Electrical impedance mammography

Abstract

An example relates to a method that includes virtually purifying an experimental electrical impedance signal relative to a target tissue using a non-target tissue model. A method includes obtaining a purified impedance transfer function based on an electrical impedance model having one or more non-target tissue components and not having a target tissue component; and purifying the experimental electrical impedance signal using a purified impedance transfer function to obtain a virtual purified signal for the target tissue component. In one embodiment, a method includes accessing an impedance transfer function dependent on non-tumor tissue based on a measure of age and / or obesity; the experimental electrical impedance signal is modified using the obtained impedance transfer function to obtain a signal that depends less on non-tumor tissue and more on tumor tissue impedance.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to electrical impedance spectroscopy, electrical impedance imaging, and electrical impedance mammography.

[0002] Embodiments of the present invention relate to apparatus, computer programs and methods for electrical impedance imaging. Specifically, they relate to apparatus, computer programs and methods for electrical impedance imaging of a female human breast to facilitate detection of changes within a breast mass, including changes in the breast such as tumors, abnormalities and malignant changes such as cancer.

[0003] background

[0004] Electrical impedance mammography (EIM) or electrical impedance imaging (EII), also known as electrical impedance tomography (EIT), electrical impedance scanner (EIS) and applied potential tomography (APT), is an imaging technique used particularly in medical applications.

[0005] The technique images the spatial distribution of electrical impedance inside an object, such as the human body. The technique is attractive as a medical diagnostic tool because it is non-invasive and does not use ionizing radiation as in X-ray tomography or generate a strong, highly uniform magnetic field as in magnetic resonance imaging (MRI).

[0006] Typically, a two-dimensional (2D) or three-dimensional (3D) array of evenly spaced electrodes is attached to the object to be imaged around a region of interest (ROI). Either an input voltage is applied across pairs of "input" electrodes and an output current is measured at the "output" electrodes, or an input current is applied between pairs of "input" electrodes and an output voltage is measured between the "output" electrodes or between pairs of output electrodes. For example, when a very small alternating current is applied between a pair of "input" electrodes, the potential difference between all other pairs of "output" electrodes is measured. A current is then applied between a different pair of "input" electrodes, and the potential difference between all other pairs of "output" electrodes is measured. An image is constructed using appropriate image reconstruction techniques.

[0007] The spatial variations revealed in the electrical impedance images may be caused by changes in impedance between healthy and non-healthy tissue, changes in impedance between different tissues and organs, or changes in apparent impedance due to anisotropic effects caused by, for example, muscle arrangement.

[0008] Tissue changes or cellular changes associated with cancer cause significant local changes in electrical impedance and can be imaged. WO 00 / 12005 discloses an example of an electrical impedance imaging device that can be used to detect breast cancer or other cancers.

[0009] Changes in individual impedance characteristics can be used to analyze the structure of the object. For example, in the case of human tissue, changes in individual impedance characteristics can indicate the presence of abnormalities, because this produces electrical characteristics at the macroscopic level (millimeter range) or at the microscopic level (micrometer range) that are different from those exhibited by normal healthy tissue, showing different special cell groups, which will be associated with early abnormal changes, especially abnormal changes in malignant tumors such as cancer.

[0010] However, the amount of change in individual impedance characteristics at the macroscopic level or microscopic level may not be sufficient to enable accurate analysis of tissue / cell structure or abnormalities through in vitro or in vivo based EIM techniques. For example, the amount of change in cell membrane capacitance (C) or relaxation frequency (fr) may not be sufficient to be easily detected, for example, in an image of the object constructed based on those individual impedance characteristics.

[0011] Therefore, it would be advantageous to virtually purify the signal prior to processing and be able to identify abnormal changes including both benign and malignant lesions.

[0012] Therefore, it would be advantageous to ensure that the signal is determined primarily by changes in impedance of the target tissue rather than other tissues to allow identification of benign and malignant changes.

[0013] Brief Overview

[0014] According to various, but not necessarily all, embodiments, examples are provided as claimed in the appended claims.

[0015] This is the first time that a histopathology-guided bioimpedance model has been used to purify and identify abnormal cell-based tissues on 2D and 3D EIM imaging by using an in vitro database after "tissue purification".

[0016] Brief Description

[0017] Some examples will now be described with reference to the accompanying drawings, in which:

[0018] Figure 1 An example of the subject matter described herein is shown;

[0019] Figure 2 An example of the subject matter described herein is shown;

[0020] FIG3 illustrates an example of the subject matter described herein;

[0021] Figure 4 An example of the subject matter described herein is shown;

[0022] Figure 5 An example of the subject matter described herein is shown;

[0023] Figure 6An example of the subject matter described herein is shown;

[0024] Detailed Description

[0025] Figure 1 is a schematic diagram of an electrical impedance tomography apparatus;

[0026] Figure 2 A graph showing the measured electrical impedance as a function of frequency with a single dispersion or with multiple dispersions;

[0027] Figure 3A and Figure 3B An exemplary electrical impedance circuit model of a subject is shown; and

[0028] Figure 4 A flow chart is shown.

[0029] Figure 5 shows a controller, and Figure 6 A computer program is shown.

[0030] Figure 1 An electrical impedance measurement or electrical impedance tomography (EIT) device 10 for measuring impedance data of a load 12 is schematically illustrated. The load 12 includes a conductive object attached to a plurality of electrodes. The term "conductive" means that the object is capable of conducting an electric current, but it does not necessarily need to conduct the electric current very well. The device 10 also includes a signal source 14, a signal detector 16, and a computer 18. In one embodiment, the signal source provides a current as an input signal, and the signal detector detects a voltage as an output signal. In another embodiment, the signal source provides a voltage as an input signal, and the signal detector detects a current as an output signal.

[0031] A computer usually comprises at least a processor and a memory. The memory stores a computer program which, when loaded into the processor, controls the computer.

[0032] An input signal is applied to the object via the electrodes using source 14, and the resulting output signal present at the same electrode or other electrodes is measured using detector 16. The process is repeated for input signals of different frequencies. For example, the electrical signal can be applied by signal source 14 at many frequencies between 0 Hz (direct current) and 100 MHz to enable frequency-dependent electrical impedance data of the object to be obtained.

[0033] The separation of the electrodes used for impedance measurement determines the resolution or scale of the object being analyzed. Electrical impedance measurements can be obtained at an expected scale of interest (e.g., micrometer or millimeter range). As an example of a scale of interest, for a biological object, we may be interested in a single cell or group cell level or tissue or histological level, such as a lobule or duct in breast tissue. Subsequently, the acquired electrical impedance data will be analyzed using the transfer function of the assumed electrical model to determine multiple electrical impedance characteristics of the object. The electrical model used may depend on the resolution / scale of the impedance measurement.

[0034] refer to Figure 2 , the electrical impedance data obtained using the above method can be plotted as a function of frequency. The graph 22 represents the impedance change of the object with frequency or transfer function. The computer 18 is operable to execute an appropriate algorithm to analyze the obtained impedance transfer function or frequency-dependent impedance characteristics and thereby determine a plurality of electrical impedance characteristics of the object. The electrical impedance characteristics typically include one or more of the following:

[0035] a) Impedance at the limit ω->0 (lower limit)

[0036] b) Impedance at the limit ω->∞ (upper limit)

[0037] c) (i) Relaxation frequency when there is a change in the impedance properties of tissue structure, membrane or cells

[0038] (ii) Impedance at the frequency of change

[0039] (iii) The gradient of impedance change, particularly at relaxation frequencies that are related to the characteristics of the cell membrane and the diverse distribution of specialized cell membrane groups.

[0040] For example, if there are N dispersions including α, β and γ dispersions of biological materials in the frequency range used [Cole KS, Permeability and impermeability of cell membranes for ions, Cold Spring Harbor Symp. Quant. Biol. 8 pp110-22, 1940], where N>1, then the dispersion frequencies ω1, ω2, ...ω N-1 ,ω N is identified, and the electrical impedance characteristics of a particular dispersion m will typically include one or more of the following:

[0041] a) Impedance at the lower (global) limit ω->0 for m=1

[0042] For m > 1, in the lower (local) limit ω->ω m -a, where a<(ω m -ωm-1 ) and may be 1 / 2(ω m -ω m-1 )

[0043] b) Impedance at the upper (global) limit ω->∞ for m=N

[0044] For m﹤N, in the upper (local) limit ω->ω m +b, where b<(ω m+1 -ω m ) and may be b~1 / 2(ω m+1 -ω m )

[0045] c) (i) Relaxation frequency ω when there is a change in impedance m (f rm )

[0046] (ii) Impedance at the frequency of change

[0047] (iii) Gradient of change.

[0048] The amount of change in one or more of these impedance characteristics can be used to analyze the structure of the object due to intracellular / extracellular or intracellular / extracellular related changes in pathological characteristics at the macroscopic or microscopic level.

[0049] In some embodiments, the object analyzed is modeled using an equivalent electrical impedance circuit. The equivalent electrical impedance circuit 20 illustrated in Fig. 3 can be used to model the object. In a non-limiting example, the object that can be modeled using the equivalent electrical impedance circuit 20 can include human tissue or animal tissue and porous or other cells or cell-like materials in a cascade structure, such as each of Z1, Z2, and Z3 will contain similar 3 elements (such as Z1-1, Z1-2, Z1-3, etc., to be able to represent a single cell at a microscopic range, a small group of cells, or a large group of cells at a macroscopic or tissue level ...).

[0050] In the illustrated embodiment, the equivalent electrical impedance circuit 20 includes a cell portion 21 connected in parallel with an extracellular portion 23. The cell portion 21 has a capacitor C and a resistor R connected in series. i The resistance C is associated with the cell membrane / boundary, and the resistance R i The extracellular portion 23 has a resistance R e . Resistance R e Associated with structures outside the cell. Resistance R e With capacitor C and resistor R connected in series i Connect in parallel.

[0051] A non-limiting example of a single dispersive impedance transfer function for this circuit is:

[0052]

[0053] In the limit ω->0, Z->R e

[0054] In the limit ω->∞, Z->R i / / R e , that is, R i R e / (R i+ R e )

[0055] There is a variation (dispersion) at the frequency fr and at the impedance Zr with a gradient α.

[0056] The transport model of more than one dispersion in biological tissues can be modeled by the Cole-Cole equation (Cole K S1940, Cole K S1941, McAdams ET et al., 1995) as follows:

[0057] Z=R∝+(R0-R∝) / (1+(jf / fr))(1-α)

[0058] In general, if a three-component electrical equivalent circuit is used to simply model a cell suspension (Fricke and Morse, 1925) or tissue, the equation can be rewritten as follows:

[0059] Z=R e ·R i / (R e +R i )+(R e -R e ·R i / (R e +R i )) / (1+(jf / fr))(1-α)

[0060] Where R∝ is R e With R i The result of parallel connection.

[0061] At frequency f ri The sum has a gradient α i The impedance Z ri There are changes (dispersion) in the

[0062] As described above, the computer 18 is operable to execute an appropriate algorithm to analyze the measured impedance data and extract a plurality of electrical impedance characteristics of the object being analyzed. For example, based on the measured impedance data, the algorithm may be operable to plot the impedance data points as a function of frequency and use the model to generate a best fit line 22 to form Figure 2 From this transfer function, the computer 18 can determine a plurality of individual impedance characteristics of the object. These impedance characteristics may include:

[0063] a) The impedance in the limit ω->0, which gives R e

[0064] b) The impedance in the limit ω->∞, which gives R i R e / (R i +R e )

[0065] c) (i) Relaxation frequency f when there is a change in impedance r

[0066] (ii) The impedance Z of the transfer function at the frequency of change r

[0067] (iii) gives the gradient α of the change in relaxation factor.

[0068] The impedance characteristic may be used to determine further impedance characteristics using a model.

[0069] For example, if R e and R i R e / (R i +R e ) are both known, then we can determine R i .

[0070] At the changing (dispersion) frequency f r The impedance Z of the transfer function at r is the case where the capacitor dominates the transfer characteristic, since it conducts significantly better with every small increase in frequency, reducing the impedance. r The impedance Z r can be modeled as 1 / (j.2πf r .C). Therefore, C can be determined as 1 / (j.2πf r .Z r ).

[0071] Individual impedance characteristics (R e , R i 、f r , Z r Changes in impedance characteristics of an object can be used to analyze the structure of an object at a specific level at the macroscopic or microscopic scale. For example, in the case of human tissue, changes in individual impedance characteristics can indicate the presence of an abnormality because this produces electrical characteristics that are different from those exhibited by normal healthy tissue.

[0072] However, the amount of variation in individual impedance characteristics may not be sufficient to enable accurate analysis of the structure. For example, the cell membrane capacitance (C) or the relaxation frequency (f r ) may not be large enough to be easily detected, for example, in an image of the object constructed based on those individual impedance characteristics.

[0073] Figure 3A A hybrid impedance model for representing different realistic modes of the analyzed object including a mixed tissue of normal stoma, glandular or abnormal tissue growth is illustrated. In the illustrated embodiment, the equivalent electrical impedance circuit 30A comprises an inclusion portion 31 connected in parallel with a non-inclusion portion 33. The inclusion portion 31 has an impedance Z1 and an impedance Z3 connected in series. The non-inclusion portion 33 has an impedance Z3. Impedance Z3 is associated with a structure outside the inclusion. Impedance Z3 is connected in parallel with impedances Z1 and Z2 connected in series.

[0074] The impedance transfer function of the circuit 30A is:

[0075]

[0076] In this model, Z1 is the impedance of the target tissue, i.e., the tissue in the target direction for detection.

[0077] Figure 3B The hybrid impedance model of the object being analyzed is illustrated in the limit where the impedance of the target tissue is zero. In the illustrated embodiment, the equivalent electrical impedance circuit 30b includes an impedance Z3 in parallel with the impedance Z2.

[0078] The impedance transfer function for this circuit 30B is:

[0079]

[0080] or

[0081]

[0082] R is an optimization factor, and impedance Z2 is a reference impedance of a first non-target tissue, and impedance Z3 is a reference impedance of a second non-target tissue different from the first non-target tissue.

[0083] let

[0084] S is the experimental data (in the frequency domain);

[0085] Z A is a model transfer function including target tissue and non-target tissue, which is based on a circuit model including the electrical impedance of the target tissue and the electrical impedance of the non-target tissue (e.g., Z1, Z2, Z3).

[0086] Z Bis a model transfer function for non-target tissue, which is based on the same circuit model, but it now includes the electrical impedances of the non-target tissue, eg, Z2, Z3, and excludes the electrical impedance Z1 of the target tissue.

[0087] N is the noise signal, which is the signal generated by non-target tissue;

[0088] C is the expected signal (experimental signal S minus noise N).

[0089] S=Z A *s

[0090] s=Z A -1 *S[Equation 1]

[0091] N=Z B *s

[0092] Substituting into equation 1

[0093] N=Z B *Z A -1 *S

[0094] C=SN=SZ B *Z A -1 *S

[0095] Therefore, C can be derived from S and Z B , Z A to be determined by knowledge.

[0096] S(t) is the experimental data (in the time domain);

[0097] S is the experimental data (in the frequency domain);

[0098] Z A It is a model transfer function including target tissue and non-target tissue, which is based on a circuit model including the electrical impedance of the target tissue and the electrical impedance of the non-target tissue (e.g., Z1, Z2, Z3).

[0099] Z B is a model transfer function for non-target tissue, which is based on the same circuit model, but it now includes the electrical impedances of the non-target tissues, eg, Z2, Z3, and does not include the electrical impedance of the target tissue Z1.

[0100] N is a noise signal, which is a signal generated by non-target tissue in the frequency domain, and N(t) is an equivalent signal in the time domain.

[0101] C is the desired signal in the frequency domain (experimental signal S minus noise N), and C(t) is the equivalent signal in the time-space domain.

[0102] S=Z A *s

[0103] s=Z A -1 *S[Equation 1]

[0104] N=Z B *s

[0105] Substituting into equation 1

[0106] N=Z B *Z A -1*S

[0107] N(t)=T -1 {Z B *Z A -1 *T{S(t)}}

[0108] C(t)=S(t)–N(t)=S(t)-T -1 {Z B *Z A -1 *T{S(t)}}

[0109] where T{} is the forward transform from spacetime to frequency, and T -1 {} is the inverse transform from frequency to spacetime.

[0110] Therefore, C(t) can be obtained from S(t) and Z B , Z A to be determined by knowledge.

[0111] Figure 4 The illustration includes the use of non-target tissue models (e.g. Figure 3B ) to virtually purify an experimental electrical impedance signal relative to a target tissue. In some instances, the target tissue is a tumor tissue.

[0112] The illustrated methods include:

[0113] Based on an electrical impedance model ( Figure 3B ) to obtain the purified impedance transfer function (Z B );

[0114] Using the purified impedance transfer function (Z B ) purifies the experimental electrical impedance signals (S, S(t)) to obtain virtual purified signals (C, C(t)) for target tissue components.

[0115] The electrical impedance model ( Figure 3A ) to use the purified impedance transfer function (Z B ) purifies the experimental electrical impedance signal (S, S(t)) to obtain a virtual purified signal for a target tissue component, wherein the target tissue is a tumor tissue and the non-target tissue is a non-tumor tissue.

[0116] Full impedance model ( Figure 3A )'s transfer function (Z A ) can be compared with the reduced electrical impedance model ( Figure 3B )'s transfer function (Z B ) to obtain a virtual purification signal (C, C(t)) for the target tissue component.

[0117] Full impedance model ( Figure 3A )'s transfer function (Z A ) is based on the target tissue impedance Z1 and the non-target tissue impedances Z2 and Z3. Reduced electrical impedance model ( Figure 3B ) is based on non-target tissue impedances Z2 and Z3.

[0118] The non-target tissue impedances Z2, Z3 may be based on a reference impedance value for pure non-target tissue modified by an optimization factor (R).

[0119] The optimization factor (R) is related to one or more characteristics of the patient, such as age and / or obesity / (different BMI).

[0120] Assigning the appropriate optimization factor (R) to a subject is a classification problem that can be solved using machine learning.

[0121] In at least some instances, the method includes:

[0122] Based on age and / or obesity (or BMI) measurements, the impedance transfer function Z that depends on non-tumor tissue (e.g., depends on Z2, Z3, R) is accessed. B ;

[0123] The impedance transfer function Z is obtained using B The experimental electrical impedance signal is modified to obtain a signal that depends less on non-tumor tissue impedance (Z1 , Z2) and more on tumor tissue impedance (Z1 ).

[0124] One measure of obesity is body mass index.

[0125] Non-tumor tissue impedance can be based on connective tissue impedance and adipose tissue impedance.

[0126] Tumor tissue impedance can be of a benign nature, such as fibroadenoma tissue impedance, or malignant nature, such as ductal carcinoma tissue, or lobular carcinoma tissue or ductal carcinoma in situ (DCIS) impedance, depending on the target.

[0127] The experimental signal can be an electrical impedance signal from an in vitro measurement. This can be used to analyze in vitro samples.

[0128] The experimental signal can be an electrical impedance signal from an in vivo measurement. This can be used to analyze living subjects.

[0129] The electrical impedance signal can be derived from a 2D or 3D region of interest (ROI) on the EIM image with known tissue area at the ROI based on a tissue impedance database or the measured BMI of that particular patient from a volunteer.

[0130] The electrical impedance signal can be used for electrical impedance imaging.

[0131] In at least some instances, the electrical impedance model is a hybrid impedance model that includes target tissue impedance (Z1) and non-target tissue impedance (Z2, Z3).

[0132] The target tissue impedance Z1 and the first non-target tissue impedance Z3 are electrically connected in series as a combination, and the combination is electrically connected in parallel with the second non-target tissue impedance Z2 - see Figure 3A .

[0133] The target tissue impedance Z1 may be one of a fibroadenoma reference impedance or a ductal carcinoma reference impedance, depending on the desired target.

[0134] The first non-target tissue impedance Z3 is one of the connective tissue impedance and the fat tissue impedance, and the second non-target tissue impedance Z2 is the other of the connective tissue impedance and the fat tissue impedance.

[0135] The method can use an in vitro sample to determine a reference impedance (Z1) of a target tissue and / or use an in vitro sample to determine a reference impedance of a non-target tissue. It can use an in vitro sample to determine a first non-target tissue reference impedance Z3. It can use an in vitro sample to determine a second non-target tissue reference impedance Z2.

[0136] The electrical impedance model can be determined by best fitting in vitro data for one or more samples. For example, the relative ratio R between the first non-target tissue reference impedance Z3 and the second non-target tissue reference impedance for the electrical impedance model is determined by optimization (best fit), wherein the impedance (Z1) of the target tissue is known.

[0137] By using the electrical impedance model ( Figure 3A ) is set to zero, and the impedance model ( Figure 3A) to a non-target model for non-target tissues ( Figure 3B ).

[0138] Non-target tissue model ( Figure 3B ) is used to estimate the non-target components and remove the non-target components from the experimental electrical impedance signal to estimate the target signal. The removal occurs in the time domain or the frequency domain.

[0139] Non-target tissue model ( Figure 3B ) is used to estimate the non-target signal, and the non-target signal is subtracted from the experimental signal to estimate the target signal.

[0140] In vivo target signals can be analyzed to identify and localize spatial electrical impedance changes indicative of tumor tissue.

[0141] In vivo target signals can be analyzed to image tumor tissue, such as on 2D or 3D EIM images.

[0142] In some examples, computer program 406 includes computer code that, when executed by one or more processors, performs one or more of the described methods.

[0143] In some examples, an imaging system includes a controller or other device for performing one or more of the described methods.

[0144] Return to reference Figure 4 .

[0145] At blocks 42, 44, and 46:

[0146] An in vitro sample is used to determine an impedance parameter Z1 of a mixed model (Z1, Z2, Z3) including target tissue (Z1) and non-target tissue (Z2, Z3).

[0147] Z scores were obtained for connective tissue, adipose tissue, fibroadenoma, ductal carcinoma, lobular carcinoma, and DCIS.

[0148] The impedance of the target tissue (fibroadenoma or ductal carcinoma or lobular carcinoma or DCIS) is used as Z1 in the modeling process.

[0149] Z2 and Z3 are connective tissue and adipose tissue, and the relative ratios were determined by optimization (best fit).

[0150] At block 42:

[0151] EIS data were collected with the following tissue types:

[0152] Table 1. Tissue types collected by EIS

[0153] 1. Pure adipose tissue 2. Pure connective tissue 3. Pure invasive ductal carcinoma (IDC) 4. Purely benign, such as fibroadenoma 5. IDC mixed with adipose tissue and / or connective tissue 6. Fibroadenomas admixed with adipose tissue and / or connective tissue

[0154] At blocks 44, 46: Z1 is determined to be either IDC or fibroadenoma, and Z2 and Z3 are determined to be either fatty or connective.

[0155] At blocks 48, 50, 52: For a particular sample, determine the associated mixed model (Z1, Z2, Z3) by best fitting to the in vitro data of the sample (find Z2, Z3 for the model, Z1 is known), for example, find the optimized parameter R.

[0156] Images from pathological sections of the tissue are examined and used to verify the reliability of this "best fit" ratio R.

[0157] At step 54:

[0158] The optimized hybrid model (Z1, Z2, Z3, for example, Figure 3A ) is converted to a non-target model for non-target tissues (Z2, Z3, Z1->0, e.g. Figure 3B ).

[0159] At block 56:

[0160] Non-target tissue models (Z2, Z3; e.g. Figure 3B ) is used to estimate the non-target signal. The non-target signal is subtracted from the experimental signal to estimate the target signal (Z1).

[0161] The experimental signal may be an in vitro signal 43 or may be a region of interest (ROI) in vivo signal 55 .

[0162] The in vitro impedance data 43 is analyzed to identify the presence of malignancies such as IDCs. In this case, the experimental signal is associated with a specific sample. Spatial variations in the impedance of the target signal indicate IDCs in the sample.

[0163] Analysis of the in vivo data 55 is used to identify and locate spatial variations ΔZ indicative of cancer or other target tissue. In this case, the experimental signal is the in vivo signal 55. Spatial variations in the impedance of the target signal are indicative of cancer in vivo.

[0164] Method blocks 40 through 54 are repeated to obtain appropriate non-target tissue models for different combinations of BMI and age.

[0165] For a particular sample and for a subject in vivo, the composition of the tissue will be similar.

[0166] The ratio R between Z2 and Z3 can be statistically classified by different age groups or BMI groups. This means that since Z1 is the reference impedance, it does not change in any classification, however, when we classify Z2 and Z3, for example by different age and / or BMI classifications, the values ​​of Z2 and Z3 have different ranges identified by R.

[0167] By searching by age / BMI or other classification parameters, non-target tissue models ( Figure 3B ) are suitable for use with experimental in vivo signals.

[0168] The non-target tissue model is then used at block 56 to obtain a better data signal 57 for subsequent analysis, such as imaging.

[0169] Figure 5 An example of a controller 400 is shown. The controller 400 may be implemented as a controller circuit. The controller 400 may be implemented solely in hardware, have certain aspects in software including firmware alone, or may be a combination of hardware and software (including firmware).

[0170] like Figure 5 As illustrated in , the controller 400 may be implemented using instructions that enable hardware functionality, for example, by using executable instructions of a computer program 406 in a general-purpose processor or a special-purpose processor 402, which may be stored on a computer-readable storage medium (disk, memory, etc.) for execution by such a processor 402.

[0171] The processor 402 is configured to read from and write to the memory 404. The processor 402 may also include an output interface via which data and / or commands are output by the processor 402 and an input interface via which data and / or commands are input to the processor 402.

[0172] The memory 404 stores a computer program 406 including computer program instructions (computer program code) that control the operation of the device 10 when loaded into the processor 402. The computer program instructions of the computer program 406 provide logic and routines that enable the device to perform the methods illustrated in the figure. The processor 402 can load and execute the computer program 406 by reading the memory 404.

[0173] Thus, the device 10 comprises:

[0174] at least one processor 402; and

[0175] at least one memory 404, the at least one memory comprising computer program code;

[0176] The at least one memory 404 and the computer program code are configured to, with the at least one processor 402 , cause the device 10 to at least perform: any one or more of the described methods.

[0177] like Figure 6, computer program 406 may arrive at device 10 via any suitable delivery mechanism 408. Delivery mechanism 408 may be, for example, a machine-readable medium, a computer-readable medium, a non-transitory computer-readable storage medium, a computer program product, a memory device, a recording medium such as a compact disc read-only memory (CD-ROM) or a digital versatile disk (DVD) or a solid-state memory, an article of manufacture including or tangibly embodying computer program 406. The delivery mechanism may be a signal configured to reliably transfer computer program 406. Device 10 may propagate or send computer program 406 as a computer data signal.

[0178] The computer program instructions are for causing an apparatus to perform at least the following or for performing any one or more of the described methods.

[0179] The computer program instructions may be included in a computer program, a non-transitory computer readable medium, a computer program product, a machine readable medium. In some but not necessarily all instances, the computer program instructions may be distributed over more than one computer program.

[0180] Although memory 404 is illustrated as a single component / circuit, it may be implemented as one or more separate components / circuit(s), some or all of which may be integrated / removable and / or may provide permanent / semi-permanent / dynamic / cache storage.

[0181] Although processor 402 is illustrated as a single component / circuit, it may be implemented as one or more separate components / circuit, some or all of which may be integrated / removable.Processor 402 may be a single-core processor or a multi-core processor.

[0182] References to "computer-readable storage medium," "computer program product," "tangibly embodied computer program," etc., or "controller," "computer," "processor," etc., should be understood to cover not only computers with different architectures (such as single-processor architecture / multi-processor architecture and sequential (von Neumann) architecture / parallel architecture), but also special-purpose circuits (such as field-programmable gate arrays (FPGAs), application-specific circuits (ASICs), signal processing devices, and other processing circuits). References to computer programs, instructions, code, etc. should be understood to cover software or firmware for a programmable processor, such as, for example, the programmable content of a hardware device, whether instructions for a processor, or configuration settings for a fixed-function device, gate array, or programmable logic device, etc.

[0183] The blocks illustrated in the figure may represent steps in a method and / or parts of code in a computer program 406. The illustration of a particular order of blocks does not necessarily imply a required or preferred order for the blocks, and the order and arrangement of the blocks may vary. In addition, some blocks may be omitted.

[0184] Where a structural feature has been described, it may be replaced by a means for performing one or more of the functions of the structural feature, whether that function or those functions are described explicitly or implicitly.

[0185] Systems, devices, methods and computer programs can use machine learning that can include statistical learning. Machine learning is a field of computer science that gives computers the ability to learn without explicit programming. If the performance of a computer at a task in a class of tasks T (as measured by P) improves with experience E, the computer learns from experience E relative to a class of tasks T and performance metric P. Computers can usually learn from previous training data to predict future data. Machine learning includes all or part of supervised learning and all or part of unsupervised learning. It can achieve discrete output (e.g., classification, clustering) and continuous output (e.g., regression). Machine learning can be achieved, for example, using different methods, such as cost function minimization, artificial neural networks, support vector machines, and Bayesian networks. Cost function minimization can be used, for example, for linear regression and polynomial regression and k-means clustering. For example, an artificial neural network with one or more hidden layers models the complex relationship between an input vector and an output vector. Support vector machines can be used for supervised learning. A Bayesian network is a directed acyclic graph that represents the conditional independence of many random variables.

[0186] The term "comprising" as used in this document has an inclusive rather than exclusive meaning. That is, any reference to X comprising Y indicates that X may comprise only one Y or may comprise more than one Y. If "comprising" is intended to be used in an exclusive sense, it will be made clear in the context by reference to "comprising only one" or by use of "consisting of".

[0187] In this specification, reference has been made to various examples. The description of the features or functions of an example indicates that those features or functions exist in the example. Whether explicitly stated or not, the use of the term "example" or "for example" or "can" or "may" in the text indicates that such features or functions exist at least in the described example, whether or not described as an example, and they may but not necessarily exist in some or all other examples. Therefore, "example", "for example", "can" or "may" refers to a specific case in a class of examples. The attributes of a case may be attributes of only the case, or attributes of a class, or attributes of a subclass of a class including some but not all cases in the class. Therefore, it is implicitly disclosed that the features described with reference to one example rather than with reference to another example can be used as part of a working combination in the other example where possible, but not necessarily in the other example.

[0188] Although examples have been described in the preceding paragraphs with reference to various examples, it should be appreciated that modifications to the examples given can be made without departing from the scope of the claims.

[0189] Features described in the preceding description may be used in combinations other than the combinations explicitly described above.

[0190] Although functions have been described with reference to certain features, those functions may be performable by other features whether described or not.

[0191] Although features have been described with reference to certain examples, those features may also be present in other examples whether described or not.

[0192] The terms "a" or "the" as used in this document have an inclusive, rather than an exclusive, meaning. That is, unless the context clearly indicates otherwise, any reference to X including one / the Y indicates that X may include only one Y or may include more than one Y. If "a" or "the" is intended to be used with an exclusive meaning, it will become clear from the context. In some cases, the use of "at least one" or "one or more" may be used to emphasize the inclusive meaning, but the absence of these terms should not be considered to infer any exclusive meaning.

[0193] The presence of a feature (or combination of features) in a claim is a reference to that feature or (combination of features) itself, and also to features that achieve substantially the same technical effect (equivalent features). Equivalent features include, for example, features that are variants and achieve substantially the same result in substantially the same way. Equivalent features include, for example, features that perform substantially the same function in substantially the same way to achieve substantially the same result.

[0194] In this specification, reference has been made to various examples of using adjectives or adjective phrases to describe characteristics of an example. Such descriptions of characteristics of an example indicate that the characteristics exist in some examples exactly as described, and in other examples substantially as described.

[0195] While an effort has been made in the foregoing description to draw attention to those features regarded as important, it will be appreciated that the applicant may seek protection via the claims with respect to any patentable feature or combination of features mentioned above and / or shown in the accompanying drawings, whether or not emphasis has been set forth therein.

[0196] appendix: The attached appendix provides further details and forms part of the specification.

[0197]

[0198] Proceedings of the International Conference on Bioelectromagnetics, Electrical Bioimpedance, and Electrical Impedance Tomography

[0199] June 28-July 1, 2022 / Kyung Hee University, Seoul, South Korea

[0200] Impedance properties of various human breast tissues

[0201] Gerald Sze 1 ,Zhao Song 1,3 ,Huijuan Zhang 2 ,Jie Wang 2 ,Qi He 2 ,Rui Guo 1 ,BarryBueles 3 and Wei Wang 1,3,4

[0202] 1 Micro Image Biotech Co., Ltd., Ningbo, China

[0203] 2 Shanghai China Welfare Institute International Peace Maternity and Child Health Hospital, China

[0204] 3 Micro Image Biotech International Ltd., Cambridge, United Kingdom

[0205] 4 Weilian Biotech Co., Ltd., Ningbo, China

[0206] Correspondence: Wei Wang, Email: w97wang@yahoo.co.uk

[0207] Abstract - In a joint clinical study with a pathology department, a series of electrical impedance spectroscopy (EIS) experiments have been performed to extract the unique characteristics of benign and malignant human breast tissue in vitro. Digital images of the pathology sections of each paraffin-embedded breast tissue have been taken, allowing the effect of unwanted tissue types mixed with target tissue types to be studied. By identifying mixed tissue in the digital pathology section images, the unwanted mixed tissue can be removed from the EIS data and the impedance pattern of the target tissue focused on, for example, cancer cells. The impedance patterns of the following four tissue types were initially extracted: adipose tissue, connective tissue, invasive ductal carcinoma (IDC), fibroadenoma. The information within this data will help achieve the goal of being able to predict potentially different diagnostic outcomes based on different measurable vital parameters for each patient, such as BMI and age.

[0208] Keywords: human breast tissue; impedance characteristics; electrical impedance spectroscopy; EIS; breast cancer.

[0209] 1. Introduction

[0210] Different types of breast tissue were isolated in vitro based on pathology from surgically removed breast tissue, and EIS data for each tissue type was digitally recorded prior to paraffin sectioning of the tissue as pathology sections. Based on the impedance pattern characteristics of human tissue published by Cole and Cole (1941), and human breast tissue published by Jossinet (1998) and Wang (2001) as references for guidance, the impedance pattern characteristics of several tissue types within human breast tissue were determined using EIS data.

[0211] One of the challenges of EIS studies of in vitro breast tissue is the mixing of different tissue types in the same biopsy or excised pathology sample, which causes huge changes in the measured overall impedance. The method adopted is to analyze digital pathology slice images to determine the different tissue types present in the sample breast sample. The direction of the current inside the sample during the EIS measurement is also analyzed. The Cole-Cole impedance parameters for different pure tissue types are used to determine the different tissue types present in the same test pathology sample. The EIS data set of different pure tissue types present in the overall tissue is then extracted from the clean pure tissue sample data set examined at the cellular level. The real EIS impedance is compared with the simulated EIS impedance, and the ratio of mixed tissues in the simulated data is highly matched with the ratio found from the digital pathology slice.

[0212] 2. Data Collection

[0213] 2.1 Sampling size

[0214] 68 EIS datasets have been acquired from the hospital, of which 31 sets were available for analysis. 11 sets (35% of the 31 sets) were taken from patients diagnosed with benign tumors, and 21 sets (68% of the 31 sets) were taken from patients diagnosed with malignant tumors. Note that some cases were diagnosed with both benign and malignant tissues.

[0215] 2.2 Quality Control of EIS Samples

[0216] All datasets ("dataset" means all EIS measurements from all tissues acquired from one patient in the surgical suite to the pathology lab) must meet the following criteria to ensure that they are of sufficient quality to be "valid". There are two types of data comparisons: one comparing different tissue types taken from the same patient; the other comparing the same tissue type taken from different patients.

[0217] Postoperative timing

[0218] EIS data must be collected within 15 minutes of surgery, with the tissue stored in a temperature-controlled chamber (6°C). After this time, the electrical impedance properties in human cells deteriorate rapidly.

[0219] Background salt water reference dataset

[0220] Prior to human tissue EIS measurements, an EIS background saline reference dataset must be collected (fill the test chamber with 1000 μS / cm saline). The saline EIS dataset is used as a reference to calibrate the internal impedance differences of the EIS test cell.

[0221] Tissue size according to the volume of the test chamber

[0222] The tissue is cut to the exact size that fits into the cylindrical test chamber (5 mm in diameter and 8 mm in length). Saline is injected to ensure good contact between the tissue and the electrodes located at both ends of the chamber. The size of the tissue sample must not be smaller than the volume of the chamber to avoid the current bypassing the tissue sample through the saline. The saline should not fill more than 10% of the volume of the chamber. Otherwise, the impedance pattern of the test tissue will be distorted, making the impedance measurement of the sample invalid.

[0223] Data comparison

[0224] There are two types of data comparison: one comparing different tissue types obtained from the same patient; the other comparing the same tissue type obtained from different patients.

[0225] 2.3 Digital Pathology Slide Images

[0226] The following diagram outlines the process used to capture digital pathology slide images of breast tissue with different fields of view (FOV):

[0227]

[0228] 2.4 Age and BMI distribution

[0229] 31 valid data sets were collected from 31 patients accordingly: age under 40 years = 7, age between 40 and 60 years = 14, age over 60 years = 10; BMI below 23 = 12, BMI between 23 and 25 = 10, BMI above 25 = 9. 21 patients were diagnosed with IDC tumors, and 11 patients were diagnosed with fibroadenoma tumors.

[0230] 3. Methods

[0231] 3.1 Process

[0232] There are three steps to verify pure tissue impedance: i. Classify tissues into "pure" and "mixed" groups based on pathological slice images (examples of mixed tissues and pure tissues are shown in Figure 2 ii. from the “pure” tissue, determine the impedance data and extract the Cole-Cole parameters as described in Section 3.2; iii. use a patented cascade model (Wang’s U.S. patent), iv. eliminate the mixed tissue impedance component from the real EIS data to produce “clean” data, and the impedance converges to the “pure” tissue impedance.

[0233] 3.2 Equations

[0234] The Cole-Cole equation (Cole 1940, Cole and Cole 1941, McAdams and Jossinet 1995) shows the change in impedance at different frequencies:

[0235] Z=R ∞ +(R0+R ∞ ) / (1+(jf / F r )) ∝ (1)

[0236] Where R ∞ and R0 = resistance at infinite frequency or zero frequency

[0237] F r = relaxation frequency

[0238] α = relaxation time.

[0239] When we assume that the measurement object is a simplified cell suspension model with a three-component (R, S, C) electrical equivalent circuit (Fricke and Morse 1925), the following equation can be used:

[0240] Z=RS / (R+S)+(R-RS / (R+S)) / (1+(jf / F r )) ∝ (2)

[0241] Where R = extracellular resistance, which is equal to R0 in (1)

[0242] S = intracellular resistance

[0243] R ∞ =R and S in parallel.

[0244] Four Cole-Cole parameters: R, S, Fr, and α are calculated from the EIS bioimpedance characteristic curve using these two equations.

[0245] 3.3 Ensemble Model

[0246] The figure below shows the integrated model composed of Z1, Z2 and Z3 organizations for calculating simulated resistivity:

[0247]

[0248] Where Z1 = impedance characteristic of the target "pure" tissue

[0249] Z2 and Z3 = impedance characteristics of the "mixed" tissue in series and parallel with the target tissue, with values ​​varying depending on the pathology slice image.

[0250] 4. Results

[0251] 4.1 Digital pathology slide images of mixed tissue and pure tissue

[0252] The following pathology slide images show the connective tissue affected by the composition of different adipose tissues:

[0253]

[0254] Figure 2 Connective tissue with different compositions of adipose tissue (upper left: heavy; upper right: moderate; lower left: slight; lower right: none).

[0255] 4.2 Calculation of Cole-Cole parameters for pure tissues

[0256] Based on the selected EIS data set, the impedance characteristic curves of four "pure" tissues: fat, connective tissue, IDC and fibroadenoma are calculated and plotted as follows:

[0257]

[0258] FIG. 3 Plots of real (left) and hypothetical (right) impedance characteristics of the average impedance from pure tissue versus frequency.

[0259] The Cole-Cole parameters for each tissue type were then calculated and listed as follows:

[0260] Table 1 Calculation of Cole-Cole parameters for four types of pure tissues

[0261]

[0262] Table 2 lists the Cole-Cole parameters for impedance of freshly excised human breast tissue published in the Jossinet paper.

[0263] Table 2 Calculation of Cole-Cole parameters for four types of pure tissues

[0264]

[0265] 4.3 Comparison between simulated and actual measurements

[0266] Below are examples of actual measured and simulated EIS impedance (real part) data:

[0267]

[0268] Figure 4 Actual (left) and simulated (right) EIS impedance versus frequency for a tissue with 34% IDC and 66% connective tissue.

[0269] 4.4 Standard deviation after purification

[0270] The Cole-Cole parameters for each tissue type were then calculated and listed as follows:

[0271]

[0272] Figure 5 Standard deviation of connective tissue (green), IDC (red), and fibroadenoma (blue) before (left) and after (right) purification. Dashed lines above and below the solid line are + / - the standard deviation.

[0273] 5. Discussion

[0274] The R and S calculated from the Shanghai connective tissue dataset dropped from more than 20 Ω·m from the Jossinet dataset to less than 4 Ω·m. This may be caused by the presence of adipose tissue within the measured connective tissue.

[0275] The relaxation frequencies of connective tissue, adipose tissue, and fibroadenoma tissues from the Shanghai dataset are much higher than those from the Jossinet dataset. At this point, we need more data to validate these new findings. However, IDC is the tissue type with the most similar Cole-Cole parameters extracted from the EIS dataset.

[0276] 6. Conclusion

[0277] The presented work is necessary before electrical impedance mammography (EIM) images can be diagnosed because each pixel in a 3DEIM image is blended with surrounding tissue and the impedance characteristics of the blended tissue must be identified with solid evidence, which can be achieved by using pathology slice images and EIS data on the same tissue.

[0278] Acknowledgements

[0279] We thank the Shanghai China Welfare Institute International Peace Maternity and Child Health Hospital for providing all data and the opportunity to conduct the clinical trial during the difficult period during the COVID pandemic.

[0280] References

[0281] Cole K 1940, Permeability and impermeability of cell membranes forions, Cold Spring Harbor Symp., Quant.Biol.8 110-22;

[0282] Cole K and Cole R 1941, Dispersion and absorption in dielectrics, J.ofChem.Phys.5 341-350;

[0283] Jossinet J 1998, The impedivity of freshly excised human breasttissue, Physiol.Meas.1961-75;

[0284] McAdams E and Jossinet J 1995, Tissue impedance: a historical overview, Physiol.Meas.16(suppl)A1-A13;

[0285] Wang W et al., 2001, Preliminary results from an EIT breast imaging simulation system, Physiol. Meas. 22(1) 39-48;

[0286] Wang W, 2005.02, Apparatus and method for detecting abnormalities in bodily matter, US6856842 (B1).

Claims

1. A method comprising using a non-target tissue model to virtually purify an experimental electrical impedance signal relative to a target tissue.

2. The method according to claim 1, wherein: The target tissue is tumor tissue.

3. A method comprising: obtaining a purified impedance transfer function based on an electrical impedance model having one or more non-target tissue components and having no target tissue component; An experimental electrical impedance signal is purified using the purified impedance transfer function to obtain a virtual purified signal for the target tissue component.

4. The method according to claim 3, comprising: The experimental electrical impedance signal is purified using the purified impedance transfer function to obtain a virtually purified signal for the target tissue component by assuming an electrical impedance model including the target tissue component and the one or more non-target tissue components, wherein the target tissue is tumor tissue and the non-target tissue is non-tumor tissue.

5. A method comprising: Depending on the measurement of age and / or obesity, access depends on the impedance transfer function of non-tumor tissue; The experimental electrical impedance signal is modified using the obtained impedance transfer function to obtain a signal that depends less on non-tumor tissue and more on tumor tissue impedance.

6. The method according to claim 5, wherein: The measure of obesity is body mass index.

7. A method according to any preceding claim, wherein: The non-tumor tissue impedance is based on connective tissue impedance and adipose tissue impedance.

8. A method according to any preceding claim, wherein: The tumor tissue impedance is benign, such as fibroadenoma tissue impedance, or cancerous, such as IDC tissue impedance.

9. A method according to any preceding claim, wherein: The experimental signal is an electrical impedance signal from in vitro measurement.

10. The method according to any one of claims 1 to 8, wherein: The experimental signal is an electrical impedance signal from in vivo measurement.

11. A method according to any preceding claim, wherein: The electrical impedance signal is used for electrical impedance imaging.

12. A method according to any preceding claim, wherein: The electrical impedance model is a mixed impedance model including target tissue impedance (Z1) and non-target tissue impedance (Z2, Z3).

13. The method according to claim 12, wherein: The target tissue impedance Z1 and the first non-target tissue impedance Z3 are electrically connected in series as a combination, and the combination is electrically connected in parallel with the second non-target tissue impedance Z2.

14. The method according to claim 13, wherein: Depending on the desired target, the target tissue impedance Z1 is one of benign such as fibroadenoma impedance or cancer such as IDC impedance, the first non-target tissue impedance Z3 is one of connective tissue impedance and adipose tissue impedance, and the second non-target tissue impedance Z2 is the other of connective tissue impedance and adipose tissue impedance.

15. A method according to any preceding claim, comprising using an in vitro sample to determine the impedance (Z1) of the target tissue and / or using an in vitro sample to determine the impedance of a non-target tissue, optionally using an in vitro sample to determine the first non-target tissue impedance Z3 and / or using an in vitro sample to determine the impedance of the second non-target tissue impedance Z2.

16. A method according to any preceding claim, wherein: The electrical impedance model is determined by best fitting to in vitro data for one or more samples.

17. The method according to any one of claims 13 to 16, wherein: The relative ratio of a first non-target tissue impedance Z3 to a second non-target tissue impedance for the electrical impedance model is determined by optimization (best fit), wherein the impedance (Z1) of the target tissue is known or is derived from a purified tissue database determined and calibrated based on cell-based tissue properties identified by pathology.

18. A method according to any preceding claim, wherein: The electrical impedance model is converted into a non-target model for non-target tissue by setting the impedance (Z1) of the target tissue in the electrical impedance model to zero.

19. A method according to any preceding claim, wherein: The non-target tissue model (Z2, Z3) is used to estimate the non-target component, and the non-target component is removed from the experimental electrical impedance signal to estimate the target signal, wherein the removal occurs in the time domain or the frequency domain.

20. A method according to any preceding claim, wherein: The non-target tissue models (Z2, Z3) are used to estimate non-target signals, and the non-target signals are subtracted from the experimental signals to estimate the target signals.

21. The method of claim 20, comprising analyzing in vivo target signals to identify and locate spatial electrical impedance changes indicative of tumor tissue.

22. The method of claim 20, comprising imaging the tumor tissue.

23. An apparatus comprising means for performing the method of any preceding claim.

24. A computer program comprising computer code which is then executed by one or more processors to perform the method according to any one of claims 1 to 22.

25. An imaging system comprising means for performing the method of any preceding claim.

Images