then the window segment is marked as a window segment containing motion artifact signals; (4.4) Store the window segments containing motion artifact signals detected in steps (4.2) and (4.3) into a set. In the middle; using polynomials to calculate the length of Multinomial least squares fitting is performed on the data within the window segment. All time points of the window segment containing motion artifact signals are substituted into the fitting formula to obtain a dataset composed of the corrected window segment. ; (4.5) The corrected window segment dataset The signal is reconstructed by concatenating it with a window segment dataset free of motion artifacts in a time series manner; (4.6) Calculate the mean of each window segment in the reconstructed signal. Use the difference between the mean and the mean of the previous window segment as the vertical displacement for parallel displacement. After correction, the complete artifact-corrected data is obtained. ; (5) Data after artifact correction Filtering is performed, and the filtered optical signal is converted into relative change data of hemoglobin concentration by correcting the Beer-Lambert law. ; (6) Data based on the number of different taste stimuli The data from each channel is averaged in segments, and the slope feature of the data from the 2nd to the 7th second after segment averaging is extracted. ; (7) The slope feature is used as a quantitative indicator for statistical testing. Based on the statistical test results, the brain response intensity and specific response area detected by the multi-channel fNIRS system are obtained.

2. The method for acquiring brain response information under taste stimulation based on fNIRS according to claim 1, characterized in that, Step (2) specifically involves: (2.1) Remove channels and trials with abnormal overall trends; (2.2) Calculate the coefficient of variation for the remaining channels and trials. The specific calculation formula is as follows: ； in, Represents the coefficient of variation. Indicates the standard deviation and mean of a channel or number of trials; (2.3) Channel variation coefficient coefficient of variation in trials Data that is considered abnormal will be removed.

3. The method for acquiring brain response information under taste stimulation based on fNIRS according to claim 1, characterized in that, The formula for calculating the moving standard deviation is as follows: ； in, This represents the standard deviation of the movement of a window segment. Indicates the first [number]th [part] within the window fragment The signal corresponding to the time, This represents the center moment within the window segment.

Images