CN110547768A - Near-infrared brain function imaging quality control method and control system - Google Patents

Abstract

The invention belongs to the field of brain detection, and relates to a near-infrared brain functional imaging quality control method and a control system, comprising the following steps: converting the original light intensity data collected by a near-infrared imaging device into optical density data; The standard deviation of a sliding window; judging the abnormal point; determining the abnormal time point according to the abnormal point; among the multiple detected abnormal time points, intercepting the distance between the two adjacent abnormal time points with the farthest distance The optical density data of each test channel; if the time series length of the optical density time series data is greater than or equal to a certain fixed time length, it is determined that the intercepted optical density time series data initially meets the quality requirements. The present invention conducts preliminary quality control by intercepting time series with relatively stable signal quality, effectively eliminating the influence of abnormal time points on signal quality evaluation, and ensuring the reliability of signal quality.

Description

A quality control method and control system for near-infrared brain functional imaging

technical field

The invention relates to a near-infrared brain function imaging quality control method and a control system, belonging to the field of brain function testing.

Background technique

At present, there are mainly the following methods for evaluating the quality of signal data collected by near-infrared brain functional imaging technology: 1) Evaluate the signal quality according to the magnitude of the signal change in the concentration of oxyhemoglobin (HbO) and deoxygenated hemoglobin (HbR); 2) Evaluate the signal quality according to the correlation of the time series change trend of the concentration signals of HbO and HbR; 3) evaluate the signal quality according to the signal-to-noise ratio of the original light intensity data; 4) calculate the power spectrum according to the concentration signals of HbO and HbR, Look for heartbeat-related peaks to assess signal quality.

Among them, method 2) cannot exclude the influence of noise signals in the concentration data of HbO and HbR on the trend of time series changes, and the quality of concentration signal data has no objective standards and relies on subjective judgment; several other methods evaluate concentration signals The emphasis on quality is different, so that only using a certain method for quality control may cause errors in the quality evaluation of the signal. In addition, the above methods are all based on the quality evaluation of the signal based on a relatively complete time series, which cannot avoid the influence of the unstable time point of data acquisition or the time point of large motion artifacts on the signal quality, resulting in inaccurate concentration data signals .

Contents of the invention

In response to the above problems, the purpose of the present invention is to provide a quality control method for near-infrared brain functional imaging, by intercepting time series with relatively stable signal quality for subsequent quality control, effectively eliminating the impact of these time points on signal quality evaluation. influences.

To achieve the above object, the present invention takes the following technical solutions:

The invention provides a method for controlling the quality of near-infrared brain functional imaging, comprising the following steps: 1) converting the original light intensity data collected by a near-infrared imaging device into optical density data; 2) recording the optical density data that changes with time , to obtain the optical density time series, adopt the sliding window method to calculate the standard deviation of each of the sliding windows for the optical density time series data; 3) assume that the standard deviation of the sliding window obeys a normal distribution, the standard deviation of the The numerical range formed after the normal distribution mean value plus and minus the predetermined number of standard deviations is defined as the normal value range, and the data points outside the normal value range are judged as abnormal points; 4) collect all test channel density time series data, if When a time point exceeds the predetermined number of test channels, the abnormal point appears, then this time point is determined as the abnormal time point; 5) among the detected multiple abnormal time points, intercept the two phases farthest The time series between adjacent abnormal time points, and judge whether the optical density time series data initially meet the requirements according to the length of the time series.

Further, in the step 5), the judging method for judging whether the optical density time series data meets the requirements is as follows: if the length of the time series is greater than or equal to the predetermined time length, then determine the intercepted optical density time series data The quality requirement is preliminarily met; if the length of the time series is less than the predetermined time length, the optical density time series data does not meet the quality requirement.

Further, signal-to-noise ratio detection, channel concentration signal correlation analysis, and power spectrum heartbeat peak detection are performed on the optical density time series data that initially meet the quality requirements.

Further, if the optical density time series data meets any two or more conditions of the signal-to-noise ratio not less than the preset value, the concentration signal correlation of each channel is positive, or the heartbeat peak appears in the power spectrum, the said Densitometric time series data met quality requirements.

Further, the detection of the signal-to-noise ratio includes: defining the signal-to-noise ratio of the signal by the ratio of the average value of the original light intensity time-series data to the standard deviation of the optical density time-series data, and calculating the signal-to-noise ratio of each test channel at different wavelengths respectively. Noise ratio, to reflect the signal acquisition quality collected by each test channel.

Further, the correlation analysis of the test channel concentration signal includes: normalizing the optical density time series data that initially meets the quality requirements, and then removing high-frequency noise and low-frequency drift through a 0.01-0.1 Hz band-pass filter, and then correcting The Beer-Lambert law transforms the filtered data into oxyhemoglobin, deoxygenated hemoglobin, and total hemoglobin concentration data, selects the concentration data that needs to calculate the signal correlation matrix, and calculates the time series of the concentration data of all test channels related to each other Get the corresponding concentration correlation coefficient matrix.

Further, the detection of the heartbeat peak value of the power spectrum includes: performing preliminary bandpass filtering of 0-3 Hz on the optical density time series data that initially meets the quality requirements, and converting the filtered data into oxygenated hemoglobin concentration, The time series of deoxygenated hemoglobin concentration; the time series of the oxygenated hemoglobin concentration and the deoxygenated hemoglobin concentration are resampled to 5Hz, and then the time series of the resampled oxygenated hemoglobin concentration and deoxygenated hemoglobin concentration are banded at 0.01-2Hz through filtering; Fourier transform is performed on the time series of the filtered oxyhemoglobin concentration and deoxyhemoglobin concentration, so that the time series of the oxygenated hemoglobin concentration and deoxygenated hemoglobin concentration are converted from time domain signals to frequency domain signals, and oxygen The amplitude corresponding to the frequency of the combined hemoglobin concentration and the deoxygenated hemoglobin concentration; the amplitude corresponding to the frequency curve of the oxygenated hemoglobin concentration and the deoxygenated hemoglobin concentration is squared to obtain the power corresponding to the frequency, and the power corresponding to the frequency is normalized A normalized power spectrum is obtained and corresponding said power spectrum is displayed for each test channel at a frequency of 0.5-1.5 Hz.

Further, in the step 4), if the abnormal time point occurs within the first few seconds of the time series or within the last few seconds of the time series, the abnormal time point is determined as an unstable time point in data collection; if the If the abnormal time point appears within the rest of the time period, the abnormal time point is a motion artifact time point.

The present invention also provides a quality control system for near-infrared brain functional imaging, including: a near-infrared imaging module for collecting original light intensity data and converting it into optical density data; a standard deviation calculation module for converting the optical density Data conversion of optical density time series data, and adopting the sliding window method to calculate the standard deviation of each of the sliding windows for the optical density time series data; the abnormal time point determination module is used to determine the standard deviation data according to the normal distribution Abnormal time point: a first judging module, configured to select an appropriate length of the optical density time series according to the abnormal time point data, and judge whether the optical density time series data initially meets the requirements according to the length of the time series.

Further, it also includes a secondary judgment module, which is used to perform signal-to-noise ratio detection, channel concentration signal correlation analysis, and power spectrum heartbeat peak detection on the optical density time series data that initially meet the quality requirements.

Due to the adoption of the above technical solutions, the present invention has the following advantages: 1) Converting the original light intensity data into optical density data improves the accuracy of data detection. 2) The outliers are judged by the standard deviation of the optical density time series data, which makes the judgment of outliers simpler, more convenient and more accurate. 3) Preliminary quality control is carried out by intercepting time series with relatively stable signal quality, which effectively eliminates the impact of abnormal time points on signal quality evaluation. 4) The present invention combines three methods of signal-to-noise ratio detection, channel concentration signal correlation analysis and power spectrum heartbeat peak detection to further evaluate the signal quality, and control the signal quality from multiple angles to ensure the reliability of the data signal quality sex.

Description of drawings

Fig. 1 is each test channel optical density time series in an embodiment of the present invention;

Fig. 2 is the signal-to-noise ratio curve of each test channel optical density data in an embodiment of the present invention;

Fig. 3 is each test channel optical density data correlation figure in an embodiment of the present invention;

Fig. 4 is a power spectrum diagram of each test channel with a frequency of 0.5-1.5 Hz in an embodiment of the present invention.

Detailed ways

The present invention will be described in detail below in conjunction with the accompanying drawings. However, it should be understood that the accompanying drawings are provided only for better understanding of the present invention, and they should not be construed as limiting the present invention. In describing the present invention, it should be understood that the terms used are for the purpose of description only, and should not be understood as indicating or implying relative importance.

An embodiment of the present invention provides a method for quality control of near-infrared brain functional imaging, comprising the following steps:

1) Convert the original light intensity data collected by the near-infrared imaging device into optical density data;

2) record the optical density data that changes with time, obtain the optical density time series, adopt the sliding window method to calculate the standard deviation of each sliding window for the optical density time series data;

3) Assuming that the standard deviation of the sliding window obeys the normal distribution, the value range formed by adding and subtracting 5 standard deviations to the normal distribution mean of the standard deviation is defined as the normal value range, and the data points outside the normal value range are judged as abnormal point;

4) Collect all test channel density time series data, if abnormal points appear in the test channels exceeding 1/3 of the total number of channels at a time point, then this time point is set as an abnormal time point;

5) Among the multiple detected abnormal time points, intercept the time series between the two adjacent abnormal time points with the furthest distance, and judge whether the optical density time series data meets the requirements according to the length of the time series. In the above-mentioned embodiment, since the original light intensity data obtained by the infrared imaging device is converted into optical density data, the accuracy of data detection is improved; the abnormal point is judged by the standard deviation of the optical density time series data, which makes the judgment of the abnormal point easier It is convenient and more accurate; by intercepting time series with stable signal quality for preliminary quality control, the influence of abnormal time points on signal quality evaluation is effectively eliminated, and the reliability of optical density data is initially guaranteed.

The judging method for judging whether the optical density time meets the requirements in step 5) of this embodiment is as follows: if the time series length is greater than or equal to the preset time length, then determine that the intercepted optical density time series data initially meets the quality requirements; if the time series length If it is less than the preset time length, the optical density time series data does not meet the quality requirements.

The abnormal time point in step 4) of this embodiment includes an unstable time point and a motion artifact time point, wherein the unstable time point refers to an abnormal point at the beginning and end of the test due to the instability of the instrument; the motion artifact The trace time point refers to the abnormal point caused by environmental factors in the test process. In this embodiment, if the abnormal time point occurs within the first 5 seconds of the time series or within the last 5 seconds of the time series, the abnormal time point will be judged as an unstable time point in data collection; if the abnormal time point occurs within the rest of the time period , then the abnormal time point is the motion artifact time point.

In step 5) of this embodiment, a certain fixed time length is preferably 300s, and the window length of the sliding window is the time for collecting the original light intensity data twice. It should be noted that this time length is only a preferred time length set based on comprehensive data accuracy and efficiency, and those skilled in the art can select an appropriate time length according to their specific testing needs.

As shown in Figure 1, in order to facilitate the operator to quickly and easily understand whether the test data is available, step 5) of this embodiment also includes a visual display device that can display the data. When the intercepted optical density time series data initially meets the quality requirements, Then, the intercepted data is displayed on the display screen of the display device, so that the user can perform subsequent processing. The black point in Figure 1 is the abnormal time point, and the optical density time series data between the two farthest and adjacent abnormal time points is intercepted, as shown in Figure 1, 10-630s is the intercepted optical density time series data , the time series of the data is greater than 300s, so it is determined that the data initially meets the quality requirements; if the intercepted optical density time series data does not meet the quality requirements, it will be displayed on the display that the data does not meet the requirements, or the user is reminded to perform the test again.

In another embodiment of the present invention, signal-to-noise ratio detection, channel concentration signal correlation analysis, and power spectrum heartbeat peak detection are also required for the optical density time series data that initially meet the quality requirements. The signal quality is evaluated by combining three methods of signal-to-noise ratio detection, channel concentration signal correlation analysis and power spectrum heartbeat peak detection, and the signal quality is controlled from multiple angles to further ensure the reliability of the data signal quality.

Among them, the signal-to-noise ratio detection includes: defining the signal-to-noise ratio of the signal by the ratio of the average value of the original light intensity time-series data to the standard deviation of the optical density time-series data, and calculating the signal-to-noise ratio of each test channel at different wavelengths, to It reflects the signal quality collected by each test channel. The higher the SNR, the better the signal quality. As shown in Figure 2, under normal circumstances, if the signal-to-noise ratio is greater than or equal to 2, it is determined that the measured data meets the quality requirements. Specifically in Figure 2, as long as the signal-to-noise ratio of the measured data is higher than the bold black solid line in the figure, the measured data is considered to meet the quality requirements.

The correlation analysis of the concentration signal of the test channel includes: normalizing the optical density time series data that initially meets the quality requirements, and then removing high-frequency noise and low-frequency drift through a 0.01-0.1Hz band-pass filter, and then according to the corrected Beer-Lambert The law converts the filtered data into oxyhemoglobin, deoxygenated hemoglobin, and total hemoglobin concentration data, selects the concentration data that needs to calculate the signal correlation matrix, and obtains the corresponding concentration correlation coefficient by calculating the time series of concentration data of all test channels related to each other matrix. In the concentration correlation coefficient matrix, the larger the correlation coefficient value, the greater the correlation between the channel signal and other channel signals. Through this method, it is possible to start from the whole, and detect channels that fail to collect data due to poor equipment contact according to the correlation mode between the overall signals. Usually, only when the correlation coefficient is very small, as shown in Figure 3, it is considered that the test channel has not collected data when it is close to -1, and the area marked by the bold solid line in Figure 3 is the test that has not collected data aisle. However, in order to ensure the accuracy of data quality in this embodiment, it is preferable that the correlation coefficient of each channel is greater than or equal to 0, that is, the test data meets the quality standard.

As shown in Figure 4, the power spectrum heartbeat peak detection includes: preliminary band-pass filtering of 0-3 Hz on the optical density time series data that meets the quality requirements, and converting the filtered data into oxyhemoglobin according to the modified Beer-Lambert law Concentration, time series of deoxyhemoglobin concentration.

The time series of oxyhemoglobin concentration and deoxyhemoglobin concentration were resampled to 5 Hz, and then the resampled time series of oxyhemoglobin concentration and deoxyhemoglobin concentration were band-pass filtered at 0.01-2 Hz.

Fourier transform is performed on the time series of filtered oxyhemoglobin concentration and deoxygenated hemoglobin concentration, so that the time series of oxyhemoglobin concentration and deoxygenated hemoglobin concentration are converted from time domain signals to frequency domain signals, and the oxygenated hemoglobin concentration, deoxygenated hemoglobin concentration The frequency corresponds to the amplitude of the hemoglobin concentration.

The amplitude of the oxyhemoglobin concentration and deoxyhemoglobin concentration-frequency curves is moduloed and squared to obtain the power corresponding to the frequency, and the power corresponding to the frequency is normalized to obtain a normalized power spectrum, and it is displayed that the frequency in each test channel is 0.5- Corresponding power spectrum at 1.5 Hz.

As shown in Figure 4, the corresponding power spectrum with a frequency of 0.5-1.5 Hz in each test channel shows the power spectrum diagram of 48 test channels at the same time. Through the power spectrum diagram, it can be seen at a glance which test channels appear in the power spectrum Which heartbeat power peaks do not appear, among them, the signal corresponding to the power spectrum of the test channel with the heartbeat power peak is the signal that meets the quality requirements. Using this method to sort the power spectrum diagrams in rows and columns avoids the inconvenience of checking the power spectrum of each channel multiple times, and the operation is more convenient. In addition, the frequency of 0.5-1.5 Hz is chosen because this frequency range covers the power peak of the heartbeat.

If the signal can meet the requirements of the quality standard in the above-mentioned signal-to-noise ratio detection, channel concentration signal correlation analysis, and power spectrum heartbeat peak detection, it is the final determined signal that meets the quality standard. This situation is only an ideal situation. In actual operation, as long as any two quality standards are met among the three quality standards, the signal can be determined as a signal that finally meets the standard.

Another embodiment of the present invention also provides a quality control system for near-infrared brain functional imaging, including:

The near-infrared imaging module is used to collect raw light intensity data and convert it into optical density data;

The standard deviation calculation module is used to convert the optical density data into optical density time series data, and calculate the standard deviation of each sliding window by using a sliding window method for the optical density time series data;

The abnormal time point determination module is used to determine the abnormal time point according to the standard deviation data of the normal distribution;

The first judging module is used to select an appropriate optical density time series length according to the abnormal time point data, and judge whether the optical density time series data initially meets the requirements according to the length of the time series.

Among them, the control system also includes a secondary judgment module, which is used to perform signal-to-noise ratio detection, channel concentration signal correlation analysis, and power spectrum heartbeat peak detection for the optical density time series data that initially meets the quality requirements.

Above-mentioned each embodiment is only for illustrating the present invention, and wherein each step etc. all can be changed to some extent, all equivalent transformations and improvements carried out on the basis of the technical solution of the present invention, all should not be excluded in the protection scope of the present invention outside.

Claims (10)

1. A near-infrared brain function imaging quality control method is characterized by comprising the following steps:

1) Converting original light intensity data acquired by a near-infrared imaging device into optical density data;

2) Recording the optical density data changing along with time to obtain an optical density time sequence, and calculating the standard deviation of each sliding window by adopting a sliding window method for the optical density time sequence data;

3) Assuming that the standard deviation of the sliding window obeys normal distribution, determining a numerical range formed by adding or subtracting a preset number of standard deviations from the normal distribution mean of the standard deviation as a normal value range, and judging data points outside the normal value range as abnormal points;

4) Acquiring density time series data of all the test channels, and if the abnormal point occurs when a time point exceeds a preset number of the test channels, determining the time point as an abnormal time point;

5) and intercepting a time sequence between two adjacent abnormal time points with the farthest distance from the plurality of detected abnormal time points, and judging whether the optical density time sequence data preliminarily meet the requirements according to the length of the time sequence.

2. The method for controlling quality of near-infrared brain function imaging according to claim 1, wherein in the step 5), the method for determining whether the optical density time-series data meets the requirement is as follows: if the time sequence length is greater than or equal to a preset time length, determining that the intercepted optical density time sequence data preliminarily meets the quality requirement; and if the time sequence length is less than the preset time length, the optical density time sequence data does not meet the quality requirement.

3. The method according to claim 1 or 2, wherein the optical density time-series data meeting the quality requirement is subjected to signal-to-noise ratio detection, channel concentration signal correlation analysis and power spectrum heartbeat peak detection.

4. the method according to claim 3, wherein the optical density time-series data satisfy two or more conditions selected from a signal-to-noise ratio not less than a predetermined value, a positive correlation of the concentration signals of the channels, and a peak value of a heartbeat in the power spectrum, and the optical density time-series data are considered to satisfy the quality requirement.

5. The method of claim 3, wherein the SNR detection comprises: and respectively calculating the signal-to-noise ratio of each test channel under different wavelengths to reflect the signal acquisition quality acquired by each test channel by defining the signal-to-noise ratio of the signal through the average value of the original light intensity time sequence data and the standard deviation ratio of the light density time sequence data.

6. the method of claim 3, wherein the analysis of correlation of test channel concentration signals comprises: normalizing the optical density time-series data which preliminarily meet the quality requirement, removing high-frequency noise and low-frequency drift through a 0.01-0.1Hz band-pass filter, converting the filtered data into oxyhemoglobin, deoxyhemoglobin and total hemoglobin concentration data according to a modified Beer-Lambert law, selecting the concentration data needing to calculate a signal correlation matrix, and calculating the concentration data time series of all the test channels which are correlated pairwise to obtain a corresponding concentration correlation coefficient matrix.

7. The near-infrared brain function imaging quality control method according to claim 3, wherein the power spectrum heartbeat peak detection includes: performing primary band-pass filtering of 0-3Hz on the optical density time-series data which preliminarily meet the quality requirement, and converting the filtered data into time series of oxyhemoglobin concentration and deoxyhemoglobin concentration according to a modified Beer-Lambert law;

Resampling the time series of the oxyhemoglobin concentration and the deoxyhemoglobin concentration to 5Hz, and then performing band-pass filtering of 0.01-2Hz on the time series of the resampled oxyhemoglobin concentration and deoxyhemoglobin concentration;

carrying out Fourier transform on the time series of the filtered oxyhemoglobin concentration and the filtered deoxyhemoglobin concentration, converting the time series of the oxyhemoglobin concentration and the deoxyhemoglobin concentration from a time domain signal to a frequency domain signal, and obtaining amplitudes corresponding to the frequencies of the oxyhemoglobin concentration and the deoxyhemoglobin concentration;

And performing modulus extraction on the amplitudes of the oxygenated hemoglobin concentration frequency curves and the deoxygenated hemoglobin concentration frequency curves, then performing square operation to obtain power corresponding to the frequencies, performing normalization on the power corresponding to the frequencies to obtain normalized power spectrums, and displaying the corresponding power spectrums with the frequencies of 0.5-1.5Hz of each test channel.

8. The method according to claim 1 or 2, wherein in the step 4), if the abnormal time point occurs within a plurality of seconds before the time series or within a plurality of seconds after the time series, the abnormal time point is determined as an instable data acquisition time point; and if the abnormal time point appears in the rest time periods, the abnormal time point is a motion artifact time point.

9. A near-infrared brain function imaging quality control system, comprising:

The near-infrared imaging module is used for acquiring original light intensity data and converting the original light intensity data into optical density data;

the standard deviation calculation module is used for converting the optical density data into optical density time series data and calculating the standard deviation of each sliding window by adopting a sliding window method for the optical density time series data;

The abnormal time point determining module is used for determining an abnormal time point according to the standard deviation data of the normal distribution;

and the first judgment module is used for selecting proper optical density time sequence length according to the abnormal time point data and judging whether the optical density time sequence data preliminarily meet the requirements or not according to the length of the time sequence.

10. the near-infrared brain function imaging quality control system according to claim 9, further comprising a secondary judgment module for performing signal-to-noise ratio detection, channel concentration signal correlation analysis and power spectrum heartbeat peak detection on the optical density time-series data preliminarily meeting the quality requirement.