CN116869504A - Data compensation method for cerebral ischemia conductivity distribution reconstruction - Google Patents
Data compensation method for cerebral ischemia conductivity distribution reconstruction Download PDFInfo
- Publication number
- CN116869504A CN116869504A CN202310731047.4A CN202310731047A CN116869504A CN 116869504 A CN116869504 A CN 116869504A CN 202310731047 A CN202310731047 A CN 202310731047A CN 116869504 A CN116869504 A CN 116869504A
- Authority
- CN
- China
- Prior art keywords
- dehydration
- layer
- voltage
- output
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 43
- 201000006474 Brain Ischemia Diseases 0.000 title claims abstract description 29
- 206010008120 Cerebral ischaemia Diseases 0.000 title claims abstract description 29
- 206010008118 cerebral infarction Diseases 0.000 title claims abstract description 29
- 230000018044 dehydration Effects 0.000 claims abstract description 58
- 238000006297 dehydration reaction Methods 0.000 claims abstract description 58
- 238000005259 measurement Methods 0.000 claims abstract description 54
- 210000004761 scalp Anatomy 0.000 claims abstract description 24
- 238000013528 artificial neural network Methods 0.000 claims abstract description 20
- 230000003044 adaptive effect Effects 0.000 claims abstract description 8
- 238000012545 processing Methods 0.000 claims abstract description 7
- 238000001914 filtration Methods 0.000 claims abstract description 6
- 238000002593 electrical impedance tomography Methods 0.000 claims description 20
- 210000002569 neuron Anatomy 0.000 claims description 18
- 238000012549 training Methods 0.000 claims description 18
- 206010048962 Brain oedema Diseases 0.000 claims description 14
- 208000006752 brain edema Diseases 0.000 claims description 13
- 210000005013 brain tissue Anatomy 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 13
- 230000005284 excitation Effects 0.000 claims description 11
- 210000001519 tissue Anatomy 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 6
- 210000003625 skull Anatomy 0.000 claims description 6
- 208000028867 ischemia Diseases 0.000 claims description 5
- 238000003325 tomography Methods 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000007917 intracranial administration Methods 0.000 abstract description 13
- 230000002490 cerebral effect Effects 0.000 abstract description 4
- 238000012360 testing method Methods 0.000 abstract description 3
- 208000005156 Dehydration Diseases 0.000 description 43
- 210000004556 brain Anatomy 0.000 description 6
- 238000003384 imaging method Methods 0.000 description 6
- 238000012544 monitoring process Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000002595 magnetic resonance imaging Methods 0.000 description 5
- 241000283973 Oryctolagus cuniculus Species 0.000 description 4
- 238000002610 neuroimaging Methods 0.000 description 4
- FBPFZTCFMRRESA-KVTDHHQDSA-N D-Mannitol Chemical compound OC[C@@H](O)[C@@H](O)[C@H](O)[C@H](O)CO FBPFZTCFMRRESA-KVTDHHQDSA-N 0.000 description 3
- 229930195725 Mannitol Natural products 0.000 description 3
- 238000002059 diagnostic imaging Methods 0.000 description 3
- 235000010355 mannitol Nutrition 0.000 description 3
- 239000000594 mannitol Substances 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 238000002591 computed tomography Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 210000003128 head Anatomy 0.000 description 2
- 241001075517 Abelmoschus Species 0.000 description 1
- 206010008111 Cerebral haemorrhage Diseases 0.000 description 1
- 241000965478 Darksidea epsilon Species 0.000 description 1
- 206010030113 Oedema Diseases 0.000 description 1
- 230000001154 acute effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000001175 cerebrospinal fluid Anatomy 0.000 description 1
- 238000003759 clinical diagnosis Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000004217 heart function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000002347 injection Methods 0.000 description 1
- 239000007924 injection Substances 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 231100000915 pathological change Toxicity 0.000 description 1
- 230000036285 pathological change Effects 0.000 description 1
- 230000007170 pathology Effects 0.000 description 1
- 230000002685 pulmonary effect Effects 0.000 description 1
- 230000002285 radioactive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000009423 ventilation Methods 0.000 description 1
- 238000007794 visualization technique Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves
- A61B5/053—Measuring electrical impedance or conductance of a portion of the body
- A61B5/0536—Impedance imaging, e.g. by tomography
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0033—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room
- A61B5/004—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room adapted for image acquisition of a particular organ or body part
- A61B5/0042—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room adapted for image acquisition of a particular organ or body part for the brain
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B2576/00—Medical imaging apparatus involving image processing or analysis
- A61B2576/02—Medical imaging apparatus involving image processing or analysis specially adapted for a particular organ or body part
- A61B2576/026—Medical imaging apparatus involving image processing or analysis specially adapted for a particular organ or body part for the brain
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Physics & Mathematics (AREA)
- Surgery (AREA)
- Veterinary Medicine (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Radiology & Medical Imaging (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Pathology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Neurology (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention discloses a data compensation method for reconstructing cerebral ischemia conductivity distribution, which comprises the steps of sending a cerebral full-field boundary measurement voltage value into a trained fully-connected neural network, determining the dehydration degree, finding a corresponding model boundary measurement voltage value caused by independent scalp dehydration according to the dehydration degree and a preceding test sequence, then compensating, combining coordinate information of pixels, reconstructing an intracranial cerebral ischemia image, and finally performing binarization processing on an image reconstruction result by using an adaptive threshold filtering algorithm.
Description
Technical Field
The invention belongs to the technical field of electrical tomography, and particularly relates to a data compensation method for reconstructing cerebral ischemia conductivity distribution.
Background
Medical imaging is a key part of clinical procedures, provides higher accuracy for monitoring human diseases, and provides powerful scientific basis for clinical diagnosis and biomedical research. A number of medical imaging techniques have been developed. Conventional imaging methods include computed tomography (Computer tomography, CT) and magnetic resonance imaging (Magnetic resonance imaging, MRI). However, CT is radioactive, and MRI is time consuming and expensive. In recent years, electrical impedance tomography (Electrical Impedance Tomography, EIT) has attracted considerable attention from many researchers due to its low cost and high temporal resolution. EIT is more suitable for bedside monitoring of patients than CT and MRI, and therefore, EIT has shown great potential in medical applications, such as in the fields of pulmonary ventilation, brain imaging, and cardiac function imaging.
Cerebral edema is a serious complication of acute cerebral ischemia. Cerebral edema caused by ischemia is estimated to account for 50% of the mortality rate of patients. Thus, accurate detection of cerebral ischemia is critical for patient rehabilitation. In order to reduce patient mortality and improve prognostic outcome, accurate imaging examinations are critical in clinical treatment. However, conventional imaging methods do not allow continuous monitoring of intracranial pathology. In order to show real-time efficacy during treatment, a bedside diagnostic method is urgently needed to monitor cerebral edema. As a visualization technique, EIT is favored because of its ability to monitor intracranial pathological changes in real time. With this technique, images representing changes in conductivity distribution can be reconstructed and physiological changes in the brain region can be provided, thereby providing an alternative to brain imaging, which also makes it a promising candidate for monitoring edema treatment. However, it was found that when brain tissue is dehydrated during the dehydration process for treating cerebral edema, the scalp is dehydrated, which will have a great influence on the boundary measurement. Therefore, the quality of the reconstruction of cerebral oedema is severely affected. Aiming at the technical problem, the invention provides a data compensation method for reconstructing cerebral ischemia conductivity distribution, which aims to reduce the influence of scalp dehydration on intracranial monitoring.
In the prior art, bin Yang et al, 2019 published in Abelmoschus nerve imaging journal (NeuroImage Clinical), volume 23, page 101909, entitled "Electrical impedance tomography and intracranial pressure comparison during dehydration treatment of cerebral edema (Comparison of electrical impedance tomography and intracranial pressure during dehydration treatment of cerebral edema), which shows that EIT can monitor changes in cerebral water content to reveal the severity of cerebral edema, can provide real-time and non-invasive imaging tools for early identification of cerebral edema and evaluation of mannitol dehydration, and although mannitol dehydration is effective in treating cerebral edema, both brain tissue layers and scalp layers are dehydrated during mannitol dehydration treatment of cerebral edema, scalp dehydration will cause an increase in impedance of the scalp layers, further impeding current injection into the brain layers, which will have a certain effect on accurate monitoring of intracranial dehydration. However, the above-described method is mainly directed to the study of the relationship between intracranial pressure and EIT image at the time of dehydration, and relatively few studies are conducted on the accurate reconstruction of cerebral ischemia conductivity distribution by suppressing the scalp dehydration effect using direct processing of voltage data.
In order to improve the spatial resolution of a reconstructed image of a cerebral ischemia patient and the quality of brain imaging, the invention provides a data compensation method for reconstructing cerebral ischemia conductivity distribution, wherein the method uses the compensated boundary voltage for image reconstruction.
Disclosure of Invention
The invention provides a data compensation method for reconstructing cerebral ischemia conductivity distribution, which is characterized in that a cerebral full-field boundary measurement voltage value is sent into a trained fully-connected neural network, so that the dehydration degree is determined, a corresponding model boundary measurement voltage value caused by independent scalp dehydration is found by the dehydration degree and then by a preceding test sequence, then compensation is carried out, the reconstruction of an intracranial cerebral ischemia image is realized by combining coordinate information of pixels, and finally, an adaptive threshold filtering algorithm is used for carrying out binarization processing on an image reconstruction result.
The invention adopts the following technical proposal to solve the technical problems, and is a data compensation method for reconstructing cerebral ischemia conductivity distribution, which is characterized by comprising the following specific steps:
s1, determining the shape and structure information of the cranium, constructing a standard 2D elliptical cranium model on a computer, and fusing conductivity information of different tissues of the cranium into tissue structures of different layers, wherein the model is of a three-layer structure of scalp, skull and brain tissue, and the conductivities of a scalp layer, a skull layer and a brain tissue layer are respectively set to be 0.44S/m, 0.012S/m and 0.163S/m so as to simulate cerebral edema;
s2, adopting an electrical impedance tomography system with 16 electrodes, placing a No. 1 electrode at the forefront point of a craniocerebral model, then attaching 16 electrodes on the scalp layer in an equidistant mode in a anticlockwise mode, under the relative current excitation and adjacent voltage measurement modes, firstly applying safe current excitation to the No. 1 electrode, and simultaneously grounding the No. 9 electrode, and carrying out detection on the conditions of 2-3, 3-4, 6-7 and 7-8;10-11, 11-12. 14-15, 15-16 total 12 electrode pairs voltage measurements, 12 voltage measurements being obtained as a first set of measurement data, and so on, 2-16 electrodes being excited sequentially in the same way, grounding the electrode opposite to the excitation electrode, measuring voltage values on the rest electrode pairs to obtain 16 groups of measurement data, wherein each group of measurement data comprises 12 voltage measurement values, and after each electrode is excited by traversing, 192 voltage measurement values are obtained in total;
step S3, obtaining a craniocerebral full-field boundary voltage measurement value U of the brain tissue and scalp layer in the simultaneous dehydration process by the electrical impedance tomography mode bs Wherein the degree of dehydration d= { d 1 ,d 2 ,…d i Subscript i is used to mark the size of its corresponding dehydration level, and the value of i is: i= {1,2, ··9,10}, the spacing step is set to 1%, i.e. d 1 Represents the degree of dehydration of 1%, d 2 Indicating a degree of dehydration of 2%, and so on, the measured value U of the boundary voltage of the full-field of the cranium measured each time bs And the corresponding dehydration processDegree d i Forming a sample S, measuring a plurality of groups of samples to form a training data set D, using the training data set D for training of the network, and then measuring individual scalp dehydration dE [1%,10%]The resulting model boundary measurement voltage value U s The method comprises the steps of carrying out a first treatment on the surface of the Measuring voltage value U by model boundary s And the corresponding degree of dehydration d i Constructing a priori sequence P s ;
S4, constructing a fully-connected neural network, which mainly comprises a straightening layer, an input layer, an hidden layer and an output layer;
step S401, in the forward propagation process, to match the input layer, the craniocerebral full-field boundary voltage measurement is performedBy straightening the layer to become +>And measuring the craniocerebral full-field boundary voltage (I)>As input to a fully connected neural network;
in step S402, the hidden layer has 60 neurons activated by a linear rectifying unit (ReLU) function for generating a nonlinear map, which is mathematically expressed as:
where m represents the input, the neuron is not activated if m is negative or equal to 0, otherwise the output is m,
there are ten neurons in the output layer, the activation function is SoftMax, which converts multiple inputs into output data that sums to 1, and for the ith neuron, softMax is described as:
wherein S is i The output layer is the output of the output layer, e is the output of the neurons, j is the number of the neurons, and after the softMax function is activated, the output of the output layer shows the probabilities of ten different classifications;
step S403, measuring the craniocerebral full-field boundary voltage U bs Corresponding degree of dehydration d i As the output of the fully connected neural network, 10 layers are set;
step S5, training the fully-connected neural network, wherein a loss function H used for training is as follows:
wherein p (n) represents a flag value, and q (n) represents a network output value;
step S6, training by using a Adam (Adaptive Momentum Estimation) optimizer, wherein the learning rate and regularization parameters are set to be 0.0001 and 0.000001 respectively;
step S7, obtaining actual dehydration boundary voltage measurement values U of different cerebral ischemia patients or different time periods of the same cerebral ischemia patient according to the EIT measurement mode real Will actually dehydrate the boundary voltage measurement U real Inputting into a trained fully connected neural network, and obtaining the dehydration degree d at the moment through forward propagation i ;
Step S8, prior verification sequence P s Find the corresponding dehydration degree d i Model boundary measurement voltage value U of (2) s Resulting in a compensated voltage U ', i.e., U' =u real -U s ;
Step S9, regarding the electrical tomography problem as an inverse problem u'. Apprxeq.A.g, wherein A is a sensitivity matrix, g is a conductivity variation, and based on an L1 regularization method, the conductivity distribution is estimated asIn the formula, lambda is regularization parameter for balancing fidelity term +.>And penalty term g 1 Weights in between;
step S10, solving the optimal conductivity distribution in the step S9 by adopting an alternate direction multiplier method, fusing the obtained optimal conductivity distribution with the position information of the image pixels, and reconstructing a compensated conductivity distribution image of the craniocerebral ischemia;
and S11, performing binarization processing on the compensated image reconstruction result by adopting an adaptive threshold filtering algorithm, wherein the value higher than the threshold is set to be 1, and the value lower than the threshold is set to be 0.
Compared with the prior art, the invention has the following advantages and beneficial effects: the invention provides a data compensation method for reconstructing cerebral ischemia conductivity distribution for the first time, which comprises the steps of sending a cerebral full-field boundary measurement voltage value into a trained fully-connected neural network to determine the dehydration degree, finding a corresponding model boundary measurement voltage value caused by independent scalp dehydration according to the dehydration degree and a preceding test sequence, then compensating, combining coordinate information of pixels, reconstructing an intracranial image, and finally performing binarization processing on an image reconstruction result by using a self-adaptive threshold filtering algorithm. The method can effectively improve the spatial resolution of the reconstructed image of the intracranial cerebral hemorrhage, further improve the brain imaging quality, and has great application potential in brain functional medical imaging.
Drawings
FIG. 1 is a block flow diagram of a data compensation method for cerebral ischemia conductivity distribution reconstruction provided by the present invention;
FIG. 2 is a diagram showing a single-section measured field, electrode distribution, excitation current and measurement voltage pattern of the 2D elliptical rabbit head model of the present invention;
FIG. 3 is a block diagram of a fully connected neural network;
FIG. 4 is a schematic diagram of the image reconstruction results of the three models under the noise-free condition, when not compensated, and according to the method of the present invention;
fig. 5 shows the Correlation Coefficient (CC) and the mean square error (RMSE) of the reconstruction results in the uncompensated and present method, respectively, in three models.
Detailed Description
The invention provides a data compensation method for reconstructing cerebral ischemia conductivity distribution, which is described in detail with reference to the accompanying drawings and the embodiment.
The data compensation method for reconstructing cerebral ischemia conductivity distribution aims at improving cerebral ischemia image reconstruction quality during intracranial dehydration, and aims at solving the problems of image artifact reconstruction, incapability of presenting a target object and the like during simultaneous dehydration of scalp and brain tissue, and the like, and improves spatial resolution of cerebral ischemia image reconstruction and effectively improves quality of reconstructed images by using voltage after data compensation for image reconstruction.
As shown in fig. 1, a flow chart of a data compensation method for reconstructing cerebral ischemia conductivity distribution provided by the invention comprises the following specific steps:
s1, determining the shape and structure information of the cranium, constructing a standard 2D elliptical cranium model on a computer, and fusing conductivity information of different tissues of the cranium into tissue structures of different layers. The model is a three-layer structure of scalp, skull and brain tissue, and the conductivities of the scalp layer, the skull layer and the brain tissue layer are respectively set to be 0.44S/m, 0.012S/m and 0.163S/m so as to simulate cerebral edema.
And S2, using a 16-electrode electrical impedance tomography system, placing a No. 1 electrode at the forefront point of the cranium brain model, and then attaching the 16 electrodes to the cortex in an equidistant mode in a anticlockwise mode. Under the relative current excitation and adjacent voltage measurement modes, firstly, applying safe current excitation to the electrode No. 1, and simultaneously grounding the electrode No. 9, and carrying out current excitation on the electrode No. 2-3, 6-7 and 7-8;10-11, 11-12. 14-15, 15-16 total 12 electrode pair voltage measurements, 12 voltage measurements were obtained as a first set of measurement data. And similarly, exciting 2-16 electrodes in turn according to the same method, grounding the electrode opposite to the exciting electrode, and measuring voltage values on the rest electrode pairs to obtain 16 groups of measurement data. Each set of measurement data contained 12 voltage measurements, which were obtained for a total of 192 voltage measurements after each electrode was stimulated in a traversal.
Step S3, passing through the electrical impedance tomography methodObtaining a measurement value U of a craniocerebral full-field boundary voltage when brain tissue and scalp layers are dehydrated simultaneously bs Wherein the degree of dehydration d= { d 1 ,d 2 ,…d i Subscript i is used to mark the size of its corresponding degree of dehydration. The value of i is as follows: i= {1,2, ··9,10}, the spacing step is set to 1%, i.e. d 1 Represents the degree of dehydration of 1%, d 2 Indicating a degree of dehydration of 2% and so on. The measured value U of the craniocerebral full-field boundary voltage measured each time bs And the corresponding degree of dehydration d i One sample S is formed and a plurality of sets of samples are measured to form a training data set D. The training dataset D is used for training of the network. Next, measure individual scalp dehydration d.epsilon.1%, 10%]The resulting model boundary measurement voltage value U s The method comprises the steps of carrying out a first treatment on the surface of the Measuring voltage value U by model boundary s And the corresponding degree of dehydration d i Constructing a priori sequence P s 。
And S4, constructing a fully-connected neural network, which mainly comprises a straightening layer, an input layer, an hidden layer and an output layer.
Step S401, in the forward propagation process, to match the input layer, the craniocerebral full-field boundary voltage measurement is performedBy straightening the layer to become +>And measuring the craniocerebral full-field boundary voltage (I)>As input to a fully connected neural network.
In step S402, the hidden layer has 60 neurons, activated by a linear rectifying unit (ReLU) function. The ReLU function is used to generate a nonlinear map, which is expressed mathematically as:
where m represents an input. If m is negative or equal to 0, the neuron is not activated. Otherwise, outputting as m;
there are ten neurons in the output layer and the activation function is SoftMax, which converts multiple inputs into output data that sums to 1. For the ith neuron, softMax is described as:
wherein S is i Is the output of the output layer, e is the output of the neuron, and j is the number of neurons. After the softMax function is activated, the output of the output layer displays the probabilities of ten different classifications;
step S403, measuring the craniocerebral full-field boundary voltage U bs Corresponding degree of dehydration d i As the output of the fully connected neural network, 10 layers are set;
step S5, training the fully-connected neural network, wherein a loss function H used for training is as follows:
wherein p (n) represents a flag value, and q (n) represents a network output value;
step S6, training by using a Adam (Adaptive Momentum Estimation) optimizer, wherein the learning rate and regularization parameters are set to be 0.0001 and 0.000001 respectively;
step S7, obtaining actual dehydration boundary voltage measurement values U of different cerebral ischemia patients or different time periods of the same cerebral ischemia patient according to the EIT measurement mode real Will actually dehydrate the boundary voltage measurement U real Inputting into a trained fully connected neural network, and obtaining the dehydration degree d at the moment through forward propagation i ;
Step S8, prior verification sequence P s Find the corresponding dehydration degree d i Model boundary measurement voltage value U of (2) s Resulting in a compensated voltage U ', i.e., U' =u real -U s ;
Step S9, regarding the electrical tomography problem as an inverse problem u'. Apprxeq.A.g, wherein A is a sensitivity matrix, g is a conductivity variation, and based on an L1 regularization method, the conductivity distribution is estimated asIn the formula, lambda is regularization parameter for balancing fidelity term +.>And penalty term g 1 Weights in between;
and S10, solving the optimal conductivity distribution in the step S9 by adopting an alternate direction multiplier method. Reconstructing a compensated conductivity distribution image of the craniocerebral ischemia by fusing the position information of the pixels of the obtained optimal conductivity distribution fusion image;
and S11, performing binarization processing on the compensated image reconstruction result by adopting an adaptive threshold filtering algorithm, wherein the value higher than the threshold is set to be 1, and the value lower than the threshold is set to be 0.
As shown in fig. 2, for a single section of EIT rabbit head model of 2D elliptical 16 electrodes, the 16 electrodes are uniformly distributed outside the field using a mode of relative excitation and adjacent measurement.
As shown in fig. 3, the fully connected neural network diagram includes a straightening layer, an input layer, a hiding layer and an output layer.
Three different positions of the target object on the brain tissue layer are selected as an embodiment, the real distribution of the target object in the field is shown in the first row of fig. 4, in this embodiment, COMSOL Multiphysics 5.4.5.4 and MATLAB R2016a are used for combined simulation modeling, simulation parameter setting refers to real rabbit biological tissue parameters, brain background conductivity is set to be 0.149S/m of rabbit brain cerebrospinal fluid conductivity, and conductivity of ischemia target inclusion is set to be 0.06S/m. The first behavior is a reconstructed image obtained when uncompensated, the background is not clear, a large number of electrode artifacts appear on the boundary, and the size and the position of a target object are hardly reconstructed. In contrast, the second behavior is that the reconstructed image obtained using the method of the present invention, for all three models A, B, C, the object can be reconstructed well with few artifacts in the background. The result shows that the method can effectively improve the quality of intracranial cerebral ischemia image reconstruction, can improve the resolution of image reconstruction, and has important guidance for determining the illness state in time in clinic.
As shown in fig. 5, the correlation coefficients (correlation coefficient, CC) and root mean square errors (Root Mean Square Error) of the four models when uncompensated and the reconstruction result of the method according to the present invention are shown. (a) CC values for the four models and (b) RMSE values for the four models. The expression of CC is shown in the following formula, and the larger the correlation coefficient value of the reconstructed image is, the better the reconstructed image quality is.
In the formula g c G for the calculated conductivity a Representing the actual electrical conductivity of the material,and->Respectively represent the conductivity value on the e-th element, ">And->Respectively represent g c And g is equal to a Average value of (2).
The RMSE expression is shown in the following formula, and the smaller the root mean square error value of the reconstructed image is, the better the reconstructed image quality is.
Where n=7271 is the reconstructed effective pixel point,and Δg i The ith pixel of predicted and true conductivity distribution variation in the target area, respectively.
It can be seen that the correlation coefficient of the reconstructed image by the method is far greater than that of the reconstructed image before uncompensated, which further proves the superiority of the method and can effectively improve the quality of the reconstructed image. Even in the case where the degree of dehydration is low, good performance can be exhibited.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather to enable any modification, equivalent replacement, improvement or the like to be made within the spirit and principles of the invention.
Claims (1)
1. A data compensation method for reconstructing cerebral ischemia conductivity distribution is characterized by comprising the following specific steps:
s1, determining the shape and structure information of the cranium, constructing a standard 2D elliptical cranium model on a computer, and fusing conductivity information of different tissues of the cranium into tissue structures of different layers, wherein the model is of a three-layer structure of scalp, skull and brain tissue, and the conductivities of a scalp layer, a skull layer and a brain tissue layer are respectively set to be 0.44S/m, 0.012S/m and 0.163S/m so as to simulate cerebral edema;
s2, adopting an electrical impedance tomography system with 16 electrodes, placing a No. 1 electrode at the forefront point of a craniocerebral model, then attaching 16 electrodes on the scalp layer in an equidistant mode in a anticlockwise mode, under the relative current excitation and adjacent voltage measurement modes, firstly applying safe current excitation to the No. 1 electrode, and simultaneously grounding the No. 9 electrode, and carrying out detection on the conditions of 2-3, 3-4, 6-7 and 7-8;10-11, 11-12. 14-15, 15-16 total 12 electrode pairs voltage measurements, 12 voltage measurements being obtained as a first set of measurement data, and so on, 2-16 electrodes being excited sequentially in the same way, grounding the electrode opposite to the excitation electrode, measuring voltage values on the rest electrode pairs to obtain 16 groups of measurement data, wherein each group of measurement data comprises 12 voltage measurement values, and after each electrode is excited by traversing, 192 voltage measurement values are obtained in total;
step S3, obtaining a craniocerebral full-field boundary voltage measurement value U of the brain tissue and scalp layer in the simultaneous dehydration process by the electrical impedance tomography mode bs Wherein the degree of dehydration d= { d 1 ,d 2 ,…d i Subscript i is used to mark the size of its corresponding dehydration level, and the value of i is: i= {1,2, ··9,10}, the spacing step is set to 1%, i.e. d 1 Represents the degree of dehydration of 1%, d 2 Indicating a degree of dehydration of 2%, and so on, the measured value U of the boundary voltage of the full-field of the cranium measured each time bs And the corresponding degree of dehydration d i Forming a sample S, measuring a plurality of groups of samples to form a training data set D, using the training data set D for training of the network, and then measuring individual scalp dehydration dE [1%,10%]The resulting model boundary measurement voltage value U s The method comprises the steps of carrying out a first treatment on the surface of the Measuring voltage value U by model boundary s And the corresponding degree of dehydration d i Constructing a priori sequence P s ;
S4, constructing a fully-connected neural network, which mainly comprises a straightening layer, an input layer, an hidden layer and an output layer;
step S401, in the forward propagation process, to match the input layer, the craniocerebral full-field boundary voltage measurement is performedBy straightening the layer to become +>And measuring the craniocerebral full-field boundary voltage (I)>As input to a fully connected neural network;
in step S402, the hidden layer has 60 neurons activated by a linear rectifying unit (ReLU) function for generating a nonlinear map, which is mathematically expressed as:
where m represents the input, the neuron is not activated if m is negative or equal to 0, otherwise the output is m,
there are ten neurons in the output layer, the activation function is SoftMax, which converts multiple inputs into output data that sums to 1, and for the ith neuron, softMax is described as:
wherein S is i The output layer is the output of the output layer, e is the output of the neurons, j is the number of the neurons, and after the softMax function is activated, the output of the output layer shows the probabilities of ten different classifications;
step S403, measuring the craniocerebral full-field boundary voltage U bs Corresponding degree of dehydration d i As the output of the fully connected neural network, 10 layers are set;
step S5, training the fully-connected neural network, wherein a loss function H used for training is as follows:
wherein p (n) represents a flag value, and q (n) represents a network output value;
step S6, training by using a Adam (Adaptive Momentum Estimation) optimizer, wherein the learning rate and regularization parameters are set to be 0.0001 and 0.000001 respectively;
step S7, obtaining actual dehydration boundary voltage measurement values U of different cerebral ischemia patients or different time periods of the same cerebral ischemia patient according to the EIT measurement mode real Will actually dehydrate the boundary voltage measurement U real Inputting into a trained fully connected neural network, and obtaining through forward propagationThe degree of dehydration d at this time is obtained i ;
Step S8, prior verification sequence P s Find the corresponding dehydration degree d i Model boundary measurement voltage value U of (2) s Resulting in a compensated voltage U ', i.e., U' =u real -U s ;
Step S9, regarding the electrical tomography problem as an inverse problem u'. Apprxeq.A.g, wherein A is a sensitivity matrix, g is a conductivity variation, and based on an L1 regularization method, the conductivity distribution is estimated asIn the formula, lambda is regularization parameter for balancing fidelity term +.>And penalty term g 1 Weights in between;
step S10, solving the optimal conductivity distribution in the step S9 by adopting an alternate direction multiplier method, fusing the obtained optimal conductivity distribution with the position information of the image pixels, and reconstructing a compensated conductivity distribution image of the craniocerebral ischemia;
and S11, performing binarization processing on the compensated image reconstruction result by adopting an adaptive threshold filtering algorithm, wherein the value higher than the threshold is set to be 1, and the value lower than the threshold is set to be 0.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310731047.4A CN116869504A (en) | 2023-06-19 | 2023-06-19 | Data compensation method for cerebral ischemia conductivity distribution reconstruction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310731047.4A CN116869504A (en) | 2023-06-19 | 2023-06-19 | Data compensation method for cerebral ischemia conductivity distribution reconstruction |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116869504A true CN116869504A (en) | 2023-10-13 |
Family
ID=88268895
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310731047.4A Pending CN116869504A (en) | 2023-06-19 | 2023-06-19 | Data compensation method for cerebral ischemia conductivity distribution reconstruction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116869504A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117523089A (en) * | 2023-11-06 | 2024-02-06 | 南京航空航天大学 | Method for solving reconstruction matrix based on discrete training set, three-dimensional imaging method and storage medium |
-
2023
- 2023-06-19 CN CN202310731047.4A patent/CN116869504A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117523089A (en) * | 2023-11-06 | 2024-02-06 | 南京航空航天大学 | Method for solving reconstruction matrix based on discrete training set, three-dimensional imaging method and storage medium |
CN117523089B (en) * | 2023-11-06 | 2024-04-30 | 南京航空航天大学 | Method for solving reconstruction matrix based on discrete training set, three-dimensional imaging method and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101564294B (en) | Method for structural information fused electrical impedance tomography | |
Kauppinen et al. | Sensitivity distribution visualizations of impedance tomography measurement strategies | |
Antonakakis et al. | The effect of stimulation type, head modeling, and combined EEG and MEG on the source reconstruction of the somatosensory P20/N20 component | |
CN111311703B (en) | Electrical impedance tomography image reconstruction method based on deep learning | |
CN111616708A (en) | Image reconstruction method for accurately identifying cerebral apoplexy intracranial lesion area | |
US11605162B2 (en) | Systems and methods for determining a fluid and tissue volume estimations using electrical property tomography | |
CN116869504A (en) | Data compensation method for cerebral ischemia conductivity distribution reconstruction | |
CN114708350A (en) | Conductivity visualization method for electrical impedance tomography of brain | |
CN114463493A (en) | Transcranial magnetic stimulation electric field rapid imaging method and model based on coding and decoding structure | |
CN114663355A (en) | Hybrid neural network method for reconstructing conductivity distribution image of cerebral hemorrhage | |
CN110811596B (en) | Noninvasive cardiac potential reconstruction method based on low rank and sparse constraint and non-local total variation | |
CN113902825A (en) | Lung electrical impedance imaging method based on VDD-Net | |
Li et al. | Electrical-impedance-tomography imaging based on a new three-dimensional thorax model for assessing the extent of lung injury | |
CN113470812A (en) | Heart transmembrane potential reconstruction method based on graph convolution neural network and iterative threshold contraction algorithm | |
Li et al. | SAR-CGAN: Improved generative adversarial network for EIT reconstruction of lung diseases | |
Medani et al. | Realistic head modeling of electromagnetic brain activity: an integrated Brainstorm-DUNEuro pipeline from MRI data to the FEM solutions | |
CN114041773A (en) | Apoplexy position classification method based on electrical impedance tomography measurement framework | |
Bayford et al. | Focus on recent advances in electrical impedance tomography | |
CN110393522B (en) | Non-invasive heart electrophysiological inversion method based on total variation constraint of graph | |
CN115736877A (en) | Focusing type cardiac electrical impedance imaging method | |
Fernández-Corazza et al. | A novel bounded EIT protocol to generate inhomogeneous skull conductivity maps non-invasively | |
Liu et al. | Elastography mapped by deep convolutional neural networks | |
CN116369890B (en) | Imaging method, solving method and imaging system based on twin electrode detection | |
CN115908598A (en) | Compensation method for single-electrode contact impedance change in electroencephalogram impedance tomography | |
CN115359142A (en) | Method for improving image reconstruction quality in cerebral edema treatment process |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |