阅读说明：本技术 一种脑部电阻抗磁感应多频成像数据前处理方法 (Brain electrical impedance magnetic induction multi-frequency imaging data preprocessing method ) 是由 宣和均 石崇源 谢家勋 于 2021-07-05 设计创作，主要内容包括：本发明公开了一种脑部电阻抗磁感应多频成像数据前处理方法,包括获取各频率对应的互感数据,获取算法处理前的频差数据,按照相对天线位置排列频差数据,获得每个发射天线对应的接收天线的频差数据的趋势信号并减去,把去除趋势信号的数据还原成绝对天线位置的排列顺序,得到处理完成后的数据步骤。本发明通过提取互感数据中的噪声信号的具体算法,去除差分数据中信号更强的噪声信号,使微弱的特征信号能够在和其接近的值域中凸显出来。其中按照相对天线位置排列频差数据,能让重新排列后的数据更符合物理意义,同时重新排列后的数据也更贴合二次函数曲线,二阶多项式拟合效果更好。(The invention discloses a brain electrical impedance magnetic induction multi-frequency imaging data preprocessing method, which comprises the steps of obtaining mutual inductance data corresponding to each frequency, obtaining frequency difference data before algorithm processing, arranging the frequency difference data according to relative antenna positions, obtaining trend signals of the frequency difference data of receiving antennas corresponding to each transmitting antenna, subtracting the trend signals, reducing the data with the trend signals removed into an arrangement sequence of absolute antenna positions, and obtaining the processed data. According to the method, the specific algorithm for extracting the noise signals in the mutual inductance data is adopted, the noise signals with stronger signals in the differential data are removed, and the weak characteristic signals can be highlighted in a value range close to the weak characteristic signals. The frequency difference data are arranged according to the relative antenna positions, the rearranged data can better accord with physical significance, meanwhile, the rearranged data are more fit with a quadratic function curve, and the second-order polynomial fitting effect is better.)

1. A brain electrical impedance magnetic induction multi-frequency imaging data preprocessing method is characterized by comprising the following steps:

1) acquiring mutual inductance data corresponding to each frequency;

2) acquiring frequency difference data before algorithm processing;

3) arranging the frequency difference data according to the relative antenna position;

4) obtaining and subtracting a trend signal of frequency difference data of a receiving antenna corresponding to each transmitting antenna;

5) and restoring the data of the trend signal removal into the arrangement sequence of the absolute antenna positions to obtain the processed data.

2. The brain electrical impedance magnetic induction multi-frequency imaging data preprocessing method of claim 1,

the acquiring of the mutual inductance data corresponding to each frequency specifically comprises: and subtracting the air mutual inductance data corresponding to each frequency from the detection mutual inductance data to obtain the mutual inductance data after basic calibration.

3. The method for pre-processing brain electrical impedance magnetic induction multi-frequency imaging data according to claim 1, wherein the frequency difference data is obtained by subtracting 2 frequencies of basic calibrated mutual inductance data.

4. The brain electrical impedance magnetic induction multi-frequency imaging data preprocessing method of claim 1, wherein the arranging the frequency difference data according to the relative antenna positions comprises: sorting according to the distance of the receiving antenna to the transmitting antenna, wherein n antenna data are n x n in total, removing the data of the same receiving transmitting antenna, arranging from near to far according to the position of the receiving antenna from the transmitting antenna from one direction, and rearranging the frequency difference data into n-1 receiving antennas corresponding to n transmitting antennas, n (n-1) data.

5. The brain electrical impedance magnetic induction multi-frequency imaging data preprocessing method according to claim 4, wherein the sequencing according to the distance from the receiving antenna to the transmitting antenna specifically comprises: the antenna numbers are 0-n-1, the data signal quality detection signals transmitted and received by the antennas do not penetrate through a target to a large extent, so that the target characteristic influence is basically avoided, the default is 0, the calculation is not involved, the receiving antenna sequence corresponding to the transmitting antenna number 0 is 1,2, 3 … … n-1, n, the receiving antenna sequence corresponding to the transmitting antenna number 1 is 2, 3, 4 … … n,0, the receiving antenna sequence corresponding to the transmitting antenna number 2 is 3, 4, 5 … … 0, 1, and the receiving antenna sequence corresponding to the transmitting antenna number n-1 is 0, 1,2 … … n-3 and n-2.

6. A brain electrical impedance magnetic induction multi-frequency imaging data preprocessing method according to claim 1,2, 3, 4 or 5, characterized in that the trend signal of the frequency difference data of the receiving antenna corresponding to each transmitting antenna is obtained and subtracted specifically as follows: and respectively carrying out 2-order polynomial fitting on the receiving antenna data corresponding to each transmitting antenna to a y-x function (x is 1, 2.., n), obtaining a trend line corresponding to each antenna, and subtracting the calculated trend from the original data.

7. The brain electrical impedance magnetic induction multi-frequency imaging data preprocessing method according to claim 6, wherein the data of trend signal removal is reduced to the arrangement sequence of absolute antenna positions, and the obtained processed data is specifically: arranging original n x (n-1) data according to a sequence from a number 0 to a number n-1 receiving antenna, setting the data of the transmitting and receiving antenna with the same serial number as 0: for example, the transmission No. 0 corresponds to the reception No. 0-n-1, the transmission No. 1 corresponds to the reception No. 0-n-1, and so on, to obtain n × n data after the preprocessing.

Technical Field

The invention belongs to the field of multi-frequency imaging, and particularly relates to a pretreatment method of brain electrical impedance magnetic induction multi-frequency imaging data.

Background

In the prior art, the brain electrical impedance magnetic induction multi-frequency imaging is to directly acquire mutual inductance data of 2 frequencies and directly perform imaging by using the difference value of the mutual inductance data. And carrying out polynomial fitting on the mutual inductance data of one frequency to the mutual inductance data of the other frequency to enable the value ranges of the mutual inductance data to be close, imaging by using the fitted frequency difference, and directly using the difference between the frequencies or the difference after the polynomial fitting to extract a tiny signal of focus change in a complex environment such as a human brain. The existing imaging method has low resolution, and the imaging cannot be clearly realized under the influence of strong noise due to the fact that weak focus signals are submerged.

FIG. 1 is a graph of two frequency direct differential data; FIG. 2 is the difference data after two frequency polynomials have been fitted; it can be seen that in the prior art, the amplitude of the data obtained using the difference between the frequencies is about 0.5, the amplitude of the data obtained using the polynomial fitting method is about 0.4, and the signal of lesion change is about 0.1, with signal intensities differing by a factor of 4-5, causing the lesion signal to be buried in strong noise. After the data processing method is used, noise is used as a main imaging object in imaging, and weak signals of focuses cannot be reflected.

Disclosure of Invention

In order to solve the technical problems in the prior art, the invention provides a brain electrical impedance magnetic induction multi-frequency imaging data preprocessing method which removes noise signals with large interference on imaging, reserves weak characteristic signals and improves the signal-to-noise ratio on the premise of not influencing the characteristics of data.

In order to realize the purpose of the invention, the invention is realized by the following technical scheme:

the invention discloses a brain electrical impedance magnetic induction multi-frequency imaging data preprocessing method, which comprises the following steps:

1) acquiring mutual inductance data corresponding to each frequency;

2) acquiring frequency difference data before algorithm processing;

3) arranging the frequency difference data according to the relative antenna position;

4) obtaining and subtracting a trend signal of frequency difference data of a receiving antenna corresponding to each transmitting antenna;

5) and restoring the data of the trend signal removal into the arrangement sequence of the absolute antenna positions to obtain the processed data.

As a further improvement, the obtaining of the mutual inductance data corresponding to each frequency specifically includes: and subtracting the air mutual inductance data corresponding to each frequency from the detection mutual inductance data to obtain the mutual inductance data after basic calibration.

As a further improvement, the frequency difference data of the present invention is obtained by subtracting the mutual inductance data after the basic calibration of 2 frequencies.

As a further improvement, the frequency difference data arranged according to the relative antenna positions according to the invention is: sorting according to the distance of the receiving antenna to the transmitting antenna, wherein n antenna data are n x n in total, removing the data of the same receiving transmitting antenna, arranging from near to far according to the position of the receiving antenna from the transmitting antenna from one direction, and rearranging the frequency difference data into n-1 receiving antennas corresponding to n transmitting antennas, n (n-1) data.

As a further improvement, the invention specifically comprises the following steps according to the distance sequence of the receiving antenna to the transmitting antenna: the antenna numbers are 0-n-1, the data signal quality detection signals transmitted and received by the antennas do not penetrate through a target to a large extent, so that the target characteristic influence is basically avoided, the default is 0, the calculation is not involved, the receiving antenna sequence corresponding to the transmitting antenna number 0 is 1,2, 3 … … n-1, n, the receiving antenna sequence corresponding to the transmitting antenna number 1 is 2, 3, 4 … … n,0, the receiving antenna sequence corresponding to the transmitting antenna number 2 is 3, 4, 5 … … 0, 1, and the receiving antenna sequence corresponding to the transmitting antenna number n-1 is 0, 1,2 … … n-3 and n-2.

As a further improvement, the present invention obtains a trend signal of the frequency difference data of the receiving antenna corresponding to each transmitting antenna and subtracts the trend signal specifically as follows: and respectively carrying out 2-order polynomial fitting on the receiving antenna data corresponding to each transmitting antenna to a y-x function (x is 1, 2.., n), obtaining a trend line corresponding to each antenna, and subtracting the calculated trend from the original data.

As a further improvement, the data of the trend-removed signal is restored to the arrangement sequence of the absolute antenna positions, and the obtained processed data is specifically: arranging original n x (n-1) data in sequence from No. 0 to No. n-1 receiving antennas, setting the data of the transmitting and receiving antennas with the same serial number as 0: for example, the number 0 transmission corresponds to the number 0-n-1 reception, the number 1 transmission corresponds to the number 0-n-1 reception, and so on, to obtain n × n data after the preprocessing is completed.

The invention has the following beneficial effects:

according to the method, the specific algorithm for extracting the noise signals in the mutual inductance data is adopted, the noise signals with stronger signals in the differential data are removed, and the weak characteristic signals can be highlighted in a value range close to the weak characteristic signals. Specifically, the noise signal with larger amplitude in the acquired mutual inductance data difference values of 2 frequencies is removed. Because the characteristic signal and the noise signal are in a superposition state, the noise signal ratio is far greater than that of the characteristic signal, and the extraction of the characteristic signal cannot be influenced by removing the noise signal with larger amplitude according to the integral trend. The data is more fit to an imaging algorithm, and the proportion of noise in the image is greatly reduced.

The frequency difference data are arranged according to the relative antenna positions, the rearranged data can better accord with physical significance, meanwhile, the rearranged data are more fit with a quadratic function curve, and the second-order polynomial fitting effect is better.

Drawings

FIG. 1 is a graph of two frequency direct differential data;

FIG. 2 is a graph of difference data after two frequency polynomials have been fitted;

FIG. 3 is a graph comparing 15 raw data and fitted data for transmission # 0;

FIG. 4 is a data plot of 15 of transmission # 0 after algorithm processing;

FIG. 5 is a comparison of all raw data and fitted data;

FIG. 6 is a graph of the overall data after algorithm processing;

FIG. 7 is a CT view of a patient with cerebral hemorrhage;

FIG. 8 is a graph of direct 2-frequency differential imaging without processing;

FIG. 9 is a polynomial fit plot of 2 frequency data;

fig. 10 is an image using the algorithm of the present invention.

Detailed Description

The invention discloses a brain electrical impedance magnetic induction multi-frequency imaging data preprocessing method, which comprises the following steps:

1) acquiring mutual inductance data corresponding to each frequency;

2) acquiring frequency difference data before algorithm processing;

3) arranging the frequency difference data according to the relative antenna position;

4) obtaining and subtracting a trend signal of frequency difference data of a receiving antenna corresponding to each transmitting antenna;

5) and restoring the data of the trend signal removal into the arrangement sequence of the absolute antenna positions to obtain the processed data.

Specifically, the method comprises the following steps:

1) acquiring mutual inductance data corresponding to each frequency: and subtracting the air reference mutual inductance data corresponding to each frequency from the detection mutual inductance data to obtain the mutual inductance data after basic calibration.

2) Acquiring frequency difference data before algorithm processing: subtracting the mutual inductance data after the basic calibration corresponding to the 2 frequencies to obtain frequency difference data before algorithm processing, wherein for example, the frequency 1 data is f1, the frequency 2 data is f2, and the frequency 2-f1 are obtained.

3) The sequencing according to the distance from the receiving antenna to the transmitting antenna specifically comprises the following steps:

n antenna data are n x n in total, data of the same receiving and transmitting antenna are removed, and the data are arranged from near to far from one direction according to the position of the receiving antenna from the transmitting antenna (the antenna number is 0-n-1, and the data signal quality detection signals transmitted and received by the same transmitting antenna do not penetrate through a target so that the target characteristic influence is basically avoided, the default is 0, and the data do not participate in calculation): the sequence of the receiving antenna corresponding to the transmitting antenna No. 0 is 1,2, 3 … … n-1, n, the sequence of the receiving antenna corresponding to the transmitting antenna No. 1 is 2, 3, 4 … … n, the sequence of the receiving antenna corresponding to the transmitting antenna No. 2 is 3, 4, 5 … … 0, 1, and so on, and the sequence of the receiving antenna corresponding to the transmitting antenna No. n-1 is 0, 1,2 … … n-3, n-2. And rearranging the frequency difference data into n transmitting antennas corresponding to n-1 receiving antennas (n-1) data. For example, 16 antenna data are 16 × 16 in total, data of the same receiving and transmitting antenna are removed, and the data are arranged from near to far from one direction according to the positions of the receiving antenna and the transmitting antenna (the antenna number is 0-15, and the data signal quality detection signal transmitted and received by the same transmitting antenna does not penetrate through the target so that the target characteristics are not affected basically, the default is 0, and the data do not participate in calculation): the sequence of the receiving antenna corresponding to the transmitting antenna No. 0 is 1,2, 3 … … 14, 15, the sequence of the receiving antenna corresponding to the transmitting antenna No. 1 is 2, 3, 4 … … 15, 0, the sequence of the receiving antenna corresponding to the transmitting antenna No. 2 is 3, 4, 5 … … 0, 1, and so on, and the sequence of the receiving antenna corresponding to the transmitting antenna No. 15 is 0, 1,2 … … 13, 14. The rearranged frequency difference data is that 16 transmitting antennas correspond to 15 receiving antennas, namely 240 data.

4) Obtaining a trend signal of frequency difference data of a receiving antenna corresponding to each transmitting antenna and subtracting the trend signal specifically as follows:

and respectively carrying out 2-order polynomial fitting on the receiving antenna data corresponding to each transmitting antenna to a y-x function (x is 1, 2.., n), obtaining a trend line corresponding to each antenna, and subtracting the calculated trend from the original data. Fig. 3 is a comparison graph of 15 original data of number 0 transmission and fitted data, where number 1-15 reception corresponding to number 0 transmission is selected, and the rest 15 transmissions are analogized, it can be seen that the amplitude of the overall trend is about 0.5, the amplitude of the original data is about 0.6, and the amplitude proportion of the trend exceeds 80%.

Fig. 4 is a plot of data for 15 shots from transmission No. 0 after algorithm processing, with raw data minus fitted trends.

FIG. 5 is a comparison graph of the total raw data and the fitted data, which is a comparison of the raw data and the fitted data trend signal curves.

5) Restoring the data of the trend signal removal into the arrangement sequence of the absolute antenna positions, and obtaining the processed data specifically as follows: arranging original n x (n-1) data in sequence from No. 0 to No. n-1 receiving antennas, setting the data of the transmitting and receiving antennas with the same serial number as 0: for example, the number 0 transmission corresponds to the number 0-n-1 reception, the number 1 transmission corresponds to the number 0-n-1 reception, and so on, to obtain n × n data after the preprocessing is completed.

Such as: arranging the original 240 data according to the sequence from No. 0 to No. 15 receiving antennas, setting the data of the transmitting and receiving antennas with the same serial number as 0: for example, the transmission of number 0 corresponds to the reception of numbers 0 to 15, the transmission of number 1 corresponds to the reception of numbers 0 to 15, and so on, to obtain 256 pieces of data after the preprocessing.

The processed data is as shown in fig. 6, fig. 6 is a graph of all data processed by the algorithm, the data fluctuation oscillates up and down on a straight line with y equal to 0, no obvious overall fluctuation exists, and small signals in the data are highlighted.

The specific application is as follows: carrying out brain electrical impedance magnetic induction multi-frequency imaging on patients with cerebral hemorrhage:

FIG. 7 is a CT image of a patient with cerebral hemorrhage, wherein the lesion is a left white block area corresponding to the curve of FIG. 1;

FIG. 8 is a graph obtained by dividing 2 frequency differences into two images without processing, wherein the lesion site cannot be observed according to the curve of FIG. 2;

FIG. 9 is a polynomial fit plot of 2 frequency data, corresponding to the curve of FIG. 6, with a small lesion location and not primary imaging;

FIG. 10 is an image of an algorithm of the present invention that accurately shows the location of a lesion, consistent with CT.

It is obvious that the direct difference and direct polynomial fitting images in imaging are mainly dark images of noise and cannot correspond to the focus part, and although the images after polynomial fitting have a small part of light-colored focus images at the upper left, the contrast noise is very small. The current algorithm can eliminate the dark color image of noise under a large condition and highlight the light color image of a focus signal.

The above description is only a preferred embodiment of the present invention, and the present invention is not limited to the above embodiments, and other modifications and variations directly derived or suggested by those skilled in the art without departing from the spirit and concept of the present invention should be considered to be included in the protection scope of the present invention.