Electrical impedance tomography method, device, terminal and medium based on sparse recovery

Abstract

The electrical impedance tomography method, device, terminal and medium based on sparse recovery provided by the present invention relate to the field of medical imaging technology. The method comprises: calculating the boundary voltage difference according to the boundary voltage of the object to be measured obtained, which includes noise-free boundary voltage difference, Gaussian noise and sparse abnormal interference, constructing the target relationship between the boundary voltage difference and the conductivity change, based on the target relationship, and using the L0 norm and L2 norm to construct the L0 norm term and the L2 norm term, solving the conductivity change and sparse abnormal interference in the optimization problem constructed based on the L0 norm term and the L2 norm term, and then performing electrical impedance imaging based on the solved target conductivity change. The present invention models the boundary voltage change caused by body motion interference or poor electrode contact as non-Gaussian noise, uses different norm constraints according to the characteristics of different interferences, and can obtain high-quality EIT imaging when there is a serious error in the impedance boundary voltage due to non-Gaussian interference.

Description

Technical Field

[0001] The present invention relates to the field of medical imaging technology, and in particular to an electrical impedance tomography method, device, terminal and medium based on sparse recovery. Background Art

[0002] As a non-invasive imaging method, chest EIT (Electrical Impedance Tomography) is an ill-conditioned and underdetermined mathematical inverse problem, which makes it highly sensitive to measurement noise. Even small abnormal changes in boundary voltage can have a great impact on the imaging results. Since chest EIT is susceptible to various clinical interferences, this sensitivity makes EIT difficult to standardize as a medical tool.

[0003] Interference caused by poor electrode contact, patient movement, medical intervention or nursing procedures can induce severe fluctuations in the measured voltage. These fluctuations destroy the respiratory system EIT impedance signal, thus posing challenges to the interpretation and evaluation of EIT data. For example, patients need to be examined after lung parenchymal resection to evaluate changes in postoperative lung ventilation. Despite the use of minimally invasive techniques, patient movement caused by incision pain after anesthesia leads to serious errors in the EIT boundary voltage, thus affecting the evaluation of ventilation indicators based on EIT. For example, in clinical trials using EIT to evaluate patients with chronic obstructive pulmonary disease, patients need to exhale forcefully during the standard forced vital capacity (FVC) test, which leads to obvious chest movement. This movement may destroy the electrode contact and strongly interfere with the EIT boundary voltage, resulting in severe artifacts in the EIT image, thus affecting the accurate calculation of the lung function index. Therefore, it is necessary to enhance the robustness of EIT by reducing the impact of strong interference on the EIT image for clinical application.

[0004] Currently, in order to reduce interference and improve the robustness of imaging, time series reconstruction is combined with electrode motion compensation to improve image quality when multiple electrodes move. However, this method cannot effectively suppress severe electrode interference. Or, in order to offset strong interference, the L1 norm is used instead of the L2 norm to minimize the estimation error, thereby obtaining a robust EIT imaging method. This algorithm has good imaging performance under low-intensity interference conditions, but the imaging quality is severely degraded under high-intensity interference conditions.

[0005] In summary, the existing EIT imaging methods can obtain high-quality EIT images in the absence of clinical interference such as pulse noise. However, when there is high-intensity non-Gaussian interference such as pulse noise caused by body motion interference or poor electrode release, the EIT imaging performance will be reduced and high-quality EIT images cannot be obtained.

[0006] Therefore, how to provide a solution to the above technical problems is a problem that those skilled in the art need to solve at present. Summary of the invention

[0007] The technical problem to be solved by the present invention is to provide an electrical impedance tomography method, device, terminal and medium based on sparse recovery in view of the above-mentioned defects of the prior art, which can obtain robust and high-quality EIT imaging when there are serious errors in the impedance boundary voltage due to non-Gaussian interference.

[0008] The technical solution adopted by the present invention to solve the technical problem is as follows:

[0009] An electrical impedance tomography method based on sparse recovery, wherein the method comprises:

[0010] Obtaining a boundary measurement voltage of a test area of ​​the test object, and calculating a corresponding boundary voltage difference according to the boundary measurement voltage; wherein the boundary voltage difference includes a noise-free boundary voltage difference, Gaussian noise, and sparse abnormal interference to be solved;

[0011] Constructing a relationship between the boundary voltage difference and the conductivity change to be solved, and obtaining a corresponding target relationship;

[0012] Based on the target relationship, an L0 norm term and an L2 norm term are constructed by using the L0 norm and the L2 norm;

[0013] Constructing a corresponding optimization problem based on the L0 norm term and the L2 norm term, and solving the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem to determine a target conductivity change;

[0014] Electrical impedance imaging is performed based on the target conductivity change to obtain corresponding imaging results.

[0015] In one implementation, the boundary voltage difference is: ;

[0016] Furthermore, the target relationship between the boundary voltage difference and the conductivity change to be solved is: ;

[0017] in, Indicates time point Moments and time points The boundary voltage difference at time Indicates time point The boundary measurement voltage at the moment, Indicates time point The boundary measurement voltage at the moment, represents the noise-free boundary voltage difference, represents Gaussian noise, represents the sparse abnormal interference to be solved, represents the conductivity change to be solved, represents the Jacobian matrix.

[0018] In one implementation, based on the target relationship, L0 norm term and L2 norm term are constructed by using L0 norm and L2 norm, including:

[0019] Determining a target term representing the Gaussian noise based on the target relationship;

[0020] Based on the target term, a first L2 norm term is constructed using the L2 norm;

[0021] Based on the conductivity change to be solved, a second L2 norm term is constructed using the L2 norm;

[0022] Using the L0 norm to constrain the sparse abnormal interference to be solved to construct an L0 norm term based on the constrained sparse abnormal interference;

[0023] In one implementation, constructing a corresponding optimization problem based on the L0 norm term and the L2 norm term includes:

[0024] Constructing a corresponding optimization problem based on the L0 norm, the first L2 norm term, and the second L2 norm term;

[0025] The optimization problem is:

[0026]

[0027]

[0028] ;

[0029] in, represents the first L2 norm term, represents the second L2 norm term, represents the L0 norm term, represents the first penalty parameter and greater than zero, represents the second penalty parameter and greater than zero, represents the inverse matrix of the covariance of Gaussian noise, express The matrix decomposition factor of express The transposed matrix of represents the inverse matrix of the covariance of the desired conductivity, express The matrix decomposition factor of express The transposed matrix of .

[0030] In one implementation, solving the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem to determine a target conductivity change includes:

[0031] Based on a preset nearest neighbor alternating minimization method, the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem are alternately solved to determine a target conductivity change.

[0032] In one implementation, the alternately solving the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem based on a preset nearest neighbor alternating minimization method to determine a target conductivity change includes:

[0033] Based on the conductivity change to be solved and the sparse abnormal interference to be solved, and using the L2 norm, a first nearest neighbor constraint term and a second nearest neighbor constraint term are constructed;

[0034] Based on the optimization problem, a first linear least squares problem and a second linear least squares problem are constructed by using the first nearest neighbor constraint item and the second nearest neighbor constraint item;

[0035] Alternately solving the conductivity change to be solved in the first linear least squares problem and the sparse abnormal interference to be solved in the second linear least squares problem to determine a target conductivity change;

[0036] Among them, the first nearest neighbor constraint item is: ;

[0037] The second nearest neighbor constraint is: ;

[0038] The first linear least squares problem is:

[0039] ;

[0040] The second linear least squares problem is:

[0041] ;

[0042] in, Indicates The conductivity change matrix of the iterative optimization is: Indicates The conductivity change matrix of the iterative optimization is: Indicates Sparse anomaly interference of iterative optimization, Indicates Sparse anomaly interference of iterative optimization, represents the neighbor parameter and Greater than zero.

[0043] In one implementation, solving the conductivity change to be solved in the first linear least squares problem includes:

[0044] Constructing a corresponding closed-form solution based on the first linear least squares problem, and solving the conductivity change to be solved based on the closed-form solution;

[0045] Among them, the closed-form solution is:

[0046] ;

[0047] and, represents the transposed matrix of the Jacobian matrix, Represents the identity matrix.

[0048] In one implementation, solving the sparse abnormal interference to be solved in the second linear least squares problem includes:

[0049] scalarizing the second linear least squares problem to obtain a corresponding scalar formula;

[0050] Based on the scalar formula, and using the preset The rule solves the sparse abnormal interference to be solved;

[0051] Wherein, the scalar formula is:

[0052] ;

[0053] ;

[0054] ;

[0055] in, express No. item, Indicates sparse abnormal interference No. item, Indicates Sparse anomaly interference for sub-iterative optimization No. item, Indicates Sparse anomaly interference for sub-iterative optimization No. item, express The Line Column elements, express The square root of .

[0056] The present invention also discloses an electrical impedance tomography device based on sparse recovery, wherein the device comprises:

[0057] A boundary voltage acquisition module is used to obtain the boundary measurement voltage of the test area of ​​the test object;

[0058] A voltage difference calculation module, used to calculate a corresponding boundary voltage difference according to the boundary measurement voltage; wherein the boundary voltage difference includes a noise-free boundary voltage difference, Gaussian noise and a sparse abnormal interference to be solved;

[0059] A relationship building module, used to build a relationship between the boundary voltage difference and the conductivity change to be solved, and obtain a corresponding target relationship;

[0060] A norm term construction module, used to construct an L0 norm term and an L2 norm term based on the target relationship and using an L0 norm and an L2 norm;

[0061] An optimization problem construction module, used for constructing a corresponding optimization problem based on the L0 norm term and the L2 norm term;

[0062] A problem solving module, used for solving the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem to determine a target conductivity change;

[0063] The electrical impedance imaging module is used to perform electrical impedance imaging based on the target conductivity change to obtain corresponding imaging results.

[0064] The present invention also discloses a terminal, which includes: a memory, a processor, and a sparse recovery-based electrical impedance tomography program stored in the memory and executable on the processor, wherein the sparse recovery-based electrical impedance tomography program, when executed by the processor, implements the steps of the sparse recovery-based electrical impedance tomography method as described above.

[0065] The present invention also discloses a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program can be executed to implement the steps of the electrical impedance tomography method based on sparse recovery as described above.

[0066] The present invention provides an electrical impedance tomography method, device, terminal and medium based on sparse recovery. The electrical impedance tomography method based on sparse recovery includes: obtaining the boundary measurement voltage of the test area of ​​the test body, and calculating the corresponding boundary voltage difference according to the boundary measurement voltage; wherein the boundary voltage difference includes the noise-free boundary voltage difference, Gaussian noise and the sparse abnormal interference to be solved; constructing the relationship between the boundary voltage difference and the conductivity change to be solved to obtain the corresponding target relationship; based on the target relationship, and using the L0 norm and L2 norm to construct the L0 norm term and the L2 norm term; constructing the corresponding optimization problem based on the L0 norm term and the L2 norm term, and solving the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem to determine the target conductivity change; performing electrical impedance imaging based on the target conductivity change to obtain the corresponding imaging result. It can be seen that the present invention models the boundary voltage change caused by body motion interference or poor electrode contact as non-Gaussian noise. First, the boundary voltage difference is calculated based on the acquired boundary voltage of the body to be measured, and the boundary voltage difference is divided into three parts, namely, noise-free boundary voltage difference, Gaussian noise, and sparse abnormal interference, that is, the external interference is divided into Gaussian noise and sparse abnormal value interference, and the external interference is classified and processed. Then, the target relationship between the boundary voltage difference and the conductivity change to be solved is constructed. Based on the target relationship, the L2 norm and the L0 norm are used for constraints to construct a corresponding optimization problem, and then the optimization problem is solved to determine the target conductivity change. Finally, electrical impedance imaging is performed based on the conductivity change, thereby solving the technical problem of reduced EIT imaging performance when there is body motion interference or poor electrode contact in the clinical environment, that is, the technical solution of the present application can have good EIT imaging performance under different interference conditions, and can obtain robust and high-quality EIT imaging results when there is a serious error in the impedance boundary voltage due to the presence of high-intensity non-Gaussian interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a flow chart of a preferred embodiment of the electrical impedance tomography method based on sparse recovery in the present invention;

[0068] Figure 2 This is a schematic diagram of an electrode belt worn by a human body disclosed in the present invention;

[0069] Figure 3 is a schematic diagram of an EIT image reconstructed by different EIT imaging algorithms on an experimental object 1 disclosed in the present invention;

[0070] Figure 4 is a schematic diagram of an EIT image reconstructed by different EIT imaging algorithms on an experimental object 2 disclosed in the present invention;

[0071] Figure 5 It is a functional principle block diagram of a preferred embodiment of an electrical impedance tomography device based on sparse recovery in the present invention;

[0072] Figure 6 It is a functional principle block diagram of a preferred embodiment of the terminal in the present invention. DETAILED DESCRIPTION

[0073] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0074] See also Figure 1 , Figure 1 Flow chart of the electrical impedance tomography method based on sparse recovery in the present invention. Figure 1 As shown, the electrical impedance tomography method based on sparse recovery described in the embodiment of the present invention includes:

[0075] Step S11, obtaining the boundary measurement voltage of the test area of ​​the test object, and calculating the corresponding boundary voltage difference according to the boundary measurement voltage; wherein the boundary voltage difference includes the noise-free boundary voltage difference, Gaussian noise and sparse abnormal interference to be solved.

[0076] In this embodiment, the boundary measurement voltage measured in the test area on the test object is obtained, and then the corresponding boundary voltage difference, that is, the boundary voltage change, is calculated according to the boundary measurement voltage, wherein the boundary voltage difference includes the noise-free boundary voltage difference, Gaussian noise and sparse abnormal interference, that is, the boundary voltage difference is composed of the noise-free boundary voltage difference, Gaussian noise and sparse abnormal interference. It can be understood that the external interference is divided into Gaussian noise and sparse abnormal value interference, and the external interference is classified and processed.

[0077] It should be pointed out that in the existing EIT imaging technology, external interference is classified into one category, and no classification processing is performed on the external interference, resulting in the existing EIT imaging technology having good imaging performance under low-intensity interference conditions, but the imaging quality decreases under high-intensity interference conditions. However, in this embodiment, the additive noise is expressed as two components, namely Gaussian noise (dense Gaussian component) and sparse abnormal interference (sparse abnormal component), to achieve classified processing of external interference.

[0078] For example, see Figure 2 As shown, 16 electrodes are placed on the surface of the test area on the test object, one pair of opposite electrodes generates current excitation, and the remaining 12 pairs of adjacent electrodes measure the surface voltage. Then, the excitation pairs are turned on until all the opposite electrode pairs are used as excitations, thereby obtaining the voltage measured at the measurement time point. The boundary voltage value is measured, and then the boundary voltage difference between the two measured time points is calculated to establish a boundary voltage change model at the two measured time points, namely, the boundary voltage change model, wherein the boundary voltage difference is divided into three parts, namely, the noise-free boundary voltage difference , Gaussian noise and sparse anomaly interference ,and , ,as well as .

[0079] The boundary voltage difference is calculated as follows: ;

[0080] in, Indicates time point Moments and time points The boundary voltage difference at time Indicates time point The boundary measurement voltage at the moment, Indicates time point The boundary measurement voltage at the moment, represents the noise-free boundary voltage difference, represents Gaussian noise, represents the sparse abnormal interference to be solved.

[0081] Step S12: constructing a relationship between the boundary voltage difference and the conductivity change to be solved, and obtaining a corresponding target relationship.

[0082] In this embodiment, after the boundary voltage difference between two measurement time points is calculated, a relationship between the boundary voltage difference and the conductivity change to be solved is constructed to obtain a corresponding target relationship.

[0083] Furthermore, the target relationship between the boundary voltage difference and the conductivity change to be solved can be: ;

[0084] in, represents the conductivity change to be solved, represents the Jacobian matrix.

[0085] Step S13: Based on the target relationship, L0 norm term and L2 norm term are constructed by using L0 norm and L2 norm.

[0086] In this embodiment, the external interference is divided into Gaussian noise and sparse outlier interference. After the external interference is classified and processed, different norms are used to constrain the characteristics of different interferences, that is, based on the target relationship between the constructed boundary voltage difference and the conductivity change to be solved, the L0 norm and L2 norm are constructed using the L0 norm and L2 norm. Specifically, based on the target relationship, the target term representing Gaussian noise is determined; based on the target term, the first L2 norm term (i.e., Gaussian noise suppression term) is constructed using the L2 norm; based on the conductivity change to be solved, the second L2 norm term is constructed using the L2 norm; the sparse abnormal interference to be solved is constrained using the L0 norm to construct the L0 norm term (i.e., sparse abnormal interference penalty term) based on the constrained sparse abnormal interference. It can be understood that the L2 norm is used to suppress Gaussian noise, and in order to describe sparsity, the non-convex and non-differentiable L0 norm is used to constrain sparse abnormal interference, so as to resist serious errors. That is, the L2 norm is used to constrain Gaussian noise, and the L0 norm is used to constrain sparse abnormal interference, so that the sparse abnormal interference It is sparse. It should be pointed out that since EIT has a high time resolution and the conductivity changes smoothly between two time measurement points, the L2 norm can also be used to constrain the conductivity change rate.

[0087] Step S14: construct a corresponding optimization problem based on the L0 norm term and the L2 norm term, and solve the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem to determine a target conductivity change.

[0088] In this embodiment, based on the target relationship between the constructed boundary voltage difference and the conductivity change to be solved, and after constructing the L0 norm term and the L2 norm term using the L0 norm and the L2 norm term, a corresponding optimization problem is constructed based on the L0 norm term and the L2 norm term. Specifically, a corresponding optimization problem is constructed based on the L0 norm, the first L2 norm term and the second L2 norm term. It can be understood that the EIT imaging task of the present application can be expressed as a non-convex optimization problem consisting of an L0 norm term and an L2 norm term.

[0089] It should be pointed out that one subproblem in the non-convex optimization problem contains an L2 norm term, and the other subproblem contains an L0 norm regularization term (L0 norm term). The Gauss-Newton (GN) method can be used to solve the subproblem containing the L2 norm term, and the Projected Gradient Descent (PGD) method can be used to directly solve the other subproblem containing the L0 norm regularization term. The Gauss-Newton method is an iterative algorithm for solving nonlinear least squares problems, that is, by iteratively updating parameters, the objective function gradually approaches the minimum value; the Projected Gradient Descent method is a simple method for solving convex optimization problems, which is applicable to problems on arbitrary domains. PGD iteratively updates parameters to gradually approach the optimal solution.

[0090] The optimization problem is:

[0091]

[0092]

[0093] ;

[0094] in, represents the first L2 norm term, represents the second L2 norm term, represents the L0 norm term, represents the conductivity change to be solved and , represents the first penalty parameter and greater than zero, represents the second penalty parameter and greater than zero, represents the inverse matrix of the covariance of Gaussian noise and , express The matrix decomposition factor of , express The transposed matrix of represents the Jacobian matrix and , represents the inverse matrix of the covariance of the desired conductivity and , express The matrix decomposition factor of , express The transposed matrix of . And, Can be used to control sparse abnormal interference The degree of sparseness.

[0095] In this embodiment, after obtaining a non-convex optimization problem consisting of an L0 norm term and an L2 norm term, the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem can be solved to determine the target conductivity change. Specifically, the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem are alternately solved based on a preset neighbor alternating minimization method to determine the target conductivity change. Among them, the basic principle of the neighbor alternating minimization method is to gradually approach the optimal solution by alternately minimizing two sub-problems. It can be understood that for the two variables in the formula corresponding to the above optimization problem, the parts related to the conductivity change and the sparse abnormal interference can be extracted respectively, and the corresponding neighbor constraint terms are added respectively to alternately optimize the conductivity change and the sparse abnormal interference, that is, the neighbor alternating minimization (PAM) method is used as a solution method, thereby generating an alternating optimization process.

[0096] Moreover, the specific process of alternately solving the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem based on the preset nearest neighbor alternating minimization method is as follows: based on the conductivity change to be solved and the sparse abnormal interference to be solved, the first nearest neighbor constraint term and the second nearest neighbor constraint term are constructed using the L2 norm; based on the optimization problem, the first linear least squares problem and the second linear least squares problem are constructed using the first nearest neighbor constraint term and the second nearest neighbor constraint term; the conductivity change to be solved in the first linear least squares problem and the sparse abnormal interference to be solved in the second linear least squares problem are alternately solved to determine the target conductivity change.

[0097] Among them, the first nearest neighbor constraint is: ;

[0098] The second nearest neighbor constraint is: ;

[0099] The first linear least squares problem is:

[0100] ;

[0101] The second linear least squares problem is:

[0102] ;

[0103] in, Indicates The conductivity change matrix of the iterative optimization is: Indicates The conductivity change matrix of the iterative optimization is: Indicates Sparse anomaly interference of iterative optimization, Indicates Sparse anomaly interference of iterative optimization, represents the neighbor parameter and Greater than zero.

[0104] Wherein, solving the conductivity change to be solved in the first linear least squares problem may specifically include: constructing a corresponding closed-form solution based on the first linear least squares problem, and solving the conductivity change to be solved based on the closed-form solution.

[0105] The closed-form solution is:

[0106] ;

[0107] represents the transposed matrix of the Jacobian matrix, Represents the identity matrix.

[0108] It should be pointed out that When the value is set to a very small value, it is convenient to analyze the convergence and the conductivity change is calculated for the formula corresponding to the first linear least squares problem. Taking the derivative and setting the gradient to 0, we can get the formula corresponding to the above closed-form solution.

[0109] Furthermore, solving the sparse abnormal interference to be solved in the second linear least squares problem may specifically include: scalarizing the second linear least squares problem to obtain a corresponding scalar formula, and then based on the scalar formula and using a preset The rules solve the sparse abnormal interference to be solved.

[0110] The scalar formula is:

[0111] ;

[0112] ;

[0113] ;

[0114] in, express No. item, Indicates sparse abnormal interference No. item, Indicates Sparse anomaly interference for sub-iterative optimization No. item, Indicates Sparse anomaly interference for sub-iterative optimization No. item, express The Line Column elements, express The square root of .

[0115] It should be pointed out that when solving the second linear least squares problem, it is necessary to first analyze , to transform the second linear least squares problem into a solvable problem, where, due to is a diagonal matrix, so Matrix decomposition factor is also a diagonal matrix, that is , so the formula corresponding to the second linear least squares problem can be equivalent to the above scalar formula, and in the scalar formula The range is between 1 and 192, that is, .

[0116] And the solution to the scalar expression can be obtained by the following formula, namely:

[0117] ;

[0118] Since the solution of this formula is subject to , so when solving this formula, we can first determine , will soon If it is set to 0, the formula is:

[0119] ;

[0120] And, in order to obtain a robust confidence interval in the presence of outliers, the standard deviation can be calculated using the normalized median absolute deviation, that is:

[0121] ;

[0122] Among them, Med( ) is the median operator.

[0123] Then, Updated to:

[0124] ;

[0125] ;

[0126] in, represents a scaling factor that controls the range of the confidence interval, and Greater than zero, that is .

[0127] Finally, using The rule performs outlier detection to determine the abnormal values ​​in order to filter out the abnormal values ​​in the solution results. Among them, since 99.7% of the values ​​fall within So it can be determined that it falls within Values ​​outside of this range are outliers.

[0128] Step S15: performing electrical impedance imaging based on the target conductivity change to obtain corresponding imaging results.

[0129] In this embodiment, after solving the optimization problem to obtain the target resistivity change, electrical impedance imaging can be performed based on the target conductivity change to obtain the corresponding imaging result, that is, the target conductivity change is converted into an EIT image that can display the area to be tested. For example, for the specific scenario of chest lung ventilation monitoring, the target conductivity is converted into an EIT image that can display the ventilation lung area.

[0130] It can be seen that in the embodiment of the present invention, the boundary voltage change caused by body motion interference or poor electrode contact is modeled as non-Gaussian noise. First, the boundary voltage difference is calculated according to the acquired boundary voltage of the body to be measured, and the boundary voltage difference is divided into three parts, namely, noise-free boundary voltage difference, Gaussian noise, and sparse abnormal interference, that is, the external interference is divided into Gaussian noise and sparse abnormal value interference, and the external interference is classified and processed. Then, a target relationship between the boundary voltage difference and the conductivity change to be solved is constructed. Based on the target relationship, the L2 norm and the L0 norm are used for constraints to construct a corresponding optimization problem, and then the optimization problem is solved to determine the target conductivity change. Finally, electrical impedance imaging is performed based on the conductivity change, thereby solving the technical problem of decreased EIT imaging performance when there is body motion interference or poor electrode contact in a clinical environment. That is, the technical solution of the present application can have good EIT imaging performance under different interference conditions, and can obtain robust and high-quality EIT imaging results when there is a serious error in the impedance boundary voltage due to the presence of high-intensity non-Gaussian interference.

[0131] It should be pointed out that EIT is a medical imaging method that attaches conductive surface electrodes to the skin around body parts to obtain information, and combines current stimulation and surface voltage to display the internal impedance distribution, that is, the tomographic image of the body part. Its non-invasive and radiation-free characteristics make it a real-time imaging tool that can be used at the patient's bedside. For example, EIT can be used to monitor lung ventilation, track the progression of brain damage, and help early detection of breast cancer. In these applications, chest EIT can optimize tidal volume, evaluate lung recruitment capacity, quantify the impact of lung recruitment, analyze the effect of suction or rehabilitation, and evaluate local lung function. In addition, chest EIT is gradually being accepted by pulmonary and critical care medicine for its ability to provide regional information on lung ventilation and perfusion. In addition to the aforementioned chest mid-lung ventilation monitoring (respiratory monitoring), the specific application scenarios of the EIT imaging scheme of the present application can also be specifically applied to cardiac function, gastrointestinal tract monitoring, etc., and the EIT imaging scheme of the present application is aimed at non-Gaussian interference caused by body motion interference or poor electrode contact in the EIT clinical use environment, while ensuring efficiency and theoretical convergence, it can obtain robust and high-quality EIT imaging, thereby making EIT technology more suitable for clinical applications. That is, the EIT imaging scheme of the present application classifies external interference and uses different norms to constrain it according to the characteristics of different interferences, so it has stronger resistance to interference, and the reconstructed image has better robustness, which is more suitable for dealing with severe interference caused by poor electrode contact, patient movement, medical intervention or nursing procedures in clinical practice.

[0132] For example, through relevant human test experiments, the EIT imaging scheme (REIT algorithm) of this application is compared with the non-robust method EIT L2 and robust approach EIT L1 Compared with the reconstructed EIT image, it can be seen that the EIT image reconstructed by the technical solution of the present invention is better than that reconstructed by EIT L2 、EIT L1 The EIT image reconstructed by the algorithm is more robust. The specific test experiment process is as follows:

[0133] The subject 1 (healthy male) wore the corresponding EIT electrode belt between the fourth and fifth ribs. The subject stood and kept breathing steadily. After ensuring that all electrodes were in good contact with the skin, the chest EIT boundary voltage was collected for 1 min. # Insert a piece of paper between the electrode 1 and the skin to separate the electrodes and generate strong interference. Collect the chest EIT boundary voltage for 1 minute. # (Electrode No. 8) performs the same steps, and then based on the collected data, the EIT imaging scheme of this application, EIT L2 Algorithms, EIT L1The algorithm reconstructs the corresponding EIT image, see Figure 3 As shown in FIG. 1 , the EIT image reconstructed by the REIT algorithm proposed in this application is almost the same as the reference EIT image, while the EIT L2 Algorithms and EIT L1 The algorithm cannot reconstruct the correct EIT image, and then analyzes the EIT image quality indicators under different imaging schemes. That is, the EIT image evaluation index values ​​reconstructed under different EIT imaging algorithms for the experimental object 1 are shown in Table 1:

[0134] Table 1

[0135]

[0136] As can be seen from Table 1, the image error (Error), correlation coefficient (Corr), ventilation center error (CoV) of the EIT images reconstructed under different EIT imaging algorithms and the normal EIT images without interference for the experimental subject 1 are err ), global inhomogeneity error (GI err ), ventral-dorsal ratio error (VtoD err ), left-right lung ratio error (RtoL err ) six evaluation index values. Among them, the closer the correlation coefficient is to 1, the better the image quality is. The smaller the corresponding values ​​of image error, ventilation center error, global inhomogeneity error, ventral-dorsal ratio error, and left-right lung ratio error are, the better the image quality is. It can be seen that the REIT algorithm proposed in this application has an absolute advantage in image quality indicators, that is, the REIT algorithm proposed in this application has a stronger ability to resist interference, and the reconstructed image has better robustness. Moreover, when electrode 1 # and 8 # When affected, the REIT algorithm proposed in this application can still reconstruct a robust and high-quality EIT image. L2 Algorithms and EIT L1 The EIT images reconstructed by the algorithm are superior.

[0137] Subject 2 (a patient with a left lung nodule) used the corresponding EIT system to collect the EIT boundary voltage on the right side of his chest before undergoing laparoscopic lung parenchymal resection. During the data collection process, the patient maintained spontaneous breathing and was not restricted from deliberate movement, medical treatment, and nursing. Therefore, the EIT boundary voltage will inevitably be destroyed. The values ​​of the six evaluation indicators of the EIT images reconstructed under different EIT imaging algorithms for Subject 2 are shown in Table 2:

[0138] Table 2

[0139]

[0140] As can be seen from Table 2 above, for experimental subject 2, the REIT algorithm proposed in this application is better than EIT in all image quality indicators and clinical indicators. L1 Algorithms, EIT L2 The algorithm is more advantageous. Also, see Figure 4 As shown in the figure, in the EIT images reconstructed by different algorithms for the experimental subject 2, EIT L2 The algorithm had the worst imaging performance and was unable to detect both lungs. L1 In the EIT image constructed by the algorithm, only the right lung can be identified, but the shape of the right lung has changed, which is different from the EIT image. L2 Algorithms and EIT L1 Compared with the previous algorithms, the EIT images obtained by the REIT algorithm have higher quality and can better reflect the actual condition of the patients.

[0141] Based on the test results of Experimental Subject 1 and Experimental Subject 2, it can be seen that the EIT image reconstructed by the REIT algorithm proposed in this application is better than the EIT L2 Algorithms, EIT L1 The algorithm is more robust.

[0142] In one embodiment, if Figure 5 As shown, based on the above-mentioned electrical impedance tomography method based on sparse recovery, the present invention also provides an electrical impedance tomography device based on sparse recovery, including:

[0143] The boundary voltage acquisition module 11 is used to obtain the boundary measurement voltage of the test area of ​​the test object;

[0144] A voltage difference calculation module 12, configured to calculate a corresponding boundary voltage difference according to the boundary measurement voltage; wherein the boundary voltage difference includes a noise-free boundary voltage difference, Gaussian noise, and a sparse abnormal interference to be solved;

[0145] A relationship building module 13 is used to build a relationship between the boundary voltage difference and the conductivity change to be solved, and obtain a corresponding target relationship;

[0146] A norm term construction module 14, configured to construct an L0 norm term and an L2 norm term based on the target relationship and using an L0 norm and an L2 norm;

[0147] An optimization problem construction module 15, used to construct a corresponding optimization problem based on the L0 norm term and the L2 norm term;

[0148] A problem solving module 16, configured to solve the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem to determine a target conductivity change;

[0149] The electrical impedance imaging module 17 is used to perform electrical impedance imaging based on the target conductivity change to obtain corresponding imaging results.

[0150] Figure 6 A schematic diagram of the structure of a terminal provided in an embodiment of the present application. The terminal may include:

[0151] A memory 501 , a processor 502 , and a computer program stored in the memory 501 and executable on the processor 502 .

[0152] When the processor 502 executes the program, the electrical impedance tomography method based on sparse recovery provided in the above embodiment is implemented.

[0153] Furthermore, the terminal further includes:

[0154] The communication interface 503 is used for communication between the memory 501 and the processor 502 .

[0155] The memory 501 is used to store computer programs that can be executed on the processor 502 .

[0156] The memory 501 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0157] If the memory 501, the processor 502 and the communication interface 503 are implemented independently, the communication interface 503, the memory 501 and the processor 502 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0158] Optionally, in a specific implementation, if the memory 501, the processor 502 and the communication interface 503 are integrated on a chip, the memory 501, the processor 502 and the communication interface 503 can communicate with each other through an internal interface.

[0159] The processor 502 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0160] This embodiment also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the electrical impedance tomography method based on sparse recovery as described above is implemented.

[0161] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed in this application. The specification and examples are to be considered exemplary only, and the true scope and spirit of the present invention are indicated by the claims.

[0162] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0163] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, which can be embodied in any computer-readable medium for use by an instruction execution system, apparatus or device (such as a computer-based system, a system including a processor or other system that can read instructions from an instruction execution system, apparatus or device and execute instructions), or used in combination with these instruction execution systems, apparatuses or devices.

[0164] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA, Programmable Gate Array), a field programmable gate array (FPGA, Field-Programmable Gate Array), etc.

[0165] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. An electrical impedance tomography method based on sparse recovery, characterized in that: The method comprises: Obtaining a boundary measurement voltage of a test area of ​​the test object, and calculating a corresponding boundary voltage difference according to the boundary measurement voltage; wherein the boundary voltage difference includes a noise-free boundary voltage difference, Gaussian noise, and sparse abnormal interference to be solved; Constructing a relationship between the boundary voltage difference and the conductivity change to be solved, and obtaining a corresponding target relationship; Based on the target relationship, an L0 norm term and an L2 norm term are constructed by using the L0 norm and the L2 norm; Constructing a corresponding optimization problem based on the L0 norm term and the L2 norm term, and solving the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem to determine a target conductivity change; Electrical impedance imaging is performed based on the target conductivity change to obtain corresponding imaging results.

2. The electrical impedance tomography method based on sparse recovery according to claim 1, characterized in that: The boundary voltage difference is: ; Furthermore, the target relationship between the boundary voltage difference and the conductivity change to be solved is: ; in, Indicates time point Moments and time points The boundary voltage difference at time Indicates time point The boundary measurement voltage at the moment, Indicates time point The boundary measurement voltage at the moment, represents the noise-free boundary voltage difference, represents Gaussian noise, represents the sparse abnormal interference to be solved, represents the conductivity change to be solved, represents the Jacobian matrix.

3. The electrical impedance tomography method based on sparse recovery according to claim 2, characterized in that: Based on the target relationship, L0 norm term and L2 norm term are constructed by using L0 norm and L2 norm, including: Determining a target term representing the Gaussian noise based on the target relationship; Based on the target term, a first L2 norm term is constructed using the L2 norm; Based on the conductivity change to be solved, a second L2 norm term is constructed using the L2 norm; Using the L0 norm to constrain the sparse abnormal interference to be solved to construct an L0 norm term based on the constrained sparse abnormal interference; Wherein, constructing a corresponding optimization problem based on the L0 norm term and the L2 norm term includes: Constructing a corresponding optimization problem based on the L0 norm, the first L2 norm term, and the second L2 norm term; The optimization problem is: ； ； ； in, represents the first L2 norm term, represents the second L2 norm term, represents the L0 norm term, represents the first penalty parameter and greater than zero, represents the second penalty parameter and greater than zero, represents the inverse matrix of the covariance of Gaussian noise, express The matrix decomposition factor of express The transposed matrix of represents the inverse matrix of the covariance of the desired conductivity, express The matrix decomposition factor of express The transposed matrix of .

4. The electrical impedance tomography method based on sparse recovery according to claim 3, characterized in that: Solving the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem to determine a target conductivity change includes: Based on a preset nearest neighbor alternating minimization method, the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem are alternately solved to determine a target conductivity change.

5. The electrical impedance tomography method based on sparse recovery according to claim 4, characterized in that: The method of alternately solving the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem based on a preset nearest neighbor alternating minimization method to determine a target conductivity change includes: Based on the conductivity change to be solved and the sparse abnormal interference to be solved, and using the L2 norm, a first nearest neighbor constraint term and a second nearest neighbor constraint term are constructed; Based on the optimization problem, a first linear least squares problem and a second linear least squares problem are constructed by using the first nearest neighbor constraint item and the second nearest neighbor constraint item; Alternately solving the conductivity change to be solved in the first linear least squares problem and the sparse abnormal interference to be solved in the second linear least squares problem to determine a target conductivity change; Among them, the first nearest neighbor constraint item is: ; The second nearest neighbor constraint is: ; The first linear least squares problem is: ； The second linear least squares problem is: ； in, Indicates The conductivity change matrix of the iterative optimization is: Indicates The conductivity change matrix of the iterative optimization is: Indicates Sparse anomaly interference of iterative optimization, Indicates Sparse anomaly interference of iterative optimization, represents the neighbor parameter and Greater than zero.

6. The electrical impedance tomography method based on sparse recovery according to claim 5, characterized in that: Solving the conductivity change to be solved in the first linear least squares problem includes: Constructing a corresponding closed-form solution based on the first linear least squares problem, and solving the conductivity change to be solved based on the closed-form solution; Among them, the closed-form solution is: ； and, represents the transposed matrix of the Jacobian matrix, Represents the identity matrix.

7. The electrical impedance tomography method based on sparse recovery according to claim 6, characterized in that: Solving the sparse abnormal interference to be solved in the second linear least squares problem includes: scalarizing the second linear least squares problem to obtain a corresponding scalar formula; Based on the scalar formula, and using the preset The rule solves the sparse abnormal interference to be solved; Wherein, the scalar formula is: ； ； ； in, express No. item, Indicates sparse abnormal interference No. item, Indicates Sparse anomaly interference for sub-iterative optimization No. item, Indicates Sparse anomaly interference for sub-iterative optimization No. item, express The Line Column elements, express The square root of .

8. An electrical impedance tomography device based on sparse recovery, characterized in that: The device comprises: A boundary voltage acquisition module is used to obtain the boundary measurement voltage of the test area of ​​the test object; A voltage difference calculation module, used to calculate a corresponding boundary voltage difference according to the boundary measurement voltage; wherein the boundary voltage difference includes a noise-free boundary voltage difference, Gaussian noise and a sparse abnormal interference to be solved; A relationship building module, used to build a relationship between the boundary voltage difference and the conductivity change to be solved, and obtain a corresponding target relationship; A norm term construction module, used to construct an L0 norm term and an L2 norm term based on the target relationship and using an L0 norm and an L2 norm; An optimization problem construction module, used for constructing a corresponding optimization problem based on the L0 norm term and the L2 norm term; A problem solving module, used for solving the conductivity change to be solved and the sparse abnormal interference to be solved in the optimization problem to determine a target conductivity change; The electrical impedance imaging module is used to perform electrical impedance imaging based on the target conductivity change to obtain corresponding imaging results.

9. A terminal, characterized in that: include: A memory, a processor, and an electrical impedance tomography program based on sparse recovery stored in the memory and executable on the processor, wherein the electrical impedance tomography program based on sparse recovery, when executed by the processor, implements the steps of the electrical impedance tomography method based on sparse recovery as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which can be executed to implement the steps of the electrical impedance tomography method based on sparse recovery according to any one of claims 1 to 7.

Images