Cerebral hemorrhage magnetic induction tomography tensor compressed sensing acquisition method and system

Abstract

The invention discloses a cerebral hemorrhage magnetic induction tomography tensor compressed sensing acquisition method and system, which is applied to the technical field of magnetic induction tomography, and comprises the following steps: arranging a coil around the human brain, acquiring an induction electric signal generated on the coil, and converting the induction electric signal into a three-dimensional signal; carrying out sparse processing on the three-dimensional signal, constructing and obtaining a three-dimensional tensor under recovered sparse representation by a sparse representation method, and carrying out sparse sampling on the three-dimensional tensor to obtain measurement data; and based on measurement data, performing reconstruction processing on the three-dimensional tensor under the recovery sparse representation by adopting an inverse discrete cosine transform method, and performing construction to obtain a reconstructed three-dimensional tensor signal. In this way, the signals on the sensor array are expressed as the tensor, the tensor compressed sensing operation is carried out, the data collection amount can be effectively reduced, the data collection time is shortened, and an effective method is provided for miniaturization and real-time performance of a cerebral hemorrhage MIT system.

Description

Technical Field

[0001] The present disclosure relates to the technical field of magnetic induction tomography, and in particular to a tensor compression sensing acquisition method and system for magnetic induction tomography of cerebral hemorrhage. Background Art

[0002] As a sudden brain disease, cerebral hemorrhage develops quickly and has a short golden treatment time. Rapidly locating the bleeding location and bleeding range and other key information can buy precious treatment time for patients and improve their survival rate. Currently, commonly used medical imaging equipment such as CT, MRI and other technical equipment are bulky and slow in imaging speed, and are not suitable for emergency treatment. Magnetic Induction Tomography (MIT) technology is an emerging non-destructive testing technology that uses the principle of electromagnetic induction to detect changes in the conductivity of biological tissues. It has the characteristics of no radiation to the human body, no contact, low testing cost, small size, and fast imaging speed, which enables it to effectively make up for the shortcomings of existing medical imaging technology and is suitable for imaging examinations of sudden diseases such as cerebral hemorrhage.

[0003] At present, the MIT data acquisition for cerebral hemorrhage mainly adopts the method of multi-channel separate acquisition. However, this method has some significant disadvantages. First, due to the need to collect data from multiple channels at the same time, the amount of data collected is large, which brings huge challenges to data storage and processing. This not only takes up a large storage space, but also increases the complexity of data storage and processing. Secondly, the data acquisition speed of the multi-channel acquisition system is relatively slow, and may not meet the requirements for rapid positioning of cerebral hemorrhage. In addition, in order to support multi-channel acquisition, the circuit design needs to be more complex, which increases the technical difficulty and cost of the system. Finally, multi-channel acquisition systems usually require a larger device volume, which may not be ideal for some scenarios with high requirements on device volume, such as mobile medical or portable monitoring.

[0004] Therefore, there is an urgent need for a method and a corresponding system that can effectively solve the problems of high circuit workload and poor real-time performance caused by the MIT data acquisition process. Summary of the invention

[0005] The present invention provides a method and system for collecting compressed sensing tensor of magnetic induction tomography of cerebral hemorrhage. By adopting the OMP algorithm to process the three-dimensional signal tensor, the present invention at least solves the technical problems of high workload of system hardware circuit, difficult data transmission and poor real-time performance of MIT system caused by the large amount of data collected in MIT of cerebral hemorrhage in the prior art.

[0006] According to a first aspect of the present disclosure, a method for collecting magnetic induction tomography tensor compressed sensing for cerebral hemorrhage is provided, comprising the following steps:

[0007] Coils are placed around the human brain to collect the induced electrical signals generated by the coils and convert them into three-dimensional signals;

[0008] Sparse processing is performed on the three-dimensional signal, a three-dimensional tensor under the sparse representation is constructed by a sparse representation method, and the three-dimensional tensor is sparsely sampled to obtain measurement data;

[0009] Based on the measurement data, an inverse discrete cosine transform method is used to reconstruct the three-dimensional tensor under the restored sparse representation to construct a reconstructed three-dimensional tensor signal.

[0010] According to the above aspects and any possible implementation, there is further provided an implementation, wherein the coil includes an excitation coil and a detection coil;

[0011] A sine wave voltage is input into the excitation coil for signal excitation, thereby generating an induced electrical signal on the detection coil.

[0012] According to the above aspects and any possible implementation, an implementation is further provided, wherein the three-dimensional signal tensor is specifically: X∈R K×N×M , where X is the three-dimensional signal tensor, K is the extra dimension used for stacking three-dimensional tensors, and N and M are the signal length and the number of sensors, respectively.

[0013] According to the above aspects and any possible implementation, an implementation is further provided, wherein the process of sparsely processing the three-dimensional signal, constructing a three-dimensional tensor under the sparse representation method to restore the sparse representation, and sparsely sampling the three-dimensional tensor to obtain the measurement data is:

[0014] Perform discrete cosine transform on the three-dimensional signal tensor to obtain a sparse representation after discrete cosine transform;

[0015] The sampling rate is set and the measurement matrix is ​​selected. The three-dimensional tensor after sparse representation is sparsely sampled by the sampling rate and the measurement matrix to obtain the measurement data Y=AX, where Y is the measurement data and A is the measurement matrix.

[0016] According to the above aspects and any possible implementation, an implementation is further provided, wherein the method of performing discrete cosine transform on the three-dimensional signal tensor to obtain a sparse representation after discrete cosine transform is:

[0017]

[0018] Among them, χ(i, j, k) is the spatial domain representation of the three-dimensional tensor, Y(u, v, w) is the frequency domain representation of the three-dimensional tensor after discrete cosine transform, 0 ≤ i < t, 0 ≤ j < c, 0 ≤ k < x, 0 ≤ u < t, 0 ≤ v < c, 0 ≤ w < x, α(u), α(v), α(w) are orthogonal basis coefficients, t is the number of coil channels, c is the number of signals in each channel, and x is the number of signal points in each signal.

[0019] For the aspects and any possible implementation manners described above, a further implementation manner is provided. The process of reconstructing and processing the three-dimensional tensor under the restored sparse representation by using the inverse discrete cosine transform method to construct the reconstructed three-dimensional tensor signal is as follows:

[0020] Initialize the residual f = Y based on the measurement data;

[0021] Iteratively calculate the projection coefficients according to the residual and the measurement matrix, and select the atom index with the largest projection coefficient;

[0022] Calculate and update the reconstructed signal based on the atom index, and update the residual again based on the reconstructed signal;

[0023] Iterate on the residual, and stop the iteration and output the final reconstruction result when the condition is met.

[0024] For the aspects and any possible implementation manners described above, a further implementation manner is provided. The process of calculating and updating the reconstructed signal based on the atom index and updating the residual again based on the reconstructed signal is as follows:

[0025] Select an atom subset based on the atom index, and calculate the pseudo-inverse of the matrix composed of the column vectors of the atom subset;

[0026] Update the reconstructed signal according to the pseudo-inverse Among them, is the pseudo-inverse of the matrix composed of the column vectors of the selected atom subset, and there is

[0027] Calculate the product of the reconstructed signal and the measurement matrix, and update and iterate the residual based on the measurement data.

[0028] For the aspects and any possible implementation manners described above, a further implementation manner is provided. The method further includes: calculating the relative error and root mean square error between the reconstructed three-dimensional tensor signal and the original simulation signal to analyze the accuracy of the reconstructed three-dimensional tensor signal, specifically:

[0029]

[0030] Among them, RE is the relative error, and RMSE is the root mean square error. is the prediction result, y i is the actual result, and N is the amount of data.

[0031] According to a second aspect of the present disclosure, a magnetic induction tomography tensor compressed sensing acquisition system for cerebral hemorrhage is provided, which implements the magnetic induction tomography tensor compressed sensing acquisition method for cerebral hemorrhage as described in the first aspect, including: a three-dimensional signal acquisition module, a three-dimensional signal sparse processing module and a three-dimensional signal reconstruction module;

[0032] The three-dimensional signal acquisition module is used to arrange coils around the human brain, collect the induced electrical signals generated by the coils, and convert the induced electrical signals into three-dimensional signals;

[0033] The three-dimensional signal sparse processing module is used to perform sparse processing on the three-dimensional signal, construct a three-dimensional tensor under the sparse representation method to restore the sparse representation, and perform sparse sampling on the three-dimensional tensor to obtain measurement data;

[0034] The three-dimensional signal reconstruction module is used to reconstruct the three-dimensional tensor under the restored sparse representation by using an inverse discrete cosine transform method to construct a reconstructed three-dimensional tensor signal.

[0035] Compared with the prior art, the present invention has the following technical effects:

[0036] The present invention can effectively reduce the amount of data collection and the time of data collection by representing the signals on the sensor array as tensors and performing tensor compressed sensing operations. At the same time, it can also ensure the accuracy of the restored signals and the running time is short, thereby providing an effective method for the miniaturization and real-time performance of the MIT system for cerebral hemorrhage.

[0037] It should be understood that the contents described in the summary of the invention are not intended to limit the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, among which:

[0039] Figure 1 A schematic diagram of a process for collecting magnetic induction tomography tensor compressed sensing data for cerebral hemorrhage according to an embodiment of the present disclosure is shown;

[0040] Figure 2 A schematic diagram of the structure of a magnetic induction tomography tensor compressed sensing acquisition system for cerebral hemorrhage according to an embodiment of the present disclosure is shown;

[0041] Figure 3 A schematic diagram of coil array arrangement is shown in Example 1 of a tensor compression sensing acquisition method for magnetic induction tomography of cerebral hemorrhage according to an embodiment of the present disclosure;

[0042] Figure 4 A schematic diagram of induced electrical signals acquired according to Embodiment 1 of a tensor compressed sensing acquisition method for magnetic induction tomography of cerebral hemorrhage according to an embodiment of the present disclosure is shown;

[0043] Figure 5 A schematic diagram of a three-dimensional electrical signal reconstructed according to Embodiment 1 of a tensor compressed sensing acquisition method for magnetic induction tomography of cerebral hemorrhage according to an embodiment of the present disclosure is shown;

[0044] Figure 6 A schematic diagram of coil array arrangement according to Embodiment 2 of a tensor compressed sensing acquisition method for magnetic induction tomography of cerebral hemorrhage according to an embodiment of the present disclosure is shown;

[0045] Figure 7 A schematic diagram of a magnetic induction tomography experimental system is shown according to Embodiment 2 of a magnetic induction tomography tensor compressed sensing acquisition method for cerebral hemorrhage according to an embodiment of the present disclosure;

[0046] Figure 8 A schematic diagram of induced electrical signals acquired according to Embodiment 2 of a tensor compressed sensing acquisition method for magnetic induction tomography of cerebral hemorrhage according to an embodiment of the present disclosure is shown;

[0047] Fig. 9 A schematic diagram of a three-dimensional electrical signal reconstructed according to Example 2 of a tensor compressed sensing acquisition method for magnetic induction tomography of cerebral hemorrhage according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0049] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0050] Reference Figure 1 As shown, this embodiment provides a tensor compression sensing acquisition method for magnetic induction tomography of cerebral hemorrhage, including the following steps:

[0051] S101. Arrange coils around the human brain, collect the induced electrical signals generated on the coils, and convert the induced electrical signals into three-dimensional signals.

[0052] In this embodiment, the coil includes an excitation coil and a detection coil, and the detection coil is arranged in a ring shape. By applying a sinusoidal voltage to the excitation coil for signal excitation, an induced voltage signal is generated in the detection coil.

[0053] After collecting the signals, stack the collected induced electrical signals to generate a three-dimensional signal tensor. Specifically: the signals on the collected sensor array are the signals on each sensor in the sensor array, and they are all one-dimensional signals. The stacked three-dimensional signal tensor is X ∈ R K×N×M , where K is the additional dimension for three-dimensional tensor stacking, and N and M are the signal length and the number of sensors respectively.

[0054] S102. Perform sparse processing on the three-dimensional signal, construct a three-dimensional tensor under the recovered sparse representation using the sparse representation method, and perform sparse sampling on the three-dimensional tensor to obtain measurement data.

[0055] In this embodiment, the process of performing sparse processing on the three-dimensional signal is: perform a discrete cosine transform on the tensor of the three-dimensional signal to obtain the sparse representation of the three-dimensional signal tensor. The specific transformation method is:

[0056]

[0057]

[0058] where χ(i,j,k) is the spatial domain representation of the three-dimensional tensor, Y(u,v,w) is the frequency domain representation of the three-dimensional tensor after discrete cosine transform, 0 ≤ i < t, 0 ≤ j < c, 0 ≤ k < x, 0 ≤ u < t, 0 ≤ v < c, 0 ≤ w < x. α(u), α(v), α(w) are orthogonal basis coefficients, t is the number of coil channels, c is the number of signals in each channel, and x is the number of signal points in each signal.

[0059] After the sparse representation is completed, perform sparse sampling on the sparse-represented three-dimensional tensor to obtain measurement data Y = AX, where the sampling rate is r and the measurement matrix is A ∈ R rN×N The specific method for obtaining the measurement data is Y k = AX k×N×m , k = 1, 2, …, K, m = 1, 2, …, M.

[0060] S103. Based on the measurement data, use the inverse discrete cosine transform method to perform reconstruction processing on the three-dimensional tensor under the recovered sparse representation, and construct a reconstructed three-dimensional tensor signal.

[0061] This embodiment adopts the orthogonal matching pursuit method (Orthogonal Matching Pursuit, OMP) to reconstruct the three-dimensional tensor signal. The OMP algorithm is based on sparse signals, that is, it is believed that the signal can be accurately or approximately represented by a linear combination of a finite number of atoms of an overcomplete dictionary. The core idea is to gradually construct a sparse representation of the signal by greedily selecting the dictionary atoms that best match the current residual signal.

[0062] The convergence and performance guarantee of the OMP method are mainly reflected in:

[0063] Given an overcomplete dictionary D, if the sparsity K of the signal x under the dictionary is less than a certain threshold and the noise meets certain conditions, then OMP can accurately restore x within a fixed number of iterations K; for signals that are not completely sparse but still have sparse structures, OMP can approximate their true representation with a certain error bound. The error bound is related to the actual sparsity of the signal, the RIP constant of the dictionary, and the noise level.

[0064] The specific process is:

[0065] Initialize the residual f=Y based on the measurement data;

[0066] The projection coefficient α is obtained by iterative calculation based on the residual and measurement matrix k×rN×m =A T ·f k×rN×m , and select the atom with the largest projection coefficient i = atgmax i |α i |;

[0067] Select an atom subset based on the atom index, and calculate the pseudo-inverse of the matrix composed of the column vectors of the atom subset;

[0068] Reconstruct the signal based on pseudo-inverse updating in, is the pseudo-inverse of the column vector matrix of the selected atomic subset,

[0069] Calculate the product of the reconstructed signal and the measurement matrix, and iterate the residual based on the measurement data to obtain

[0070] When the number of iterations meets the conditions, the iteration is stopped and the final reconstruction result is output

[0071] like Figure 2 As shown, this embodiment also provides a magnetic induction tomography tensor compression sensing acquisition system for cerebral hemorrhage, including: a three-dimensional signal acquisition module 1, a three-dimensional signal sparse processing module 2 and a three-dimensional signal reconstruction module 3;

[0072] The three-dimensional signal acquisition module 1 is used to arrange coils around the human brain, collect the induced electrical signals generated by the coils, and convert the induced electrical signals into three-dimensional signals;

[0073] The three-dimensional signal sparse processing module 2 is used to perform sparse processing on the three-dimensional signal, construct a three-dimensional tensor under the sparse representation method to restore the sparse representation, and perform sparse sampling on the three-dimensional tensor to obtain measurement data;

[0074] The three-dimensional signal reconstruction module 3 is used to reconstruct the three-dimensional tensor under the restored sparse representation based on the measurement data by using the inverse discrete cosine transform method to construct a reconstructed three-dimensional tensor signal.

[0075] Example 1

[0076] In order to realize the acquisition of magnetic induction tomography detection signals for cerebral hemorrhage, this embodiment constructs the following Figure 3 In the simulation detection model shown in the figure, coil 1 is used as the excitation coil, coils 2-8 are used as detection coils, and the seven detection coils are arranged in a ring shape. A sine wave voltage of 0.2V and 1kHz is passed through coil 1. The induced voltage signals generated on coils 2-8 are collected using the data collection method proposed in the present invention. A total of 10s of data are collected, including 1000 data values. Specifically, Figure 4 shown.

[0077] First, the normally collected signals are stacked into a three-dimensional tensor, whose size is X∈R 1×1000×7 , after performing discrete cosine transform to obtain sparse representation, the sampling rate r is set to 0.5, and the measurement matrix is ​​A∈R 500×1000 , sparsely sample it to get Y, the size is Y∈R 1×500×7 , and then use it as the input of the OMP algorithm, initialize the residual to f = Y, and finally get the recovery tensor

[0078] The recovery detection signal obtained by expanding it is as follows Figure 5 The relative error and root mean square error between the restored signal obtained by the present invention and the original simulation signal are summarized in Table 1:

[0079] Table 1 Relative error and RMS error between restored signal and original signal

[0080]

[0081] In addition, the running time of the present invention is 2.15s, which can effectively shorten the data collection time compared with the prior art.

[0082] Example 2

[0083] In order to further verify the effectiveness of the present invention in the actual data collection process, this embodiment uses Figure 6 , Figure 7 The data acquisition system shown in the figure performs actual data acquisition. A sine wave voltage signal of 5V and 1KHz is passed into coil 1. Coil 2-8 is used as a detection coil to collect the detection voltage signal induced by it. The induced voltage signal generated on coil 2-8 is collected using the data acquisition method proposed in the present invention. A total of 10s of data are collected, and the original sampling rate is 10000 / s. Figure 8 shown.

[0084] First, the normally collected signals are stacked into a three-dimensional tensor, whose size is X∈R 1×10000×7 , after performing discrete cosine transform to obtain sparse representation, the sampling rate r is set to 0.5, and the measurement matrix is ​​A∈R 5000×10000 , sparsely sample it to get Y, the size is Y∈R 1×5000×7 , and then use it as the input of the OMP algorithm, initialize the residual to f = Y, and finally get the recovery tensor

[0085] The recovery detection signal obtained by expanding it is as follows Fig. 9 The relative error and root mean square error between the restored signal obtained by the present invention and the original simulation signal are summarized in Table 2:

[0086] Table 2 Relative error and RMS error between restored signal and original signal

[0087]

[0088] In addition, the running time of the present invention is 5.2s, which can effectively shorten the data collection time compared with the prior art.

[0089] It can be seen that, different from the prior art, the present invention provides a tensor compressed sensing acquisition method for cerebral hemorrhage magnetic induction tomography based on the OMP algorithm, and introduces tensor compressed sensing technology into the data acquisition of cerebral hemorrhage magnetic induction tomography. It requires a small amount of collected data, has a short running time, and has high recovery accuracy, which provides a guarantee for the real-time performance of cerebral hemorrhage magnetic induction tomography technology and the miniaturization of the system.

[0090] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present disclosure is not limited by the order of the actions described, because according to the present disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present disclosure.

[0091] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0092] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A tensor compression sensing acquisition method for magnetic induction tomography of cerebral hemorrhage, characterized in that: It includes the following steps: Arrange coils around the human brain, collect the induced electrical signals generated on the coils, and convert the induced electrical signals into three-dimensional signals; Perform sparse processing on the three-dimensional signals, construct a three-dimensional tensor under the restored sparse representation under the method of sparse representation, and perform sparse sampling on the three-dimensional tensor under the restored sparse representation to obtain measurement data; Based on the measurement data, use the inverse discrete cosine transform method to perform reconstruction processing on the three-dimensional tensor under the restored sparse representation, and construct a reconstructed three-dimensional tensor signal.

2. The method for collecting magnetic induction tomography tensor compressed sensing data for cerebral hemorrhage according to claim 1, characterized in that: The coils include excitation coils and detection coils; Signal excitation is performed by inputting a sine wave voltage to the excitation coils, so as to generate induced electrical signals on the detection coils.

3. The method for collecting magnetic induction tomography tensor compressed sensing data for cerebral hemorrhage according to claim 1, characterized in that: The three-dimensional signal tensor is specifically: X∈R K×N×M , where X is the three-dimensional signal tensor, K is the extra dimension used for stacking three-dimensional tensors, and N and M are the signal length and the number of sensors, respectively.

4. The method for collecting magnetic induction tomography tensor compressed sensing data for cerebral hemorrhage according to claim 3, characterized in that: The process of performing sparse processing on the three-dimensional signals, constructing a three-dimensional tensor under the restored sparse representation under the method of sparse representation, and performing sparse sampling on the three-dimensional tensor under the restored sparse representation to obtain measurement data is as follows: Perform discrete cosine transform on the three-dimensional signal tensor to obtain the sparse representation after discrete cosine transform; Set the sampling rate and select a measurement matrix, and perform sparse sampling on the three-dimensional tensor after sparse representation through the sampling rate and the measurement matrix to obtain measurement data Y = AX, where Y is the measurement data and A is the measurement matrix.

5. The method for collecting magnetic induction tomography tensor compressed sensing data for cerebral hemorrhage according to claim 4, characterized in that: The method of performing discrete cosine transform on the three-dimensional signal tensor to obtain the sparse representation after discrete cosine transform is as follows: Among them, χ(i, j, k) is the spatial domain representation of the three-dimensional tensor, Y(u, v, w) is the frequency domain representation of the three-dimensional tensor after discrete cosine transform, 0 ≤ i < t, 0 ≤ j < c, 0 ≤ k < x, 0 ≤ u < t, 0 ≤ v < c, 0 ≤ w < x, α(u), α(v), α(w) are orthogonal basis coefficients, t is the number of coil channels, c is the number of signals in each channel, and x is the number of signal points in each signal.

6. The method for collecting magnetic induction tomography tensor compressed sensing data for cerebral hemorrhage according to claim 5, characterized in that: The process of performing reconstruction processing on the three-dimensional tensor under the restored sparse representation based on the measurement data by using the inverse discrete cosine transform method to construct a reconstructed three-dimensional tensor signal is as follows: Initialize the residual f = Y based on the measurement data; Iteratively calculate the projection coefficients according to the residual and the measurement matrix, and select the atom index with the largest projection coefficient; Calculate and update the reconstructed signal based on the atom index, and update the residual again based on the reconstructed signal; By iterating on the residual, stop the iteration and output the final reconstruction result when the condition is met.

7. The method for collecting magnetic induction tomography tensor compressed sensing data for cerebral hemorrhage according to claim 6, characterized in that: The process of calculating and updating the reconstructed signal based on the atom index and updating the residual again based on the reconstructed signal is as follows: Select an atom subset based on the atom index, and calculate the pseudo-inverse of the matrix composed of the column vectors of the atom subset; Reconstruct the signal based on pseudo-inverse updating in, is the pseudo-inverse of the matrix consisting of the column vectors of the selected atomic subset, Calculate the product of the reconstructed signal and the measurement matrix, and update and iterate the residual based on the measurement data.

8. The method for collecting magnetic induction tomography tensor compressed sensing data for cerebral hemorrhage according to claim 1, characterized in that: The method further includes: calculating the relative error and root mean square error between the reconstructed three-dimensional tensor signal and the original simulation signal to analyze the accuracy of the reconstructed three-dimensional tensor signal, specifically: Among them, RE is relative error, RMSE is root mean square error, To reconstruct the three-dimensional tensor signal, y i is a real three-dimensional signal, and N is the amount of data.

9. A magnetic induction tomography tensor compressed sensing acquisition system for cerebral hemorrhage, which implements the magnetic induction tomography tensor compressed sensing acquisition method for cerebral hemorrhage as claimed in any one of claims 1 to 8, characterized in that: It includes: A three-dimensional signal acquisition module (1), a three-dimensional signal sparse processing module (2), and a three-dimensional signal reconstruction module (3); The three-dimensional signal acquisition module (1) is used to arrange coils around the human brain, collect induced electrical signals generated by the coils, and convert the induced electrical signals into three-dimensional signals; The three-dimensional signal sparse processing module (2) is used to perform sparse processing on the three-dimensional signal, construct a three-dimensional tensor under the restored sparse representation method, and perform sparse sampling on the three-dimensional tensor under the restored sparse representation to obtain measurement data; The three-dimensional signal reconstruction module (3) is used to reconstruct the three-dimensional tensor under the restored sparse representation based on the measurement data using an inverse discrete cosine transform method to construct a reconstructed three-dimensional tensor signal.

Images