Diagnostic system and method based on electrical impedance imaging
Patent Information
- Application Number
- CN202380018345.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-01-28
- Publication Date
- 2024-11-12
AI Technical Summary
Existing electrical impedance imaging techniques are difficult to effectively diagnose and classify the health status and disease stages of organs, especially in the liver and kidneys, and traditional methods are highly invasive, costly, and have measurement errors and pain risks.
By using a computer-impedance imaging data set, machine learning models combine conductivity characteristics and anthropometric measurement characteristics to perform group source separation and spectrum decomposition, extract organ-related conductivity characteristics, and then determine the health status of the organ and The existence of the disease and the severity of the disease.
A non-invasive, low-cost diagnosis is achieved, improving the accuracy and reliability of the healthy status of the liver and kidneys, reducing diagnostic errors, and providing portable and economical disease screening and monitoring tools.
Smart Images

Figure CN118946309A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a diagnostic system and method based on electrical impedance tomography (EIT). Background Art
[0002] Electrical impedance imaging is a medical imaging technique that can be used to determine the electrical conductivity, permittivity, and / or electrical impedance of a body part of a subject (eg, an animal, a human, etc.). Summary of the invention
[0003] In a first aspect, a computer-implemented method is provided, comprising: processing an electrical impedance imaging dataset of a subject to determine one or more conductivity characteristics associated with a tissue or organ of the subject, and determining a health state or condition of the tissue or organ of the subject based at least on the one or more determined conductivity characteristics.
[0004] Optionally, the determining comprises: determining whether the subject suffers from a disease associated with the tissue or organ based at least on the one or more determined conductivity characteristics, and optionally: further classifying the stage or severity of the disease associated with the tissue or organ.
[0005] Optionally, the determining comprises: using a machine learning based processing model to process at least the one or more determined conductivity characteristics of the subject and one or more anthropometric characteristics of the subject to determine a quantitative or qualitative parameter associated with a health state or condition of a tissue or organ of the subject.
[0006] Optionally, the determining comprises: using a machine learning based processing model to process (i) the one or more determined conductivity characteristics of the subject, (ii) one or more anthropometric characteristics of the subject, and (iii) one or more determined conductivity characteristics of one or more reference subjects and / or one or more determined conductivity characteristics of a group comprising the subject and one or more reference subjects to determine a quantitative or qualitative parameter associated with a health state or condition of a tissue or organ of the subject.
[0007] Alternatively, the processing model based on machine learning includes a regression model. Alternatively, the processing model based on machine learning includes a classification model. The processing model based on machine learning can be a cyclic model or a non-cyclic model. For example, the processing model based on machine learning can include a recursive neural network, a recurrent neural network, a long short-term memory network model, a Markov process, reinforcement learning, a gated recurrent unit model, a deep neural network, a convolutional neural network, a support vector machine, a principal component analysis, a logistic regression, a decision tree / forest, an integrated learning method (combined model), regression (Bayesian / polynomial / regression), stochastic gradient descent, linear discriminant analysis, neighbor classification or regression, naive Bayes, etc. The processing model based on machine learning can be trained to perform a specific processing or classification task associated with a diagnostic application.
[0008] Optionally, the one or more anthropometric characteristics include or are associated with one or more of: the subject's age, the subject's weight, the subject's height, the subject's waist circumference, the subject's waist-over-height ratio, the subject's body mass index (BMI), the subject's gender, and the subject's race. Other anthropometric characteristics are also possible.
[0009] Optionally, the quantitative or qualitative parameter associated with the health state or condition of the subject's tissue or organ includes: a value associated with an estimated performance of the subject's tissue or organ. For example, in the case of the kidney, the value may be an estimated glomerular filtration rate (eGFR) or a related (e.g., arithmetically related) value. For example, in the case of the liver, the value may be a controlled attenuation parameter (CAP) score or a related (e.g., arithmetically related) value.
[0010] Optionally, the determination further comprises comparing a quantitative or qualitative parameter associated with the health state or condition of the subject's tissue or organ with reference parameter data to determine whether the subject suffers from a disease associated with the tissue or organ.
[0011] Optionally, the determining further comprises classifying a stage or severity of a disease associated with the tissue or organ based on the comparison.
[0012] Optionally, the electrical impedance imaging data set includes electrical impedance imaging data acquired from a region of the subject containing the tissue or organ. Optionally, the electrical impedance imaging data set is acquired by: (a) providing an excitation signal of a frequency to the subject via an electrode attached to the region of the subject; (b) measuring a response signal received via the electrode due to providing the excitation signal; (c) repeating steps (a) and (b) for multiple frequencies. The electrical impedance imaging data set includes multiple electrical impedance imaging data subsets, each of which is associated with a corresponding one of the multiple frequencies.
[0013] Optionally, the processing includes: (i) processing the electrical impedance imaging dataset to obtain a frequency difference (fd) electrical impedance imaging dataset, which includes multiple frequency difference electrical impedance imaging data subsets; (ii) using the frequency difference electrical impedance imaging dataset and one or more reference frequency difference electrical impedance imaging datasets corresponding to one or more reference subjects to perform a source separation operation to determine the component related to the tissue or organ in the frequency difference electrical impedance imaging dataset and each component related to the tissue or organ in the one or more reference frequency difference electrical impedance imaging datasets; and (iii) using the component related to the tissue or organ in the frequency difference electrical impedance imaging dataset and optionally each component related to the tissue or organ in the one or more reference frequency difference electrical impedance imaging datasets to perform a conductivity feature extraction operation to at least determine the one or more conductivity features.
[0014] Optionally, the processing further includes: preprocessing the electrical impedance imaging dataset before the processing in (i) so that the electrical impedance imaging dataset processed in (i) is a preprocessed electrical impedance imaging dataset.
[0015] Optionally, the pre-processing of the electrical impedance imaging data set comprises filtering and / or smoothing each of the plurality of electrical impedance imaging data subsets.
[0016] Optionally, the preprocessing of the electrical impedance imaging data set includes: processing the electrical impedance imaging data set using a classifier model to determine the corresponding performance of each of the multiple electrodes, the performance being associated with the quality of the response signal or data obtained from the corresponding electrode; and preventing the response signal or data obtained via any one or more of the multiple electrodes determined to have insufficient performance from being included in the processed electrical impedance imaging data set. Optionally, the processing of the electrical impedance imaging data set in (ii) includes: for each of the corresponding one or more of the multiple processed electrical impedance imaging data subsets, determining the corresponding difference between the corresponding processed electrical impedance imaging data subset and the reference electrical impedance imaging data subset, thereby obtaining the multiple frequency difference electrical impedance imaging data subsets, each of which is respectively associated with the corresponding difference between the corresponding processed electrical impedance imaging data subset and the reference electrical impedance imaging data subset.
[0017] Optionally, the reference electrical impedance imaging data subset comprises at least one of the plurality of processed electrical impedance imaging data subsets.
[0018] Optionally, performing the set of source separation operations includes: performing a dimensionality reduction operation on the frequency difference electrical impedance imaging dataset and one or more reference frequency difference electrical impedance imaging datasets corresponding to one or more reference subjects.
[0019] Optionally, performing the conductivity feature extraction operation includes: using the component associated with the tissue or organ in the frequency difference electrical impedance imaging data set to determine one or more conductivity features of the subject. The one or more conductivity features of the subject may include one or more statistical conductivity features of the subject (e.g., the mean, median, mode, standard deviation, etc. of the frequency difference electrical impedance imaging data of the subject or any part of the electrical impedance imaging data (which can be represented as a conductivity map or image)).
[0020] Optionally, performing the conductivity feature extraction operation includes: utilizing the component associated with the tissue or organ in the frequency difference electrical impedance imaging data set and the corresponding component associated with the tissue or organ in each of the one or more reference frequency difference electrical impedance imaging data sets to determine one or more conductivity features of the group including the subject and the one or more reference subjects. The determination of the health state or condition of the tissue or organ of the subject is also based on the one or more conductivity features of the group. The one or more conductivity features of the subject may include one or more statistical conductivity features of the group (e.g., the mean, median, mode, standard deviation, etc. (which may be represented as a conductivity map or image) of the frequency difference electrical impedance imaging data or any part of the electrical impedance imaging data of the group).
[0021] Optionally, the determination of the health state or condition of the tissue or organ of the subject is also based on the one or more conductivity characteristics of the one or more reference subjects. The one or more conductivity characteristics of the one or more reference subjects may include one or more statistical conductivity characteristics of the group (e.g., the mean, median, mode, standard deviation, etc. (which may be represented as a conductivity map or image) of the frequency difference electrical impedance imaging data or any portion of the electrical impedance imaging data of the one or more reference subjects).
[0022] Optionally, the tissue or organ comprises lung, kidney, liver or heart.
[0023] In a second aspect, a system is provided, comprising one or more processors and a memory storing one or more programs configured to be executed by the one or more processors. The one or more programs include instructions for performing or facilitating the performance of the computer-implemented method of the first aspect. In a third aspect, a non-transitory computer-readable storage medium is provided, storing one or more programs configured to be executed by the one or more processors. The one or more programs include instructions for performing or facilitating the performance of the computer-implemented method of the first aspect.
[0024] In a fourth aspect, there is provided a computer program product comprising instructions which, when executed by a computer, cause or facilitate the computer to perform the computer-implemented method of the first aspect.
[0025] In a fifth aspect, a computer-implemented method is provided, comprising: processing electrical impedance imaging data acquired from a subject, the electrical impedance imaging data comprising a plurality of potential data sets, each potential data set acquired at an electrode attached to (directly or indirectly) the subject in response to an excitation signal (e.g., current) of a set frequency applied sequentially to each electrode, the applied set frequency being different for different data sets and the same for the same data set; and determining whether the subject suffers from a disease based on the processing.
[0026] Optionally, the plurality of potential data sets include at least a first potential data set associated with an excitation signal (e.g., current) of a first frequency, a second potential data set associated with an excitation signal (e.g., current) of a second frequency, and a third potential data set associated with an excitation signal (e.g., current) of a third frequency. Optionally, the processing includes: determining the difference between the first and second potential data sets to obtain a first potential difference data set; determining the difference between the first and third second potential data sets to obtain a second potential difference data set; applying the first and second potential difference data sets to a spectral unmixing model to determine parameters indicating the effects of tissue type i on the first and second potential difference data sets; and determining the value of the parameter associated with the disease based on the first and second corrected potential difference data sets and one or more anthropometric indicators of the subject. In this example, the first potential data set is used as a reference data set. The reference data set may have the highest signal-to-noise ratio among all data sets.
[0027] Optionally, the spectrum decomposition model includes where ΔV(ω) is the first and second potential difference data sets, a i is a parameter representing the effect caused by tissue type i, Δσ i (ω) is a predetermined spectrum specific to tissue type i, M is the total number of tissue types, and ∈ is an error term. In one example, the error term is 0, in which case the spectral decomposition model includes where ΔV(ω) is the first and second potential difference data sets, a i is a parameter representing the effect caused by tissue type i, Δσ i (ω) is the predetermined spectrum specific to tissue type i, and M is the total number of tissue types.
[0028] Optionally, the multiple potential data sets include at least a first potential data set associated with an excitation signal (e.g., current) of a first frequency, a second potential data set associated with an excitation signal (e.g., current) of a second frequency, a third potential data set associated with an excitation signal (e.g., current) of a third frequency, and a fourth potential data set associated with an excitation signal (e.g., current) of a fourth frequency. Optionally, the processing includes determining the difference between the first and second potential data sets to obtain a first potential difference data set; determining the difference between the first and third second potential data sets to obtain a second potential difference data set; determining the difference between the first and fourth second potential data sets to obtain a third potential difference data set; applying the first, second, and third potential difference data sets to a spectral decomposition model to determine parameters indicating the effects of tissue type i on the first, second, and third potential difference data sets; and determining values of parameters associated with the disease based on the first, second, and third corrected potential difference data sets and one or more anthropometric indicators of the subject. Optionally, the spectral decomposition model includes where ΔV(ω) is the first, second and third potential difference data sets, a i is a parameter representing the effect caused by tissue type i, Δσ i (ω) is a predetermined spectrum specific to tissue type i, M is the total number of tissue types, and ∈ is an error term. In one example, the error term is 0, in which case the spectral decomposition model includes where ΔV(ω) is the first and second potential difference data sets, a i is a parameter representing the effect caused by tissue type i, Δσ i (ω) is the predetermined spectrum specific to tissue type i, and M is the total number of tissue types.
[0029] Optionally, determining whether the subject has a disease comprises comparing the determined value with a predetermined reference scale. The predetermined reference scale may comprise predetermined values of parameters classified according to whether the disease is present and, optionally, the severity of the disease.
[0030] Optionally, the first potential difference data set may be processed to provide a conductivity variation map (eg, an average conductivity variation map) of the subject.
[0031] Optionally, the second potential difference data set may be processed to provide a conductivity variation map (eg, an average conductivity variation map) of the subject.
[0032] Optionally, the parameters associated with the disease include controlled attenuation parameters.
[0033] Optionally, the one or more anthropometric indices of the subject include an index of the ratio of waist circumference to height (i.e., the waist circumference of the subject divided by the height of the subject). Optionally, the one or more anthropometric indices of the subject include the age of the subject. Optionally, the one or more anthropometric indices of the subject include the chest circumference of the subject.
[0034] Optionally, the processing further comprises filtering the potential data set before determining the difference. The filtering may remove outliers.
[0035] Optionally, the computer-implemented method further comprises acquiring the electrical impedance imaging data from the subject.
[0036] Optionally, the disease includes liver disease, lung disease, kidney disease, etc. In one example, the disease includes fatty liver disease (eg, non-alcoholic fatty liver disease).
[0037] Optionally, the computer-implemented method further comprises determining a severity of the disease based on the processing.
[0038] Optionally, the computer-implemented method also includes presenting the determined result to the user. The presentation may include displaying the result to the user. The result may include a "yes / no" result (regarding whether the subject suffers from the disease) and optionally the severity of the disease.
[0039] Optionally, the subject is a human. Optionally, the subject is a non-human animal.
[0040] In a sixth aspect, a system is provided, comprising: one or more processors arranged to process electrical impedance imaging data acquired from a subject, the electrical impedance imaging data comprising a plurality of electrical potential data sets, each electrical potential data set being obtained at electrodes attached (directly or indirectly) to the subject in response to an excitation signal (e.g., current) of a set frequency applied sequentially to each electrode, the applied set frequency being different for different data sets and the same for the same data set; and based on the processing, determining whether the subject suffers from a disease.
[0041] Optionally, the multiple potential data sets include at least a first potential data set associated with an excitation signal (e.g., current) of a first frequency, a second potential data set associated with an excitation signal (e.g., current) of a second frequency, and a third potential data set associated with an excitation signal (e.g., current) of a third frequency. Optionally, the one or more processors are arranged to: determine the difference between the first and second potential data sets to obtain a first potential difference data set; determine the difference between the first and third second potential data sets to obtain a second potential difference data set; apply the first and second potential difference data sets to a spectral decomposition model to determine parameters indicating the effects of tissue type i on the first and second potential difference data sets; and determine the value of the parameter associated with the disease based on the first and second corrected potential difference data sets and one or more anthropometric indicators of the subject. In this example, the first potential data set is used as a reference data set. The reference data set may have the highest signal-to-noise ratio among all data sets.
[0042] Optionally, the spectrum decomposition model includes where ΔV(ω) is the first and second potential difference data sets, a i is a parameter representing the effect caused by tissue type i, Δσ i (ω) is a predetermined spectrum specific to tissue type i, M is the total number of tissue types, and ∈ is an error term. In one example, the error term is 0, in which case the spectral decomposition model includes where ΔV(ω) is the first and second potential difference data sets, a i is a parameter representing the effect caused by tissue type i, Δσ i (ω) is a predetermined spectrum specific to tissue type i, and M is the total number of tissue types. The system may include a memory storing the spectral decomposition model and operably connected to the one or more processors.
[0043] Optionally, the plurality of potential data sets include at least a first potential data set associated with an excitation signal (e.g., current) of a first frequency, a second potential data set associated with an excitation signal (e.g., current) of a second frequency, a third potential data set associated with an excitation signal (e.g., current) of a third frequency, and a fourth potential data set associated with an excitation signal (e.g., current) of a fourth frequency. Optionally, the one or more processors are arranged to: determine the difference between the first and second potential data sets to obtain a first potential difference data set; determine the difference between the first and third second potential data sets to obtain a second potential difference data set; determine the difference between the first and fourth second potential data sets to obtain a third potential difference data set; apply the first, second, and third potential difference data sets to a spectral decomposition model to determine a parameter indicating the effect of tissue type i on the first, second, and third potential difference data sets; and determine the value of the parameter associated with the disease based on the first, second, and third corrected potential difference data sets and one or more anthropometric indicators of the subject.
[0044] Optionally, the spectrum decomposition model includes where ΔV(ω) is the first, second and third potential difference data sets, a i is a parameter representing the effect caused by tissue type i, Δσ i (ω) is a predetermined spectrum specific to tissue type i, M is the total number of tissue types, and ∈ is an error term. In one example, the error term is 0, in which case the spectral decomposition model includes where ΔV(∈) is the first and second potential difference data sets, a i is a parameter representing the effect caused by tissue type i, Δσ i (ω) is a predetermined spectrum specific to tissue type i, and M is the total number of tissue types. The system may include a memory that stores the spectral decomposition model and is operably connected to the one or more processors. Optionally, the one or more processors are arranged to compare the determined value with a predetermined reference scale to determine whether the subject suffers from a disease. The predetermined reference scale may include predetermined values of parameters classified according to whether a disease exists and optionally the severity of the disease.
[0045] Optionally, the one or more processors are arranged to process the potential difference dataset to provide a conductivity variation map (eg a mean conductivity variation map) of the subject.
[0046] Optionally, the parameters associated with the disease include controlled attenuation parameters.
[0047] Optionally, the one or more anthropometric indices of the subject include an index of the ratio of waist circumference to height (i.e., the waist circumference of the subject divided by the height of the subject). Optionally, the one or more anthropometric indices of the subject include the age of the subject. Optionally, the one or more anthropometric indices of the subject include the chest circumference of the subject.
[0048] Optionally, the one or more processors are arranged to filter the potential data set before determining the difference. The filtering may remove outliers.
[0049] Optionally, the disease includes liver disease, lung disease, kidney disease, etc. In one example, the disease includes fatty liver disease (eg, non-alcoholic fatty liver disease).
[0050] Optionally, the one or more processors are arranged to determine a severity of the disease based on the processing.
[0051] Optionally, the system also includes an output device (e.g., a display) for presenting the determined result to a user. The presentation may include displaying the result to the user. The result may include a "yes / no" result (regarding whether the subject suffers from the disease) and optionally the severity of the disease.
[0052] Optionally, the subject is a human. Optionally, the subject is a non-human animal.
[0053] In a seventh aspect, a non-transitory computer-readable storage medium is provided, which stores one or more programs configured to be executed by one or more processors. The one or more programs include instructions for performing or facilitating the performance of the computer-implemented method of the fourth aspect.
[0054] In an eighth aspect, a computer program product is provided, comprising instructions which, when executed by a computer, cause or facilitate the computer to perform the computer-implemented method of the fourth aspect.
[0055] By considering the specific embodiments and the accompanying drawings, other features and aspects of the present invention will become apparent. Where appropriate and applicable, any feature described herein for one aspect or embodiment may be combined with any other feature described herein for any other aspect or embodiment. Depending on the context, degree terms such as "generally", "approximately", "substantially" or similar terms are used to take into account manufacturing tolerances, degradation, trends, tendencies, imperfect actual conditions, etc. For example, when a value is modified with a degree term (e.g., "approximately"), such an expression may include ±10%, ±5%, ±2%, or ±1% of the value.
[0056] Unless otherwise specified, the terms "connected," "coupled," "mounted," and the like are intended to include direct and indirect connections, couplings, mountings, etc. Unless otherwise specified or the context requires otherwise, the terms "conductivity" and the like refer to electrical conductivity or bioconductivity. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] Embodiments of the present invention will now be described by way of example with reference to the accompanying drawings, in which:
[0058] Figure 1 is a flow chart illustrating a computer-implemented electrical impedance imaging data processing method according to some embodiments of the present invention.
[0059] Figure 2 is a flow chart illustrating a computer-implemented electrical impedance imaging data processing method according to some embodiments of the present invention.
[0060] Figure 3 is a schematic diagram illustrating a computer-implemented electrical impedance imaging data processing method according to some embodiments of the present invention.
[0061] Figure 4 Schematic diagram showing the steps of predicting CAP using frequency difference electrical impedance tomography according to one embodiment of the present invention.
[0062] Figure 5 is a diagram showing an embodiment of the present invention, in which the total variance of the estimator is decomposed into the variance due to repetition and the variance due to measurement noise (including N R =7 reference area).
[0063] Figure 6 2 is a diagram showing CAP estimated by VCTE and CAP predicted by using a frequency difference electrical impedance tomography (blue) method and a frequency difference electrical impedance tomography (orange) method including spectral decomposition according to one embodiment of the present invention.
[0064] Figure 7 Graphs showing average CAP values of a healthy population (H) and a non-healthy population (NH) classified by CAP based on fibrosis scan (Fibroscan): (A) fibrosis scan values, (B) CAP predicted by electrical impedance imaging using general frequency difference electrical impedance imaging, (C) CAP predicted by electrical impedance imaging using a decomposed polynomial.
[0065] Figure 8 is a schematic diagram showing an example of steps for data collection.
[0066] Fig. 9 is a schematic diagram illustrating example operations of one embodiment of the present invention including electrical impedance imaging data acquisition, processing, and analysis.
[0067] Fig.10 Graphs showing in one example: (A) scatter plot of CAP and predicted CAP, (B) average CAP value of different NAFLD stages classified according to CAP predicted based on electrical impedance imaging, (C) average CAP value of different NAFLD stages classified according to true CAP, (D) receiver operating characteristic (ROC) curve of predicted CAP of healthy population relative to unhealthy population.
[0068] Fig.11 is a graph showing the performance of self-assessment scores of CAP on (A) liver fibrosis scan in one example, wherein (A1) is a scatter plot, (A2) the mean scores according to CAP classification of liver fibrosis scan, and (A3) the ROC curve of healthy versus non-healthy subjects.
[0069] Fig.12 Graphs showing in one example: (A) a CAP scatter plot including stimulation data, (B) average CAP values of different NAFLD stages classified according to CAP predicted based on electrical impedance imaging, (C) average CAP values of different NAFLD stages classified according to true CAP, and (D) receiver operating characteristic (ROC) curve of predicted CAP for a healthy population relative to a non-healthy population.
[0070] Fig.13A is a schematic diagram illustrating example operations of one embodiment of the present invention including electrical impedance imaging data acquisition, processing, and analysis.
[0071] Fig. 13B is a schematic diagram illustrating example electrical impedance imaging data processing according to one embodiment of the present invention.
[0072] Fig.14A is a graph showing the linear correlation coefficients and relative importance of various electrical impedance imaging features (conductivity features) and anthropometric features of a subject in one example.
[0073] Fig. 14B : is a graph (scatter plot) showing the relationship and classification specificity and sensitivity of the eGFR regression model of one embodiment of the present invention, associating true (blood test) eGFR values with eGFR scores determined based on electrical impedance imaging.
[0074] Fig. 14C is a graph showing the receiver operating characteristic (ROC) curve of the eGFR regression model and classification scheme according to one embodiment of the present invention.
[0075] Fig.14Dis a graph showing eGFR scores determined based on electrical impedance imaging for different stages (S1-S5) and severity of chronic kidney disease in one example.
[0076] Fig.15A : is a graph showing simulation results obtained based on the existing population eGFR distribution and model errors according to one embodiment of the present invention.
[0077] Fig. 15B is a graph (scatter plot) showing the simulated relationship and classification specificity and sensitivity of the eGFR regression model of one embodiment of the present invention, associating true (blood test) eGFR values with simulated electrical impedance imaging-based determined eGFR scores.
[0078] Fig. 15C is a graph showing simulated electrical impedance imaging-based eGFR scores for different stages (S1-S5) and severity of chronic kidney disease in one example.
[0079] Fig.16 is a block diagram of a data processing system arranged to perform (partially or completely) one or more method embodiments of some embodiments of the present invention.
[0080] Fig.17 is a block diagram of an example electrical impedance imaging system of some embodiments of the invention.
[0081] Fig.18 is a block diagram of an example electrical impedance imaging console according to some embodiments of the present invention.
[0082] Fig.19 is a block diagram of an example electrical impedance imaging console according to some embodiments of the present invention. DETAILED DESCRIPTION
[0083] Figure 1 A computer-implemented electrical impedance tomography (EIT) data processing method 100 of some embodiments of the present invention is shown. The method includes: in step 102, processing an electrical impedance tomography data set of a subject to determine a conductivity characteristic associated with a tissue / organ of the subject; and in step 104, determining a health state or condition of the tissue or organ of the subject based at least on the determined conductivity characteristic. In some embodiments, step 102 and step 104 are performed separately (sequentially). In some embodiments, step 102 and step 104 are performed at least partially simultaneously.
[0084] In some embodiments, the electrical impedance imaging data set includes electrical impedance imaging data acquired from a region (body part) of a subject containing an organ or tissue of interest. The region may be a chest region, an abdominal region, etc. In some embodiments, the electrical impedance imaging data set is acquired by: (a) providing an excitation signal (voltage, potential, current, etc.) of a frequency to the subject via electrodes attached to the region of the subject; (b) measuring a response signal (voltage, potential, current, etc.) received via the electrodes due to providing the excitation signal; and (c) repeating steps (a) and (b) for different frequencies. The electrical impedance imaging data set may include a plurality of electrical impedance imaging data subsets, each of which is associated with a corresponding one of the different frequencies.
[0085] In some implementations, step 102 may include one or more or all of the following:
[0086] (i) processing the electrical impedance imaging data set to obtain a processed electrical impedance imaging data set (which includes a plurality of processed electrical impedance imaging data subsets);
[0087] (ii) processing the processed electrical impedance imaging data set to obtain a frequency difference electrical impedance imaging data set (which includes a plurality of frequency difference electrical impedance imaging data subsets, which may correspond to a plurality of conductivity maps)
[0088] (iii) performing a source separation operation using the frequency difference electrical impedance imaging dataset and one or more reference frequency difference electrical impedance imaging datasets corresponding to one or more reference subjects to determine a component in the frequency difference electrical impedance imaging dataset associated with a tissue or organ of the subject and each component in the one or more reference frequency difference electrical impedance imaging datasets associated with a tissue or organ of the one or more reference subjects.
[0089] (iv) performing a conductivity feature extraction operation using the component associated with the tissue or organ of the subject in the frequency difference electrical impedance imaging dataset and optionally each component associated with the tissue or organ of one or more reference subjects in one or more reference frequency difference electrical impedance imaging datasets to at least determine one or more conductivity features of the subject
[0090] (v) performing a conductivity feature extraction operation using each of the processed EIM dataset or the frequency difference EIM dataset associated with the tissue or organ of the subject and optionally one or more processed EIM datasets or one or more reference frequency difference EIM datasets associated with the tissue or organ of one or more reference subjects to determine at least one or more other conductivity features
[0091] In some embodiments, step 102 (i) may include filtering and / or smoothing each electrical impedance imaging data subset.
[0092] In some embodiments, (i) in step 102 may additionally or alternatively include processing the electrical impedance imaging data set using a classifier model (e.g., a processing model based on machine learning) to determine the corresponding performance of each electrode when acquiring the electrical impedance imaging data (the performance is associated with the quality of the response signal or data acquired from the corresponding electrode), and prevent the response signal or data acquired via any one or more electrodes determined to be insufficiently performing from being included in the processed electrical impedance imaging data set. In some examples, a corresponding performance score is determined for each electrode, and the corresponding performance score is compared with the reference data to determine whether any electrode (and if so, which one) has insufficient performance when acquiring the data. In some examples, the classification model can determine the electrode that has insufficient performance when acquiring the data.
[0093] In some embodiments, (ii) in step 102 may include: for each of one or more of the processed electrical impedance imaging data subsets, determining the difference between the corresponding processed electrical impedance imaging data subset and the reference electrical impedance imaging data subset, thereby obtaining a plurality of frequency difference electrical impedance imaging data subsets, each of which is respectively associated with a corresponding difference between the corresponding processed electrical impedance imaging data subset and the reference electrical impedance imaging data subset. At least one processed electrical impedance imaging data subset may be respectively used as a reference electrical impedance imaging data subset.
[0094] In some embodiments, (iii) in step 102 may include performing a dimensionality reduction operation on the frequency difference electrical impedance imaging dataset and one or more reference frequency difference electrical impedance imaging datasets corresponding to one or more reference subjects to determine the components in the frequency difference electrical impedance imaging dataset associated with the tissue or organ of the subject and each corresponding component in the one or more reference frequency difference electrical impedance imaging datasets associated with the tissue or organ of one or more reference subjects.
[0095] In some embodiments, (iv) in step 102 may include using a component associated with the subject's tissue or organ in the frequency difference electrical impedance imaging data set to determine one or more (e.g., statistical) conductivity features associated with the subject's tissue or organ. For example, one or more (e.g., statistical) conductivity features associated with the subject's tissue or organ may include one or more of the following: the average value of the conductivity feature within a defined tissue or organ region within an area, the average value of the conductivity feature outside a defined tissue or organ region within an area, and the average value of the conductivity feature within an area. The area may be an area within a conductivity map. In some embodiments, other statistical conductivity features associated with the subject's tissue or organ (e.g., median, mode, standard deviation, etc.) may be used instead of the average value.
[0096] In some embodiments, (iv) in step 102 may include using a component associated with the tissue or organ of the subject in the frequency difference electrical impedance imaging data set and a corresponding component associated with the tissue or organ of one or more reference subjects in each of the one or more reference frequency difference electrical impedance imaging data sets to determine one or more (e.g., statistical) conductivity features associated with the tissue or organ of the group including the subject and the one or more reference subjects. In addition, the determination of the health state or condition of the tissue or organ of the subject may also be based on one or more conductivity features associated with the tissue or organ of the group. For example, one or more (e.g., statistical) conductivity features associated with the tissue or organ of the group may include one or more of the following: the average value of the conductivity features within a defined tissue or organ region within an area, the average value of the conductivity features outside a defined tissue or organ region within an area, and the average value of the conductivity features within and outside a defined tissue or organ region within an area. The area may be an area within a conductivity map (e.g., an average conductivity map determined from a conductivity map). In some embodiments, other statistical conductivity features associated with the tissue or organ (e.g., median, mode, standard deviation, etc.) may be used instead of the average value.
[0097] In some embodiments, step 104 includes determining whether the subject suffers from a disease associated with the tissue or organ based at least on the one or more determined conductivity characteristics. In some embodiments, step 104 also includes classifying the stage or severity of the disease associated with the tissue or organ.
[0098] In some embodiments, step 104 includes determining a value associated with an estimated property of a tissue or organ of the subject based on at least one or more determined conductivity characteristics. For example, with respect to a kidney, the value may be an estimated glomerular filtration rate or a related value (e.g., mathematically related). For example, with respect to a liver, the value may be a controlled attenuation parameter (CAP) score or a related value (e.g., mathematically related).
[0099] In some embodiments, step 104 includes processing at least one or more determined conductivity characteristics of the subject and one or more anthropometric characteristics of the subject using a machine learning based processing model to determine a quantitative or qualitative parameter associated with a health state or condition of a tissue or organ of the subject.
[0100] In some embodiments, step 104 includes: using a machine learning based processing model to process (i) one or more determined conductivity features of the subject, (ii) one or more anthropometric features of the subject, and (iii) one or more determined conductivity features of one or more reference subjects and / or one or more determined conductivity features of a group including the subject and one or more reference subjects to determine a quantitative or qualitative parameter associated with a health state or condition of a tissue or organ of the subject. Preferably, the one or more reference subjects do not suffer from any disease associated with the tissue or organ.
[0101] The processing model based on machine learning may include a regression model, a classification model, etc. The regression model may include a linear regression model or a nonlinear regression model.
[0102] The one or more anthropometric characteristics may include or be associated with one or more of, for example, the subject's age, the subject's weight, the subject's height, the subject's waist circumference, the subject's waist-to-height ratio, the subject's body mass index (BMI), the subject's gender, and the subject's race.
[0103] A quantitative or qualitative parameter associated with the health state or condition of a tissue or organ of a subject may include a value associated with an estimated performance of a tissue or organ of the subject. For example, with respect to a kidney, the value may be an estimated glomerular filtration rate or a related value (e.g., mathematically related). For example, with respect to a liver, the value may be a controlled attenuation parameter (CAP) score or a related value (e.g., mathematically related).
[0104] In some embodiments, step 104 includes: comparing a quantitative or qualitative parameter associated with the health state or condition of a tissue or organ of the subject with reference parameter data (one or more reference parameter values or ranges) to determine whether the subject suffers from a disease associated with the tissue or organ. In some embodiments, step 104 also includes: classifying the stage or severity of the disease associated with the tissue or organ based on the comparison.
[0105] Figure 2 A computer-implemented electrical impedance imaging data processing method 200 of some embodiments of the present invention is shown.
[0106] Method 200 includes performing initial processing (preprocessing) on an electrical impedance imaging dataset of a subject in step 202A to obtain a processed electrical impedance imaging dataset. The electrical impedance imaging dataset includes electrical impedance imaging data acquired from a region of the subject containing a tissue or organ of interest. The electrical impedance imaging dataset may be obtained using the above reference Figure 1 The initial processing may include the above reference Figure 1 The processing mentioned in method 100 with respect to step 102(i) is described.
[0107] The method 200 includes performing a frequency difference electrical impedance imaging (fd-EIT) processing operation on the processed electrical impedance imaging data set in step 204A to obtain a frequency difference electrical impedance imaging data set of the subject. The frequency difference electrical impedance imaging processing operation may include the above reference Figure 1 The processing mentioned in step 102 (ii) of the method 100 described above. The frequency difference electrical impedance imaging processing operation can be other frequency difference electrical impedance imaging methods known in the art for obtaining conductivity maps. The method 200 can also include performing initial processing (preprocessing) on the electrical impedance imaging dataset of a reference subject (who does not suffer from any disease associated with the tissue or organ of interest) in step 202B to obtain a processed electrical impedance imaging dataset. Each of the electrical impedance imaging datasets includes electrical impedance imaging data acquired from a region of the corresponding reference subject containing the tissue or organ of interest. The above reference can be used Figure 1 The method 100 described above is used to obtain an electrical impedance imaging data set. The initial processing may include the above reference Figure 1 The processing mentioned in method 100 with respect to step 102(i) is described.
[0108] The method 200 may further include performing a frequency difference electrical impedance imaging processing operation on the processed electrical impedance imaging data set in step 204B to obtain a frequency difference electrical impedance imaging data set of a reference subject. The frequency difference electrical impedance imaging processing operation may include the above reference subject. Figure 1The processing mentioned in step 102 (ii) in the described method 100. The frequency difference electrical impedance imaging processing operation can be other frequency difference electrical impedance imaging methods known in the art for obtaining conductivity maps. Method 200 includes using the frequency difference electrical impedance imaging dataset of the subject and the frequency difference electrical impedance imaging dataset of the reference subject to perform a group source separation operation in step 206. The main purpose of the group source separation operation is to separate or extract source signals (i.e., signals or data related to the tissue or organ of interest) from the frequency difference electrical impedance imaging dataset of the subject and the frequency difference electrical impedance imaging dataset of the reference subject, which may include signals of other organs or tissues. The result of step 206 is to obtain components in the frequency difference electrical impedance imaging dataset related to the tissue or organ of the reference subject and components in the frequency difference electrical impedance imaging dataset related to the tissue or organ of the subject. In one example, if the subject has one set of frequency difference electrical impedance imaging data sets and the reference subject has two sets of frequency difference electrical impedance imaging data sets, after the group source separation operation, one set of data (one graph) including components of the frequency difference electrical impedance imaging data sets associated with the tissue or organ of the subject and two sets of data (two graphs) including components of the frequency difference electrical impedance imaging data sets associated with the tissue or organ of the reference subject will be obtained. The group source separation operation may include the above reference Figure 1 The processing mentioned in step 102(iii) of method 100 is described.
[0109] Method 200 includes, in step 208, performing an EI feature extraction operation using a component in the frequency difference EI dataset associated with a tissue or organ of a reference subject and a component in the frequency difference EI dataset associated with a tissue or organ of a subject. Here, the EI feature corresponds to a conductivity feature or features. The result of step 206 is to obtain characteristics (conductivity features) of the subject and optionally of the reference subject. The EI feature extraction operation may include the above reference Figure 1 The processing mentioned in the method 100 described above for steps 102 (iv) and (v). In one example, if there is a set of data (one graph) including components of a frequency difference electrical impedance imaging dataset associated with a tissue or organ of a subject and two sets of data (two graphs) including components of a frequency difference electrical impedance imaging dataset associated with tissues or organs of two reference subjects, one or more of these graphs will be processed to determine one or more features associated with the tissue or organ. For example, a statistical operation can be performed on the entire graph including components of a frequency difference electrical impedance imaging dataset associated with a tissue or organ of a subject to obtain an average value, mean, median, etc. of the conductivity of the graph.
[0110] For example, a portion of a graph that includes only components of a frequency difference electrical impedance imaging dataset related to a tissue or organ of a subject may be subjected to statistical operations to obtain an average value, mean, median, etc. of the electrical conductivity of the portion of the graph. For example, statistical operations may be performed on all graphs that include components of a frequency difference electrical impedance imaging dataset related to tissues or organs of a subject and a reference subject to obtain an average value, mean, median, etc. of the electrical conductivity of a collection of graphs. For example, a delineation operation (e.g., thresholding) may be performed on a graph that includes components of a frequency difference electrical impedance imaging dataset related to a tissue or organ of a subject to define a region of interest and obtain an average value, mean, median, etc. of the electrical conductivity of the region of interest of the graph. These can all be obtained as electrical impedance imaging features. In some examples, the processed electrical impedance imaging dataset of the subject and / or the processed electrical impedance imaging dataset of the reference subject may also be used to perform electrical impedance imaging feature extraction operations.
[0111] Figure 3 A computer-implemented electrical impedance imaging data processing method 300 of some embodiments of the present invention is shown. The method 300 includes processing conductivity features associated with a subject and an optional reference subject and an anthropometric feature of the subject using a regression and / or classification model (e.g., a processing model based on machine learning) to determine a quantitative or qualitative parameter associated with a health state or condition of a tissue or organ of the subject. The electrical impedance imaging feature may be a reference subject or a combination of the electrical conductivity features and the anthropometric features of the subject. Figure 1 and Figure 2 Those of the methods 100, 200 described above. The anthropometric features may be Figure 1 Those of method 100 described.
[0112] Some example embodiments of the present invention are provided below. These embodiments can be considered as Figure 1 A specific embodiment of the method 100 is provided.
[0113] Example 1
[0114] Example 1 can be viewed as Figure 1 A specific embodiment of the method 100, wherein the tissue / organ of interest is the liver.
[0115] 1. Introduction
[0116] 1.1 Fatty Liver, Controlled Attenuation Parameters, and Purpose of Implementation
[0117] Non-alcoholic fatty liver disease (NAFLD), also known as hepatic steatosis, refers to the appearance of fat around liver cells (hepatocytes). NAFLD is often associated with a sedentary lifestyle. Currently, a variety of techniques are used to diagnose NAFLD. One of these techniques is a liver biopsy. Although useful, this technique is invasive, relatively expensive, prone to sampling errors, often painful, and can cause some serious complications. Non-invasive techniques such as ultrasound-based devices and vibration-controlled transient elastography are also used to diagnose NAFLD. These techniques measure the elasticity of the liver to infer hepatic steatosis in NAFLD, which is quantified using controlled attenuation parameters.
[0118] In this embodiment, frequency sweeping in electrical impedance tomography (ie, taking multiple measurements where the frequency of each measurement is different) is applied to predict controlled attenuation parameters from cross-sectional electrical impedance tomography measurements across the liver (using a general frequency difference and spectral decomposition model).
[0119] 1.2 Basic principles of electrical impedance tomography
[0120] Electrical impedance tomography (EIT) is a non-invasive imaging method based on measuring the electrical impedance (i.e. bioimpedance) of living tissue. Typically, a small electric current (e.g., about 1 mA or any other value that does not affect normal physiology) with a frequency range of typically 1 kHz to 1 MHz (other frequencies are also possible) is applied to the body through electrodes (e.g., on a belt). This current induces an electric potential that is measured at each electrode. Using this input, a conductivity map of the body can be reconstructed. These conductivity changes are used to predict the value of a controlled attenuation parameter. In the rest of this section, the electrical impedance tomography reconstruction problem will be briefly introduced.
[0121] 1) Positive question
[0122] Assume that a current is applied to the body (of the subject) via the source electrode and a response signal (e.g., potential) is received at the receiver electrode. If the conductivity in the body is known, the potential in the body can be calculated using Maxwell's equations
[0123]
[0124] where σ represents the conductivity in the bulk and V represents the electric potential. This equation can be modeled fairly accurately.
[0125] 2) Inverse Problem
[0126] The main challenge of electrical impedance tomography is that both the conductivity σ and the potential V in the body are unknown. Instead, they must measure the value V at the boundary. mes This example aims to recover only the conductivity change between two frequencies ( Figure 4). Using linear approximation, this reconstruction is easier than absolute imaging. The algorithm used in this embodiment is a linearized least squares algorithm that uses Kotre regularization and prior information extracted from computed tomography (CT). By using V mes =F(σ) to represent the forward operator that maps conductivity to the measured potential. The change in potential can be expressed as follows:
[0127]
[0128] Due to the ill-posedness of the problem, this relationship is not sufficient to recover σ from the boundary measurements. Combining the regularization term and the prior information, the final expression for the discretization of the conductivity is
[0129]
[0130] where R is the Kotre diagonal sensitivity matrix, α and p are two regularization parameters, and W x is a matrix that includes prior information.
[0131] 2. Materials and Methods
[0132] 2.1 Clinical evaluation
[0133] A total of 11 human subjects were tested, including 3 females and 8 males, whose ages ranged from 20 to 65 years old and waist circumferences ranged from 71 cm to 110 cm. Their respective clinical demographic and physical characteristics, including sex, BMI, age, waist circumference, height, weight, and history of liver disease (if any), were collected.
[0134] 2.2 Data Collection Process
[0135] First, each subject underwent a 5- to 10-minute liver fibrosis scan. Test (Echosens, France) to obtain the value of the controlled attenuation parameter (CAP). Then, the electrical impedance tomography examination is performed using a portable system including an electrical impedance tomography data acquisition console and a 16-electrode belt. The data acquisition console used in this example consists of a power management module, a current generator that provides an alternating current in the frequency range of 10KHz to 1MHz (other frequency ranges, such as the order of GHz, are also envisioned), a data acquisition module for measuring potential differences, and a control and output module for module coordination, data processing, and cloud server communication. The position of the electrode belt (and therefore the electrodes thereon) is targeted to the upper abdominal area, such as shown by the bottom boundary of the thorax.
[0136] 2.3 General frequency difference
[0137] Conductivity was reconstructed using a custom version of the pyEIT python library. The conductivity changes between two pairs of frequencies (28.75kHz to 20.00kHz) and (28.75kHz to 25.00kHz) were calculated by averaging the difference conductivity maps of regions of interest (ROI) covering the entire liver. Both the prior and the ROI were extracted from computed tomography images.
[0138] 2.4 Spectral decomposition method for conductivity
[0139] In this example, a method is provided to extract more information from the frequency difference curve. The method is based on the following principle: if the change of conductivity with respect to frequency is small, then the shape of the measured voltage change with frequency is given by a linear combination of the shape of the conductivity change with frequency. For example, the change of conductivity with frequency is approximated using the Taylor expansion:
[0140]
[0141] The measured potential is equal to V mes (ω)=F(σ(ω)), which is a discrete operator Approximately, where x N is a vector of points in the discretized domain. The potential difference can be approximated as follows:
[0142]
[0143] And the derivative can be expressed as
[0144]
[0145] It should be noted that (corresponding )and (corresponding ) is the same as the relationship between ΔV and Δσ. Therefore, the method used to reconstruct Δσ can be exactly the same as that used to reconstruct (corresponding ).
[0146] It is assumed that the body parts that contribute to the first and second order changes are different, so separating them will provide more specific information. It should be noted that the extension of this method example is up to second order. However, it can be extended to higher orders as long as the conductivity change has sufficient derivatives. It can also be extended to any function basis / free family decomposition. Assume that M tissue types contribute to the conductivity change and their respective conductivity changes are Δσ i (ω), then under the linear assumption (2), we can get the following results:
[0147]
[0148] where ΔV(ω) is the potential difference data set (excluding the reference set), a i is a parameter representing the effect of tissue type i, Δσ i (ω) is the predetermined spectrum specific to tissue type i, M is the total number of tissue types, and ∈ is the error term. Δσ i (∈) can be obtained from the tabulated values and the coefficients can be estimated by the general linear mixed model method (Table 1).
[0149] Table 1: Summary of graphs recovered with respect to a given input
[0150]
[0151] 2.5 Optimization of acquisition paradigm
[0152] Once the model for the controlled attenuation parameter is constructed, more EIT data can be acquired (without measuring the controlled attenuation parameter for VCT) to minimize the variance of the prediction. For each acquisition frequency, the duration and number of repetitions can be freely chosen. Assume that the data acquired at the jth repetition at frame i provide the following observations for the controlled attenuation parameter
[0153]
[0154] where CAP is the true value of the controlled attenuation parameter, ∈ F,i,j is the measurement error (i.e., the error due to sensor noise), and ∈ R,j is the repetition error (patients do not repeat exactly the same pattern). F To represent the number of frames acquired each time and use N R To estimate the controlled decay parameters, a Monte Carlo average is used for all frames. Assume ∈ F,i,j ,∈ R,j are independent random variables and the errors ∈ F,i,j (Correspondingly ∈ R,j ) have the same finite variance (Correspondingly ), then according to the central limit theorem, we can conclude that: when N R and N F When it is big enough, In this example, the duration of one acquisition is determined by the number of frames N. F The duration t of collecting one frame F The product of plus the setup time t s Therefore, the total duration of all acquisitions is T = N R (NF t F +t S By fixing the total time T = T0 and the above equation, the total variance can be decomposed into a part caused by measurement noise and a part caused by repetition ( Figure 5 ). The optimal theoretical value can be calculated, namely as well as Finally, the value is selected by rounding the obtained true value. For setting time t s =5s and the total acquisition time of each frequency T0 = 30s, the optimal number of repetitions is N R =3 and the acquisition time is N F t F =5 seconds.
[0155] 3. Results
[0156] 3.1 Linear regression including simple differences
[0157] Table 2: Results of linear regression of controlled attenuation parameters with respect to changes in conductivity and waist-to-height ratio (WoH)
[0158]
[0159]
[0160] In this example, the conductivity change dC is used 28.75kHz-22kHz ,dC 28.75kHz-25kHz The spatial mean of the 1000 sq.m and the anthropometric variable WoH (waist-to-height ratio) were used as features to predict the controlled attenuation parameters by linear regression. All variables were significant at the 0.05% level of the Student's t-test. The coefficient of determination was 0.932, and the adjusted coefficient was 0.919 (Table 2). The predicted controlled attenuation parameters seemed to be able to classify the state of the liver based on the controlled attenuation parameters of the liver fibrosis scan (sensitivity = 0.75, specificity = 0.71, the subjects who were misclassified were mostly close to the threshold, reference Figure 6 and Figure 7 ). Even with high values of the controlled decay parameter, the regression maintains good homoscedasticity.
[0161] As can be seen from Table 3, using both conductivity and waist-to-height ratio can greatly improve the quality of regression compared to using either conductivity or waist-to-height ratio alone. The coefficients of the linear regression using normalized data have the same magnitude, so it can be determined that they have the same importance in the prediction.
[0162] Table 3: Determination coefficients of different models
[0163]
[0164] 3.2 Linear regression involving factorized polynomial coefficients
[0165] As described above, the conductivity was reconstructed using the first and second order variations of V with respect to ω. For each patient or subject, the least squares method was used to verify To estimate α and β, which are then used to reconstruct the image. The average conductivity dC of the liver region was calculated α and dC β and used them as regressors. The polynomial models provide a good approximation for the potential change between 10kHz and 35kHz (MAPE of 4%). They perform similarly to the simple difference in predicting CAP, with an adjusted R-squared (including WoH) of 0.914 ( Figure 6 and Figure 7 ).
[0166] 4. Discussion
[0167] 4.1 Prediction of controlled attenuation parameters
[0168] The data of this example show that in addition to the already observed correlation with waist-to-height ratio (WoH), there is a strong correlation between the conductivity difference between the two frequency pairs and the controlled attenuation parameter measured using vibration-controlled transient elastography. This correlation can be explained by the conductivity changes with respect to fat content in liver tissue and is captured by electrical impedance imaging. A shape prior was used to focus on the liver region. In addition, the use of a self-administrable electrical impedance imaging device instead of vibration-controlled transient elastography allows for more affordable measurements to obtain real-time results without the need for the assistance of trained professionals to acquire the data.
[0169] 4.2 Spectral Decomposition Model
[0170] The spectral decomposition method in one embodiment provided similar results to conventional frequency difference electrical impedance tomography (fd-EIT) (adjusted R-squared of 0.914), which confirmed the validity of the method. Due to its reasonable assumptions, the method can be applied to other organs (e.g., kidney).
[0171] In summary, the present embodiment shows that by using anthropometric measurements of waist-to-height ratio (WoH) as supplementary information, multi-spectral electrical impedance imaging can predict clinical standard controlled attenuation parameters for patients with or without non-alcoholic fatty liver disease. The present embodiment also provides a spectral decomposition method that estimates controlled attenuation parameters from multi-spectral electrical impedance imaging by matching the coefficients of the function decomposition. The spectral decomposition method can be used to process other electrical impedance imaging data to diagnose different diseases (in addition to liver disease). The present embodiment also provides an algorithm for determining the optimal number of repetitions (N) that minimizes error and maximizes accuracy under time constraints. R =3) and acquisition time (N F t s =5s). In one example, these aspects are independent of each other (ie, implemented separately). In another example, two or more of these aspects are implemented simultaneously.
[0172] Example 2
[0173] Example 2 can be viewed as Figures 1 to 3 A specific embodiment of methods 100-300, wherein the tissue / organ of interest is the liver.
[0174] 1. Introduction
[0175] The present examples relate to electrical impedance imaging-based systems and methods for assessing the health state or condition of a subject's liver (eg, whether the subject has non-alcoholic fatty liver disease (NAFLD)).
[0176] 2. Materials and Methods
[0177] Figure 8 The steps for collecting data of this example are shown ("Port. Con." means portable console).
[0178] 2.1 Data Collection
[0179] The experiment in this example includes 64 human subjects. The subjects' ages ranged from 21 to 84 years old, waist circumferences ranged from 65 cm to 127 cm, and BMIs ranged from 18.9 to 43.3. Body measurement information was collected for the subjects, including gender, BMI, age, waist circumference, height, and medical history ( Figure 8 ). Then, a controlled attenuation parameter (CAP) was obtained using a liver fibrosis scan (Echosens, France), followed by three electrical impedance tomography examinations using a 16-electrode belt worn around the waist and a portable system. The frequency of the injected current ranged from 10KHz to 400KHz.
[0180] 2.2 Constant reference impedance tomography, frequency difference impedance tomography and group source separation
[0181] Fig. 9 The steps for generating features used to predict CAP in liver fibrosis scans in this example are shown ("Fd" stands for frequency difference).
[0182] First, the data is automatically filtered based on the quality of the electrical impedance imaging ( Fig. 9 ). A modified version of the pyEIT library was used to generate constant reference conductivity images and frequency difference electrical impedance imaging images. For constant reference electrical impedance imaging, the average value of the conductivity map of healthy subjects classified by CAP at a frequency of 50kHz was selected as a reference. The conductivity difference electrical impedance imaging data between 160kHz and 80kHz was selected as another feature. The prior and ROI within the entire liver were extracted and modified from the liver computed tomography (CT) image. In this example, group source separation utilizes every possible frequency difference data between 20kHz and 160kHz. The data set is divided into a training set / test set to generalize the separation model to decompose the source components.
[0183] 2.3 Regression Model
[0184] In this example, the CAP prediction (CAP prediction using electrical impedance imaging data) model is a linear regression model including the above 3 features, and the subject's waist-to-height ratio (WtHt), height, weight, age, and gender are divided into training / development / test parts. In this example, the model is built using the scikit-learn library in python.
[0185] 3. Results
[0186] 3.1 Performance of regression models for controlled attenuation parameter (CAP)
[0187] Fig.10 Scatter plots of (A) CAP and CAP predicted by electrical impedance imaging are shown. Average CAP values of different NAFLD stages classified by (B) CAP predicted by electrical impedance imaging and (C) CAP. (D) Receiver operating characteristic (ROC) curve of CAP predicted by electrical impedance imaging (healthy population vs. non-healthy population). ("Theo" means theoretical value; "Tr" means training, "Te" means test, "Theo." means theory, "H" means health, "Mi" means mild, "Mod" means moderate, "Severe" means severe, ***: p<0.001, ***: p<0.01).
[0188] Fig.10A scatter plot of all predicted CAP values (based on electrical impedance imaging) and true CAP values is shown. The sensitivity of the predicted CAP value (based on electrical impedance imaging) was 86.1% and the specificity was 71.4% (coefficient of determination = 0.25). One-way analysis of variance and post hoc Bonferroni test between the various stages of NAFLD showed that all NAFLD progression stages predicted by the two CAPs were statistically different from those of other groups (p < 0.001). The area under the curve (AUC) of the receiver operating characteristic (ROC) curve for the predicted CAP (based on electrical impedance imaging) reached 0.799.
[0189] 4. Discussion
[0190] 4.1 Model feature selection
[0191] The three features used in this example reveal more information about the state of the liver. In this example, for constant reference electrical impedance tomography, 50kHz was selected as the target frequency and reference constant reference electrical impedance tomography was used to modify the traditional time difference electrical impedance tomography. Since the average of all available healthy data classified by CAP was selected as the reference conductivity data, the unique characteristics of each subject can be captured in the features.
[0192] For group source separation, combining all frequency difference electrical impedance imaging combinations together can ideally maximize the signal difference between adipose and liver tissues, so the highlighted attributes can be separated to produce a signature corresponding to adipose tissue.
[0193] For frequency difference electrical impedance tomography, 160 kHz was selected as the frequency and 80 kHz as the reference frequency.
[0194] 4.2 Clinical significance
[0195] To demonstrate the clinical screening effect, the self-assessment scores disclosed by Y.-ho Lee et al. in PLoS ONE, Vol. 9, 2014, were compared with CAP predicted based on electrical impedance imaging.
[0196] Fig.11 Results are shown, namely the performance of self-assessment scores on (A) liver fibrosis scan CAP, with (1) a scatter plot, (2) the mean scores classified by liver fibrosis scan CAP, and (3) the ROC curve of healthy subjects versus non-healthy subjects.
[0197] like Fig.11As shown, the prediction of the self-assessment score is biased, with a sensitivity of 73.3% and a specificity of 36.8% compared to CAP. Unlike the predicted values of the method of this example, the assessment scores did not produce statistically different results between the healthy and unhealthy groups. The AUC value of the ROC classified by CAP was 0.654, which is lower than the predicted AUC value of 0.79.
[0198] In this example, to predict the performance of the model, additional data points were simulated based on the variance of the existing model to match the population distribution in the National Health and Nutrition Examination Survey (NHANES) database using the ADAPT library.
[0199] Fig.12 The relevant results are shown (A) CAP scatter plot including stimulation data. Average CAP values of different NAFLD stages classified according to (B) CAP predicted based on electrical impedance imaging and (C) true CAP. (D) ROC curve of predicted CAP in healthy and unhealthy populations ("Org" represents original data points, "Stim" represents stimulated data points, "Theo" represents theoretical data points, ***: p < 0.001)
[0200] like Fig.12 As shown, the predicted sensitivity and specificity remained at about 77% and the AUC of ROC remained at 0.80 (N=1264). The performance of the model of this example was maintained under a larger data set: This demonstrates the potential of the method and system of this example as a low-cost portable tool for home and community-based NAFLD screening and monitoring.
[0201] In summary, this example shows that electrical impedance imaging can predict CAP of liver fibrosis scans from clinical standard VCTE using anthropometric indices and conductivity features. This example also demonstrates a cost-effective and self-operable alternative for widespread diagnostic screening and monitoring of NAFLD in home and community settings.
[0202] Example 3
[0203] Example 3 can be viewed as Figures 1 to 3 A specific embodiment of methods 100-300, wherein the tissue / organ of interest is a kidney.
[0204] In this example: A data processing pipeline is provided to extract kidney signals from in vivo electrical impedance imaging data, and a regression model is provided to estimate the estimated glomerular filtration rate (eGFR) of patients with chronic kidney disease (CKD) using only electrical impedance imaging features and age. The results of this regression are used to classify the CKD stage of the patient.
[0205] 1. Materials and Methods
[0206] Fig.13A and Fig. 13B The exemplary operation of the present embodiment including electrical impedance imaging data acquisition, processing and analysis is shown. Specifically, Fig.13A An example of electrical impedance imaging data acquisition, data preprocessing, and data analysis flow is shown; Fig. 13B A detailed illustration of the eGFR prediction pipeline including group source separation is shown.
[0207] 1.1 Subjects
[0208] In this example, 21 healthy individuals (control group) and 54 patients with clinically diagnosed chronic kidney disease (CKD) were tested. Appropriate demographic and physical characteristics of the subjects were collected, including sex, age, weight, height, and waist circumference. In this example, the subjects underwent eGFR measurement and electrical impedance imaging renal assessment.
[0209] 1.2 Electrical impedance imaging acquisition
[0210] In this example, all EIT experiments were performed using a PVC electrode belt connected to a palm-sized portable EIT console via an HDMI cable. The EIT console was connected to a computer with specialized software for collecting, visualizing, and saving the collected raw EIT data and system information for further data processing and analysis. The electrode belt includes 16 modular electrode holders, each of which includes a printed circuit board that can be connected to 1 gel electrode. The electrode belt is placed circumferentially in the abdominal (upper abdomen) area. During the entire EIT data acquisition process, the subject was asked to breathe normally and remain motionless.
[0211] In total, each frequency had 208 (16 injection pairs multiplied by 13 measurement pairs per injection pair) stimulation-measurements. Each measurement was performed at a frame rate of 33fps. During the electrical impedance imaging measurements and analysis, 24 frequencies ranging from 28kHz to 300kHz were used.
[0212] To obtain a baseline truth eGFR, serum and urine samples were collected from each subject. The creatinine level (assessed using the serum sample) was then used to derive the subject's eGFR score using the known formula:
[0213] eGFR=141×min(SCr / k,1) α ×max(SCr / κ,1) -1.209 ×0.993 Age ×1.018[if female]×1.159[if black]
[0214] eGFR is the estimated glomerular filtration rate (mL / min / 1.73 m 2 ), SCr is standardized serum creatinine (mg / dL), κ = 0.7 (female) or κ = 0.9 (male), α = -0.329 (female) or α = -0.411 (male), min represents the minimum value of SCr / κ or 1, max represents the maximum value of SCr / κ or 1, and Age represents the age of the subject (in years).
[0215] CKD stages are classified based on eGFR values extracted from serum samples, with the following criteria: Stage 1 CKD: (eGFR>90); Stage 2 CKD: (eGFR: 60-89); Stage 3 CKD: (eGFR: 30-60); Stage 4 CKD: (eGFR: 15-30); Stage 5 CKD: (eGFR: <15). CKD stages are also grouped by severity as follows: normal to mild (stages 1 to 2), moderate (stage 3), and severe (stages 4 to 5).
[0216] 1.3 Data Preprocessing Electrical impedance imaging reconstruction was performed in Python using the pyEIT library and a custom function. In this example, frequency difference electrical impedance imaging was used, where the reference frequency was 30 kHz and the remaining 23 frequencies ranged from 28 kHz to 300 kHz for frequency difference reconstruction.
[0217] To ensure the quality of the data for further analysis, a measurement quality classifier and a reconstruction algorithm for arbitrary stimulus-measurement patterns were developed. For each frame, the classifier filters the data to remove undesirable effects due to uncontrollable events, such as measurement errors due to subject movement.
[0218] For each subject and each frequency, all frames were averaged, and all electrodes that were damaged in any frame were marked and the corresponding stimulation-measurement was removed after averaging the frames. Since a reconstruction algorithm was developed for arbitrary stimulation-measurement patterns, the conductivity image could be reconstructed after removing the electrodes involved in the erroneous measurement.
[0219] 1.4 Data Analysis
[0220] Using group-source separation ( Fig. 13B) to separate the signals from different internal tissues from each individual's reconstructed images. Group source separation was performed using frequency difference conductivity images of all contrast frequencies. Data from healthy subjects were used as a reference group to assist in separating the signals because the electrical responses of healthy subjects were more consistent compared to including all subjects from different CKD stages (including subjects who had undergone nephrotomy).
[0221] After isolating the signal source, the source of the kidney signal is determined from the group separation results. Extract individual kidney sources from the group results. After group source separation, the signal from the kidney is the strongest among all other signals in the extracted kidney image components. Then, the region of interest (ROI), i.e. the kidney, is extracted from the individual source. After extracting the kidney signal and the region of interest, electrical impedance imaging related features are generated, which include but are not limited to the average conductivity in the ROI, the average conductivity outside the ROI and their ratio. The data are divided into a training set and a test set in a stratified manner, which have 60 data points and 15 data points respectively. Utilize 5-fold cross validation to evaluate the linear regression model trained with the training set.
[0222] Statistical comparisons of CKD stage and severity categories were performed using one-way ANOVA followed by post hoc Bonferroni test (*p<0.05, **p<0.01, ***p<0.001).
[0223] Since I have already referred to Figure 2 The principles and operation of group source separation are described and therefore will not be repeated here.
[0224] 2. Results
[0225] Fig.14A Linear correlation coefficients and the relative importance of electrical impedance imaging features in predicting eGFR are shown. Fig. 14B The relevance and classification specificity and sensitivity of the eGFR regression model are shown. Fig. 14C ROC curves for the regression model and classification scheme are shown. Fig.14D Comparison of electrical impedance imaging-predicted eGFR between different CKD stages and CKD severity is shown.
[0226] 2.1 Relevance
[0227] Results showed that eGFR was correlated with the mean conductivity of the group kidney source, individual kidney source, and kidney ROI, with linear correlation coefficients of -0.4, 0.59, and -0.4, respectively. Age was also found to be linearly correlated with eGFR with a coefficient of -0.68, in addition to electrical impedance imaging features. Results showed that the mean conductivity within the kidney ROI was negatively linearly correlated with the mean conductivity of the individual kidney signal images with a coefficient of -0.79. See more Fig.14A It should be understood that Fig.14A In the above, “mean conductivity in ROI” refers to the mean value of only ROI in the conductivity map of the subject after group source separation (i.e. Fig. 13B ), “Average conductivity in extracted signal” refers to the average of the entire conductivity map of the subject after group source separation, and “Average conductivity of group signal” refers to the average of all conductivity maps of the subject and the reference subject after group source separation.
[0228] 2.2 Regression Model
[0229] In this example, the electrical impedance imaging features are fitted with age using the Lasso algorithm. Among all the features, age has the highest relative importance (0.45), while the average conductivity in the individual kidney source, group source, and ROI is 0.40, 0.10, and 0.05, respectively ( Fig.14A ). The cross-validation R 2 The score was 0.4, which means that the electrical impedance imaging feature prediction model had a significant correlation with the eGFR derived from serum and urine tests.
[0230] 2.3 Classification of chronic kidney disease (CKD)
[0231] CKD stage was obtained using the eGFR predicted by the regression model according to the following criteria: stage 1 CKD: (eGFR>90); stage 2 CKD: (eGFR: 60-89); stage 3 CKD: (eGFR: 30-60); stage 4 CKD: (eGFR: 15-30); stage 5 CKD: (eGFR: <15). Considering stage 1 CKD as healthy and stages 3, 4, and 5 CKD as unhealthy, a specificity of >99.9% and a sensitivity of 87.5% were obtained ( Fig. 14B ). The area under the curve (AUC) of the receiver operating characteristic (ROC) curve was obtained by setting the healthy baseline truth threshold eGFR to 60 ml / min while varying the classification threshold for classification according to the predicted eGFR. The AUC obtained was 0.89 ( Fig. 14C ).
[0232] 3. Discussion
[0233] This example demonstrates an eGFR estimation model and a CKD stage classification scheme using a portable, self-administered electrical impedance imaging device. The operation of the device does not require specialized professional training personnel and a clinical setting. In addition, the imaging device is non-invasive, ionizing radiation-free, and cost-effective. The device can be used for medical screening, early chronic kidney disease diagnosis, and longitudinal renal function monitoring without utilizing public health services. Therefore, the device can improve the quality of telemedicine and expand its application areas to renal function monitoring and chronic kidney disease. In addition, due to its portability and user-friendliness, the device can provide community-based CKD screening for individuals in areas that may be missed by the public healthcare system.
[0234] In this example, it was found that the average conductivity in the individual extracted signals had a very negative linear correlation with the average conductivity in the ROI, while its relative importance in the Lasso model was 0.4 and the linear correlation coefficient with eGFR was 0.59. This indicates that the average conductivity of the individual extracted signals is dominated by signals related to kidney function.
[0235] The evaluation of classification specificity and sensitivity was based on 21 healthy subjects. Simulations were generated based on the prediction error of the existing Lasso model and the population eGFR distribution. Data from NHANES were used as the overall (population) eGFR distribution.
[0236] FIG. 15A to FIG. 15C The simulation results based on the existing population eGFR distribution and the error of the model proposed in this example are shown. Specifically, Fig.15A The ROC curve of this simulation is shown, Fig. 15B The simulated correlation and classification specificity and sensitivity are shown. Fig. 15C Shown is the simulated comparison of electrical impedance tomography-based eGFR estimates between different CKD stages and CKD severity.
[0237] The ROC curve with AUC = 0.82 was obtained ( Fig.15A Using N = 1000 simulated data points, the sensitivity and specificity obtained were 88.67% and 93.24%, respectively ( Fig. 15B ).
[0238] In summary, in this example, a portable electrical impedance tomography device was used to obtain clinical data and electrical impedance imaging data of 54 CKD patients and 21 healthy subjects. This example provides a data processing pipeline in which a group source separation algorithm is used to separate the renal signal from the raw electrical impedance imaging data. A significant correlation was found between the standard eGFR and the eGFR predicted by a linear model using electrical impedance imaging features and age. The proposed model was used to classify the CKD stage from the estimated eGFR and obtained a sensitivity of 87.5% and a specificity of >99.9%. This example of renal function assessment demonstrates the feasibility of electrical impedance imaging in the field of telemedicine as a non-invasive method for early CKD diagnosis and the potential for longitudinal CKD monitoring.
[0239] Example System
[0240] Fig.16An example data processing system 1600 of some embodiments of the present invention is shown, which can be used to perform one or more method embodiments (partially or completely). The data processing system 1600 may be included in a server, a mobile device, a computer, etc. The data processing system 1600 generally includes appropriate components required to receive, store and execute appropriate computer instructions, commands and / or codes. The main components of the data processing system 1600 are a processor 1602 and a memory 1604. The processor 1602 may include one or more of the following: one or more central processing units (CPUs), one or more single-chip microcomputers (MCUs), one or more graphics processing units (GPUs), one or more logic circuits, one or more Raspberry Pi chips, one or more digital signal processors (DSPs), one or more application-specific integrated circuits (ASICs), one or more field programmable gate arrays (FPGAs), and / or any other digital or analog circuits configured to interpret and / or execute program instructions and / or process signals and / or information and / or data. The processor 1602 can be used to perform machine learning-based processing and non-machine learning-based processing. Memory 1604 may include one or more volatile memories (such as random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), etc.), one or more non-volatile memories (such as read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferroelectric random access memory (FRAM), magnetic random access memory (MRAM), flash memory (FLASH), solid-state drive (SSD), flash memory (NAND), non-volatile dual in-line memory module (NVDIMM), etc.), or any combination thereof. Appropriate computer instructions, commands, codes, information and / or data (e.g., one or more of the following: electrical impedance imaging data, processed electrical impedance imaging data, frequency difference electrical impedance imaging data, components of frequency difference electrical impedance imaging data associated with specific organs / tissues, one or more machine learning-based models, one or more anthropometric characteristics of the subject, one or more conductivity characteristics associated with the subject, and quantitative or qualitative parameters associated with the health state or condition of the corresponding one or more organs / tissues of the subject) can be stored in the memory 1604. Computer instructions for executing or facilitating the execution of the method embodiments of the present invention can be stored in the memory 1604. The processor 1602 and the memory 1604 can be integrated together or separated (and operably connected). Optionally, the data processing system 1600 also includes one or more input devices 1606. Examples of the input device 1606 include: a keyboard, a mouse, a stylus, an image scanner, a microphone, a tactile / touch input device (e.g., a touch screen), an image / video input device (e.g., a camera), etc.Optionally, the data processing system 1600 also includes one or more output devices 1608. Examples of the output device 1608 include: a display (e.g., a monitor, a screen, a projector, etc.), a speaker, a headset, an earphone, a printer, an additive manufacturing machine (e.g., a 3D printer), etc. The display may include an LCD display, an LED / OLED display, or other suitable display (which may or may not be touch-sensitive). The data processing system 1600 may also include one or more disk drives 1612, which may include one or more of the following: a solid-state drive, a hard disk drive, an optical drive, a flash drive, a tape drive, etc. A suitable operating system may be installed in the data processing system 1600 (e.g., in the disk drive 1612 or in the memory 1604). The memory 1604 and the disk drive 1612 may be operated by the processor 1602. Optionally, the data processing system 1600 also includes a communication device 1610, which is used to establish one or more communication links (not shown) with one or more other computing devices (e.g., a server, a personal computer, a terminal, a tablet computer, a mobile phone, a watch, an IoT device, or other wireless computing devices). The communication device 1610 may include one or more of the following: a modem, a network interface card (NIC), an integrated network interface, an NFC transceiver, a ZigBee transceiver, a Wi-Fi transceiver, etc. Transceiver, radio frequency transceiver, cellular (2G, 3G, 4G, 5G, 5G and above, etc.) transceiver, optical port, infrared port, USB connection or other wired or wireless communication interface. The transceiver can be implemented by one or more devices (integrated transmitter and receiver, separate transmitter and receiver, etc.). The communication link can be wired or wireless, which is used to transmit commands, instructions, information and / or data. In one example, the processor 1602, memory 1604 (optionally, input device 1606, output device 1608, communication device 1610 and disk drive 1612 (if present)) are directly or indirectly connected to each other via a bus, peripheral component interconnect (PCI) (such as peripheral component interconnect high speed), universal serial bus (USB), optical bus or other similar bus structure. In one embodiment, at least some of these components can be connected wirelessly, for example, via a network (e.g., the Internet or a cloud computing network), and those skilled in the art will understand that Fig.16 The data processing system 1600 shown is merely an example and in other implementations data processing system 1600 may have a different configuration (eg, include additional components, have reduced components). Fig.17An electrical impedance imaging system 1700 of some embodiments of the present invention is shown, which can be used to collect and process electrical impedance imaging data. The electrical impedance imaging system 1700 generally includes an electrical impedance imaging console 1702 (which may or may not be portable), an electrode E electrically connected to the electrical impedance imaging console 1702 to obtain a signal from a subject, and a data processing system 1704 operably connected to the electrical impedance imaging console 1702. In one example, the electrical impedance imaging console 1702 is mainly used to control the sending of signals to the subject's body and the receiving of signals from the subject's body via the electrode E. In one example, the data processing system 1704 is mainly used to process the signal obtained by the electrical impedance imaging console 1702 via the electrode E. In one example, the data processing system 1704 can be at least partially integrated with the electrical impedance imaging console so that data processing can be performed at least partially at the electrical impedance imaging console. In one example, the data processing system 1704 can be Fig.16 system 1600.
[0241] Fig.18 An example electrical impedance imaging console 1800 of some embodiments of the present invention is shown, which can be used for electrical impedance imaging data acquisition. In this example, the electrical impedance imaging console is portable. Fig.18 The electrical impedance imaging console 1800 generally includes a power management module for supplying continuous power, a current generation module for generating alternating current, a signal distribution and readout module for injecting current and reading voltage, a data acquisition module for measuring, amplifying and collecting potential differences, and a control and output module for coordinating modules, processing data and communicating with cloud servers. Fig.18 As shown, the electrical impedance imaging console 1800 in this example includes an electrical impedance imaging console having five main modules:
[0242] A power management module, which provides power to all other modules through a power outlet or batteries.
[0243] A current generation module, which mainly includes a sine wave generator and a constant current generator in sequence to generate an AC current of 1mApp and a voltage amplitude of 1Vpp. A low-pass filter may be included to suppress total harmonic distortion and environmental electromagnetic interference (eg, power line noise).
[0244] Signal distribution and readout module, which uses a set of CMOS multiplexers (MUX) to introduce the generated current into the subject via the 16-electrode strip. Four MUXs are used, two of which are used to inject current and the other two are used to read out voltage. The MUXs are configured in adjacent-scan pattern by a single-chip microcomputer (MCU).
[0245] The data acquisition module, namely the analog front end (AFE), is used to acquire, measure and amplify the potential difference from the electrodes. The AFE includes a four-stage wide-input differential amplifier and a bandpass filter with a high common-mode rejection ratio (CMRR).
[0246] · Control and output module, which includes analog-to-digital converter (ADC), MCU and wireless communication chip. The potential difference obtained from the data acquisition module is digitized by ADC, processed in the MCU unit, and transmitted to external devices (server, mobile phone, computer, etc.) for image reconstruction and processing.
[0247] Fig.19 is a block diagram of another example electrical impedance imaging console 1900 of some embodiments of the invention. Electrical impedance imaging console 1900 is essentially a generalized version of electrical impedance imaging console 1800. Electrical impedance imaging console 1900 may or may not be portable.
[0248] Although not required, the embodiments described with reference to the accompanying drawings may be implemented as an application programming interface (API) or a series of libraries for use by developers, or may be included in another software application (e.g., in a terminal or computer operating system or a portable computing device operating system). Generally, since program modules include routines, programs, objects, components, and data files that assist in performing specific functions, those skilled in the art will understand that the functionality of a software application may be distributed across multiple routines, objects, and / or components to achieve the same desired functionality described herein.
[0249] It should also be understood that when the methods and systems of the present invention are implemented entirely or in part by a computing system, any suitable computing system architecture may be used. This would include stand-alone computers, network computers, dedicated or non-dedicated hardware devices. When the terms "computing system" and "computing device" are used, these terms are intended to include (but not limited to) any suitable computer or information processing hardware arrangement capable of implementing the described functions.
[0250] Those skilled in the art will appreciate that the embodiments described and / or shown in the present invention may be changed and / or modified to provide other embodiments of the present invention. Therefore, the embodiments described and / or shown in the present invention should be considered illustrative and non-restrictive in all aspects. Example optional features of the present invention are provided in the Summary of the Invention and the Detailed Description. Some embodiments of the present invention may include one or more of these optional features (some of which are not specifically shown in the figures). Some embodiments of the present invention may not include one or more of these optional features (some of which are not specifically shown in the figures). For example, the method embodiments of the present invention are not limited to humans and can be used for other animals. The method can be used to determine the health status or condition of different tissues or organs of different subjects. Electrical impedance imaging data acquisition does not need to be performed using a portable electrical impedance imaging device (such as the device shown), and electrical impedance imaging data can be obtained using other electrical impedance imaging devices. The data processing method of the present invention can be implemented in any suitable device (including one or more of the following: server, computer, phone, electrical impedance imaging console (portable or non-portable) etc.). For example, the method of the present invention can be used for the diagnosis of different diseases associated with different tissues or organs of interest.
Claims
1. A computer-implemented method comprising: An electrical impedance imaging dataset of a subject is processed to determine one or more conductivity characteristics associated with a tissue or organ of the subject, and a health state or condition of the tissue or organ of the subject is determined based at least on the one or more determined conductivity characteristics.
2. The computer-implemented method of claim 1 , wherein: The determination includes: Based at least on the one or more determined conductivity characteristics, determining whether the subject suffers from a disease associated with the tissue or organ, and optionally: further classifying the stage or severity of the disease associated with the tissue or organ.
3. The computer-implemented method of claim 1 , wherein: The determination includes: At least the one or more determined conductivity characteristics of the subject and one or more anthropometric characteristics of the subject are processed using a machine learning based processing model to determine a quantitative or qualitative parameter associated with a health state or condition of the tissue or organ of the subject.
4. The computer-implemented method of claim 1 , wherein: The determination includes: Use machine learning based processing models to process (i) said one or more determined conductivity characteristics of said subject, (ii) one or more anthropometric characteristics of the subject, and (iii) one or more determined conductivity characteristics of one or more reference subjects and / or one or more determined conductivity characteristics of a group comprising said subject and said one or more reference subjects, To determine a quantitative or qualitative parameter associated with the health state or condition of the tissue or organ of the subject.
5. The computer-implemented method of claim 3 or 4, wherein: The machine learning based processing model includes a regression model.
6. The computer-implemented method of claim 3 or 4, wherein: The machine learning based processing model includes a classification model.
7. A computer-implemented method according to any one of claims 3 to 6, wherein: The one or more anthropometric characteristics include or are associated with one or more of the following: the subject's age, the subject's weight, the subject's height, the subject's waist circumference, the subject's waist circumference to height ratio, the subject's body mass index, the subject's gender, and the subject's race.
8. A computer-implemented method according to any one of claims 3 to 7, wherein: The quantitative or qualitative parameter associated with the health state or condition of the tissue or organ of the subject comprises a value associated with an estimated performance of the tissue or organ of the subject.
9. The computer-implemented method of claim 8, wherein: The determination also includes: The quantitative or qualitative parameter associated with the health state or condition of the tissue or organ of the subject is compared to reference parameter data to determine whether the subject suffers from a disease associated with the tissue or organ.
10. The computer-implemented method of claim 9, wherein: The determination also includes: Based on the comparison, the stage or severity of a disease associated with the tissue or organ is classified.
11. A computer-implemented method according to any one of claims 1 to 10, wherein: The electrical impedance imaging dataset includes electrical impedance imaging data acquired from a region of the subject containing the tissue or organ; The electrical impedance imaging data set is obtained by: (a) providing an excitation signal of a frequency to the subject via electrodes attached to the area of the subject, (b) measuring a response signal received via the electrode as a result of providing the excitation signal, and (c) repeating steps (a) and (b) for multiple frequencies; and The electrical impedance imaging data set includes a plurality of electrical impedance imaging data subsets, each of which is associated with a corresponding one of the plurality of frequencies.
12. The computer-implemented method of claim 11, wherein: The processing includes: (i) processing the electrical impedance imaging data set to obtain a frequency difference electrical impedance imaging data set, wherein the frequency difference electrical impedance imaging data set includes a plurality of frequency difference electrical impedance imaging data subsets; (ii) performing a source separation operation using the frequency difference electrical impedance imaging dataset and one or more reference frequency difference electrical impedance imaging datasets corresponding to one or more reference subjects to determine the component associated with the tissue or organ in the frequency difference electrical impedance imaging dataset and each component associated with the tissue or organ in the one or more reference frequency difference electrical impedance imaging datasets; and (iii) performing a conductivity feature extraction operation using the component associated with the tissue or organ in the frequency difference electrical impedance imaging dataset and optionally each component associated with the tissue or organ in the one or more reference frequency difference electrical impedance imaging datasets to at least determine the one or more conductivity features of the subject.
13. The computer-implemented method of claim 12, wherein: The processing also includes: The electrical impedance imaging dataset is preprocessed prior to the processing in (i) such that the electrical impedance imaging dataset processed in (i) is a preprocessed electrical impedance imaging dataset.
14. The computer-implemented method of claim 13, wherein: The pre-processing of the electrical impedance imaging data set comprises: Each of the plurality of electrical impedance imaging data subsets is filtered and / or smoothed.
15. The computer-implemented method of claim 13 or 14, wherein: The pre-processing of the electrical impedance imaging data set comprises: processing the electrical impedance imaging data set using a classifier model to determine a respective performance of each of the plurality of electrodes, the performance being associated with a quality of a response signal or data acquired from the respective electrode; and Response signals or data acquired via any one or more of the plurality of electrodes determined to be underperforming are prevented from being included in the processed electrical impedance imaging data set.
16. A computer-implemented method according to any one of claims 12 to 15, wherein: The processing of the electrical impedance imaging dataset in (ii) comprises: For each of the corresponding one or more of the multiple processed electrical impedance imaging data subsets, the corresponding difference between the corresponding processed electrical impedance imaging data subset and the reference electrical impedance imaging data subset is determined, thereby obtaining the multiple frequency difference electrical impedance imaging data subsets, each of the frequency difference electrical impedance imaging data subsets being respectively associated with the corresponding difference between the corresponding processed electrical impedance imaging data subset and the reference electrical impedance imaging data subset.
17. The computer-implemented method of claim 16, wherein: The reference electrical impedance imaging data subset includes at least one of the plurality of processed electrical impedance imaging data subsets.
18. A computer-implemented method according to any one of claims 12 to 17, wherein: The group source separation operation includes: A dimensionality reduction operation is performed on the frequency difference electrical impedance imaging dataset and one or more reference frequency difference electrical impedance imaging datasets corresponding to one or more reference subjects.
19. A computer-implemented method according to any one of claims 12 to 18, wherein: Performing the conductivity feature extraction operation includes: using the component related to the tissue or organ in the frequency difference electrical impedance imaging data set to determine one or more conductivity features of the subject.
20. The computer-implemented method of any one of claims 12 to 19, wherein: Performing the conductivity feature extraction operation includes: determining one or more conductivity features of a group including the subject and the one or more reference subjects using the component associated with the tissue or organ in the frequency difference electrical impedance imaging dataset and the corresponding component associated with the tissue or organ in each of the one or more reference frequency difference electrical impedance imaging datasets; and Wherein, the determination of the health state or condition of the tissue or organ of the subject is further based on the one or more conductivity characteristics of the group.
21. A computer-implemented method according to any one of claims 1 to 20, wherein: The tissue or organ includes lung, kidney, liver or heart.
22. A system comprising: one or more processors; as well as A memory storing one or more programs configured to be executed by the one or more processors, the one or more programs comprising instructions for performing or facilitating the performance of a computer-implemented method according to any one of claims 1 to 21.
23. A non-transitory computer-readable storage medium storing one or more programs configured to be executed by one or more processors, the one or more programs comprising instructions for performing or facilitating the performance of a computer-implemented method according to any one of claims 1 to 21.
24. A computer program product comprising instructions which, when the computer program is executed by a computer, cause or facilitate the computer to perform the computer-implemented method according to any one of claims 1 to 21.
Citation Information
Cited By
Tumor identification method based on multi-modal information coding
CN122023988A
