Measurement system, measurement method, and information processing method

Abstract

The present technology relates to a measurement system, a measurement method, and an information processing method that make it possible to identify a region of interest with higher accuracy in a shorter period of time. The measurement system comprises a control unit that, for each of a plurality of electrodes included in an MEA disposed in a well in which cells are cultured, identifies a region of interest by performing processing based on measurement information different from measurement information obtained via measurements performed by the electrodes, and either performs new measurements using only electrodes included in the region of interest or deletes measurement information obtained by electrodes other than those included in the region of interest from the measurement information of the plurality of electrodes. The present technology can be applied to a measurement system.

Description

Measurement system, measurement method, and information processing method

[0001] The present technology relates to a measurement system, a measurement method, and an information processing method, and more particularly to a measurement system, a measurement method, and an information processing method that enable a region of interest to be identified with higher accuracy and in a shorter time.

[0002] There is a device that arranges microelectrodes in an array and electrochemically measures the potential of the solution on those microelectrodes.

[0003] Among these, there is a device in which a culture medium is filled on microelectrodes, biological cells consisting of at least a cell body and an axon are placed on the microelectrodes, and action potentials generated by the electrical activity of the biological cells are measured; such a device is known as a microelectrode array (MEA) (see, for example, Patent Document 1).

[0004] In particular, in recent years, CMOS-MEA has attracted attention as a device that uses CMOS (Complementary Metal Oxide Semiconductor) integrated circuit technology to integrate electrodes, amplifiers, AD (Analog to Digital) converters, etc. onto a single chip and simultaneously measures potential using densely formed multipoint electrodes (see, for example, Non-Patent Documents 1 and 2).

[0005] Measurement of electrical activity in cultured human cells using CMOS-MEA is expected to become a technique that can be used to evaluate the toxicity of new compounds such as new pharmaceuticals, cosmetics, and pesticides in vitro without animal testing (see, for example, Non-Patent Document 3).

[0006] Action potentials due to electrical activity of living cells are generated by the exchange of ions between the inside and outside of the cell via ion exchange channels in the cell membrane. To capture this potential change in either the cell body or the axon, it is necessary to measure high-speed potential fluctuations at the submillisecond level. Furthermore, as mentioned above, when evaluating the cytotoxicity of a compound, it is necessary to evaluate the long-term cellular response to the administered compound, so continuous measurement over several minutes or more is required (see, for example, Non-Patent Document 3).

[0007] Therefore, if one were to attempt to record the potential measurement data for all electrodes in a CMOS-MEA, which has a total of tens to hundreds of thousands of electrodes, the measurement data generated by a single potential measurement would be on the scale of several TB (terabytes), making it difficult to handle such large-scale data.

[0008] On the other hand, with a CMOS-MEA, an electrode located very close to the cell body or axon can detect the action potential of that cell, but the action potential of that cell has almost no effect on the measurement value of an electrode located far from the cell. Therefore, when there is no cell on an electrode, the potential measured by that electrode is almost entirely noise, and the measurement value of such an electrode is unnecessary information for measuring cell activity.

[0009] Therefore, in practice, recording is not performed on all electrodes, but rather the area on the MEA chip where cells are located is identified and measurements are performed only on electrodes limited to that area, thereby reducing the amount of data recorded compared to recording measurement data on all electrodes.

[0010] For example, some commercially available CMOS-MEA systems perform pre-measurement of action potentials at all electrodes in order to select the electrodes on which cells are located when selecting 1,024 of the 26,800 electrodes that can be used for simultaneous measurement.

[0011] This CMOS-MEA system detects the electrical activity of cells, i.e., spike waveforms that can be considered spontaneous firing, at each electrode, and automatically selects 1,024 electrodes in order of spike frequency and intensity.

[0012] Japanese Patent Application Laid-Open No. 2002-31617

[0013] “Switch-Matrix-Based High-Density Microelectrode Array in CMOS Technology”, IEEE Journal of Solid State Circuits Vol.45 (2010) No.2 pp.467-482 “Active pixel sensor array for high spatio-temporal resolution electrophysiological recordings from single cell to large scale neural networks”, Lab on a Chip Vol.9 (2009) pp.2647-2651”Large-area field potential imaging having single neuron resolution using 236 880 electrodes CMOS-MEA technology”, Adv. Sci. 2023, 2207732

[0014] However, with the above-mentioned techniques, it has been difficult to specify the region of interest (ROI) to be measured, i.e., the electrodes to be used for measurement, with high accuracy and in a short time.

[0015] For example, in the method for identifying electrodes on which cells are placed, which is used in the commercially available CMOS-MEA system mentioned above, the measured potential waveform of each electrode itself is analyzed, and cell firing (occurrence of electrical activity) is estimated based on the frequency and strength of spikes, and the presence or absence of at least one cell body or axon on the electrode is determined.

[0016] However, the change in the action potential of an actual cell is at most a few tens of μV, making it difficult to distinguish it from noise signals. There is room for improvement in the following three areas.

[0017] First, the detection accuracy of cells with small changes in action potential is low.

[0018] Since electrical signals are transmitted between cultured neurons via a neural network of axons that connect the cells, if some cells are not detected and are excluded from the measurement, it is impossible to correctly analyze the data as a neural network. Therefore, it is preferable to be able to detect at least either the cell body or the axon even in action potential waveforms with low signal strength equivalent to random noise.

[0019] Second, a certain amount of time is required to measure the frequency of occurrence, which makes the pre-measurement time required for electrode selection extremely long. For example, in the CMOS-MEA system mentioned above, the pre-measurement takes about 20 minutes.

[0020] Since the activity of biological cells to be measured is easily affected by external environmental factors such as temperature, if pre-measurement takes a long time, the state of the cells will change during that time, making stable measurement difficult. In order to obtain stable and reliable measurement results, it is preferable to measure the electrical activity as soon as possible after removing the cells from the thermostatic chamber before the state changes.

[0021] Third, unless cells fire (electrical activity) in the initial state before the drug is administered, the presence or absence of cells cannot be detected.

[0022] In measurements for compound evaluation, if no firing occurs before compound administration, and a cell reacts and fires only after compound administration, the cell cannot be detected by pre-measurement.

[0023] CMOS-MEA has the property that adhesion of proteins such as cells to the array electrodes on the chip surface generates seal resistance, which causes slight changes in the noise level. However, it is difficult to determine the presence or absence of cells that are not firing from the potential measurement information of a single electrode.

[0024] If it is possible to determine the presence or absence of cells even when no cell firing (electrical activity) is occurring, it will be possible to identify the area of ​​the electrode where the cells are located even when no cell firing occurs in preliminary measurements, which is thought to enable more reliable compound evaluation.

[0025] The present technology has been made in view of such circumstances, and makes it possible to identify a region of interest with higher accuracy and in a shorter time.

[0026] The measurement system of the first aspect of the present technology is equipped with a control unit that identifies a region of interest by performing processing based on measurement information other than the measurement information obtained by measurement using the electrodes themselves for each of the multiple electrodes of an MEA placed in a well in which cells are cultured, and either performs a new measurement using only the electrodes included in the region of interest, or deletes the measurement information of electrodes other than those included in the region of interest from the measurement information of the multiple electrodes.

[0027] The measurement method of the first aspect of the present technology includes a step in which a measurement system identifies a region of interest by performing processing on each of a plurality of electrodes of an MEA placed in a well in which cells are cultured based on measurement information other than the measurement information obtained by measurement of the electrode itself, and either performs a new measurement using only the electrodes included in the region of interest, or deletes the measurement information of electrodes other than the electrodes included in the region of interest from the measurement information of the plurality of electrodes.

[0028] In a first aspect of the present technology, a region of interest is identified by performing processing based on measurement information other than the measurement information obtained by measurement of the electrode itself for each of the multiple electrodes of an MEA placed in a well in which cells are cultured, and a new measurement is performed using only the electrodes included in the region of interest, or the measurement information of electrodes other than those included in the region of interest is deleted from the measurement information of the multiple electrodes.

[0029] A measurement system according to a second aspect of the present technology includes a determination unit that determines whether the first measurement data obtained by measurement at a first electrode among a plurality of electrodes that constitute an MEA is valid data based on the first measurement data and one or more second measurement data associated with the first measurement data.

[0030] An information processing method according to a second aspect of the present technology includes a measurement system determining whether first measurement data obtained by measurement at a first electrode among a plurality of electrodes constituting an MEA and one or more second measurement data associated with the first measurement data are valid data.

[0031] In a second aspect of the present technology, a determination is made as to whether the first measurement data is valid data based on first measurement data obtained by measurement at a first electrode among the multiple electrodes that make up an MEA, and one or more second measurement data associated with the first measurement data.

[0032] 1 is a diagram for explaining the present technology. FIG. 1 is a diagram for explaining an example of the configuration of a measurement system. FIG. 2 is a diagram for explaining specification of ROI using a microscope image. FIG. 3 is a flowchart for explaining measurement processing. FIG. 3 is a diagram for explaining construction of a cell estimation model using a microscope image. FIG. 4 is a flowchart for explaining model construction processing. FIG. 4 is a diagram for explaining specification of ROI using a cell estimation model. FIG. 5 is a flowchart for explaining measurement processing. FIG. 1 is a diagram for explaining specification of ROI using measurement data of adjacent electrodes. FIG. 2 is a diagram for explaining valid data determination and valid data extraction. FIG. 3 is a diagram for explaining an outline of a data flow. FIG. 4 is a diagram for explaining an example of a calculation result of cosine similarity. FIG. 5 is a diagram for explaining valid data determination based on a difference value. FIG. 6 is a diagram for explaining valid data determination based on a random value. FIG. 6 is a diagram for explaining an example of an extraction result of valid data based on a random value. FIG. 7 is a diagram for explaining data format conversion. FIG. 7 is a diagram for explaining data format conversion. FIG. 8 is a diagram for explaining data format conversion. FIG. 8 is a diagram for explaining data format conversion. FIG. 9 is a diagram for explaining data format conversion.

[0033] Hereinafter, embodiments to which the present technology is applied will be described with reference to the drawings.

[0034] <About this technology> This technology identifies electrodes on which at least one of a cell body or an axon is located (in contact with at least one of a cell body or an axon) by using measurement information other than the measurement information of the electrode of interest itself, such as by comparative analysis of the potential waveform measured at each electrode of an MEA with other information (information other than the measured potential waveform of that electrode).

[0035] By doing this, it is possible to identify the electrode on which a cell (at least one of the cell body and the axon) is located with higher accuracy and in a shorter measurement time than determining the presence or absence of at least one of the cell body and the axon based on the potential information of a single electrode alone. In other words, it is possible to identify the region of interest with higher accuracy and in a shorter time.

[0036] For example, it is difficult to separate the change in noise level due to the presence of a cell on the electrode or the spike waveform associated with cell firing (electrical activity) from the noise using only the measured potential waveform of a single electrode of an MEA. However, by evaluating the similarity with other measured information, for example, at the same electrode position or within the area occupied by the same cell, it becomes possible to extract the influence of a cell hidden in the waveform of a single electrode on the potential waveform and easily determine the presence or absence of a cell. Note that, hereinafter, the electrical activity of a cell will also be referred to as the cell firing.

[0037] Examples of the other measurement information mentioned above will be described below.

[0038] For example, as shown by arrow Q11 in Figure 1, the MEA chip that realizes the MEA has multiple electrodes arranged in an array, such as electrode E11, electrode E12, and electrode E13, and these electrodes form the electrode array section.

[0039] This electrode array unit is placed in a well of a well plate, and cells to be measured are cultured in the well. In this example, for example, one cultured cell C11 is present on electrode E11 or electrode E12 in the well. That is, electrode E11 and electrode E12 are in contact with cell C11. In contrast, no cell is present on electrode E13.

[0040] In the MEA, the potential waveform at each electrode is measured, and measurement information (measurement data) indicating the measurement results is recorded.

[0041] As described above, when identifying the electrode in contact with a cell (at least one of the cell body or axon) in a pre-measurement before measurement or after actual measurement, i.e., the electrode on which the cell is located, it is not easy to determine whether or not a cell is present on that electrode using only the measurement information obtained for one electrode.

[0042] Therefore, as the above-mentioned "other measurement information," it is conceivable to use information on a microscopic image of cells obtained by observing the inside of the well with a microscope, as shown by arrow Q12. The information on the microscopic image of cells here is, for example, a microscopic image that includes cells as the subject.

[0043] It is also conceivable to use measurement information obtained from other electrodes in the vicinity of the electrode to be processed as "other measurement information," for example, using electrode E12 instead of electrode E11.

[0044] For example, measurement information obtained from a single electrode, such as electrode E11, has a large noise component in the measured potential waveform, making it difficult to accurately detect the presence or absence of a cell (at least either a cell body or an axon) on the electrode using only the measurement information from that single electrode. Furthermore, although the influence of the noise component can be reduced by averaging the measurement information from a single electrode, it takes a long time to detect the presence or absence of a cell.

[0045] In contrast, this technology performs processing based on "other measurement information" that differs from the measurement information obtained by the electrode itself, such as by comparing and analyzing the measurement information of the electrode being processed with "other measurement information" such as microscopic images and measurement information from other electrodes. This makes it possible to detect (identify) cells that do not spike (fire).

[0046] Hereinafter, the case where a microscopic image (information about a microscopic image of a cell) is used as "other measurement information" and the case where other electrode measurement information is used will be described in more detail.

[0047] <Example 1 of other measurement information: Microscopic observation> An example of the above-mentioned "other measurement information" is information about a microscopic image of cells on an MEA chip, i.e., a microscopic image. In microscopic observation, the cells on the MEA chip and the electrodes on the MEA chip are observed simultaneously, making it possible to identify the electrodes on which the cells are mounted.

[0048] On the other hand, cells are highly transparent to visible light, and it is usually difficult to observe unstained cells with high contrast using a reflectance microscope.

[0049] Furthermore, when stained cell samples are prepared for microscopic observation, the cells die during the preparation process. Therefore, when measuring the action potential of living cells using a CMOS-MEA, it is not possible to identify the electrode on which the cell is located by microscopic observation in advance.

[0050] Therefore, one possible method is to stain the cell sample after measurements have been taken at all electrodes of the CMOS-MEA, observe it under a microscope, identify the electrode on which the cell is located, and then retain only the data from the CMOS-MEA measurement data for the electrode on which the cell is located, deleting the other data.

[0051] In this method, the "other measurement information" is a microscopic image. More specifically, the "other measurement information" is contrast information for determining the presence or absence of cells at coordinates corresponding to the electrodes in the microscopic image. Because the microscopic image contains contrast information that can identify the area of ​​cells, this contrast information can be used as the "other measurement information." In the above method, as a process based on the microscopic image, which is the "other measurement information," a process is performed to identify whether cells are present on the electrodes based on the position of the cells in the microscopic image, thereby identifying the area of ​​interest.

[0052] Another possible method is to use machine learning based on a pair of potential measurement information from a CMOS-MEA (electrode) and microscopic images of cells (teaching data), thereby inferring the presence or absence of cells from the CMOS-MEA measurement information and identifying the electrode.

[0053] In this method, in order to construct a cell estimation model using machine learning, cell samples of the same quality as the cells to be actually measured are used, and the electrical activity of living cells is measured in advance at all electrodes using a CMOS-MEA. After that, a stained sample of the cells is created and microscopic images are taken.

[0054] Since the location of cells is clear in the microscopic image, the microscopic image can be used as training data. Then, by performing machine learning on a pair of the microscopic image and the measured potential data (measurement information), it is possible to construct a neural network model that estimates cells (at least one of the cell body and axon) as a cell estimation model.

[0055] After this preparatory step, electrodes on which cells (at least either cell bodies or axons) are located can be identified using a cell estimation model through pre-analysis or other methods, based solely on the potential data (measurement information) actually measured by the CMOS-MEA, and electrodes on which no cells are located can be excluded from measurement and recording. This prevents the generation of unnecessary data. In the above method, as processing based on the microscopic images ("other measurement information"), a cell estimation model generated based on the microscopic images is used to identify whether cells are present on the electrodes, thereby identifying the region of interest.

[0056] <Example 2 of other measurement information: Potential measurement information of nearby electrodes> Another example of the above-mentioned "other measurement information" is potential measurement data (potential measurement information) of another electrode located near the electrode that is being focused on as a target for identifying the presence or absence of a cell (at least one of a cell body or an axon).

[0057] As mentioned above, CMOS-MEAs have a high electrode density, and the distance between adjacent electrodes is equal to or less than the size of a cell. This means that the action potential generated by a single cell on the MEA chip is measured in the same way at multiple electrodes on or in close proximity to that cell, resulting in highly similar waveforms being measured at these electrodes.

[0058] By evaluating the similarity of these waveforms, it is possible to separate the measured potential waveforms originating from cells from noise waveforms that are not correlated for each electrode, and to determine whether or not cells are present.

[0059] This is because the waveforms of the potential measurements obtained from electrodes carrying the same cell are similar not only for spike waveforms generated by cell firing, but also for changes in the noise level when the cell is not firing.For this reason, for multiple electrodes carrying the same cell, the presence or absence of a cell on the electrode of interest can be determined based on the similarity of the measured potential waveforms, regardless of whether the cell is firing or not.

[0060] Therefore, in this example, as a process based on the measurement information (measurement data) of other electrodes, which is "other measurement information," the measurement information of the electrode of interest is compared with the measurement information of other electrodes different from the electrode of interest, and a process is performed to evaluate the similarity of the measured potential waveforms to identify the area of ​​interest.

[0061] As a specific method for evaluating the similarity of waveforms, cosine similarity can be used, which is described in, for example, the above-mentioned Non-Patent Document 3.

[0062] For example, the cosine similarity between the measured potential at a single electrode of interest and the measured potentials at each of the eight adjacent electrodes surrounding that electrode is calculated. If the value of this cosine similarity is equal to or greater than a certain value, it can be assumed that the waveforms are highly similar, that a cell is present on the electrode of interest, and that signals originating from the same cell are also being detected at the surrounding electrodes.

[0063] By performing the above-mentioned cosine similarity calculation and threshold processing on all electrodes of the pre-measured CMOS-MEA measurement data, it is possible to identify electrodes on the MEA chip that have cells (at least either cell bodies or axons) attached to them, and to exclude electrodes without cells from measurement and recording. This prevents the generation of unnecessary data.

[0064] First Embodiment Configuration Example of Measurement System Now, a more specific embodiment of the present technology described above will be described.

[0065] 2 is a diagram showing an example of the configuration of an embodiment of a measurement system to which the present technology is applied. In this embodiment, a microscopic image is used as the above-mentioned "other measurement information."

[0066] The measurement system 11 shown in FIG. 2 includes an MEA chip 21 , a readout circuit board 22 , an FPGA (Field Programmable Gate Array) 23 , a measurement PC (Personal Computer) 24 , and a microscope 25 .

[0067] For example, the measurement system 11 uses a well plate having wells 26 to perform in vitro measurement of electrical activity of cells cultured in the wells 26 .

[0068] The MEA chip 21 is made of a semiconductor chip having MEA functions, such as a CMOS-MEA, and has an electrode array section made up of a plurality of electrodes (high-density multi-electrodes) arranged at high density.

[0069] In this example, cells are cultured in the well 26 with the MEA chip 21, more specifically, the electrode array portion of the MEA chip 21, placed in the well 26. In particular, the MEA chip 21 is placed at the bottom of the well 26.

[0070] Therefore, among the electrodes that make up the electrode array section, cells are present on some of the electrodes, and no cells are present on other electrodes.

[0071] The MEA chip 21 reads out measurement data (measurement information) indicating the measurement results of the action potential of the cells obtained by measuring the potential at each electrode, and outputs the data to the circuit board 22 .

[0072] In the following, for simplicity of explanation, an example in which one well 26 is provided in the well plate will be described, but it is also possible to provide multiple wells 26 in the well plate and place an MEA chip 21 in each of these wells 26.

[0073] The readout circuit board 22 is composed of, for example, a circuit provided on a semiconductor substrate on which the MEA chip 21 is mounted. The readout circuit board 22 supplies a control signal supplied from the measurement PC 24 via the FPGA 23 to the MEA chip 21, and reads out measurement data of each electrode obtained by measurement from the MEA chip 21 and supplies the data to the FPGA 23.

[0074] The FPGA 23 controls the operation of the device on which the MEA chip 21 and the readout circuit board 22 are mounted, in accordance with a control signal (command) supplied from the measurement PC 24 .

[0075] For example, the FPGA 23 supplies a control signal supplied from the measurement PC 24 to the MEA chip 21 via the readout circuit board 22, thereby causing the MEA chip 21 to measure action potentials and supplying measurement data supplied from the readout circuit board 22 to the measurement PC 24.

[0076] The measurement PC 24 is, for example, a personal computer (PC) and controls the overall operation of the measurement system 11. The measurement PC 24 includes a control unit 31 and a storage 32.

[0077] The control unit 31 is composed of, for example, a CPU (Central Processing Unit) and controls the overall operation of the measurement system 11 by executing predetermined measurement software (programs).

[0078] For example, the control unit 31 generates a control signal to perform measurement of the action potential of the cell and supplies it to the FPGA 23, or supplies the measurement data supplied from the FPGA 23 to the storage 32 to record the measurement data.

[0079] Furthermore, for example, the control unit 31 controls the microscope 25 to observe the cells in the well 26 and acquires microscopic images from the microscope 25 .

[0080] The storage 32 is made up of, for example, a non-volatile memory, and records the measurement data supplied from the control unit 31 and various data such as measurement software.

[0081] The microscope 25 is, for example, a reflective optical microscope for observing the fluorescence of the cells in the well 26. The microscope 25 also has a photographing function, i.e., is provided with a camera.

[0082] The microscope 25 observes the inside of the well 26 under the control of the measurement PC 24 (control unit 31), and takes an image of a part or all of the area inside the well 26 as a subject, and supplies the resulting microscopic image to the control unit 31. This microscopic image includes the cells cultured in the well 26 as a subject.

[0083] <Description of Measurement Processing> The operation performed by the measurement system 11 will be described with reference to Figures 3 and 4. In Figure 3, parts corresponding to those in Figure 1 are given the same reference numerals, and their description will be omitted as appropriate.

[0084] In the measurement system 11, cells are cultured in the well 26, for example, as shown by the arrow Q21 in FIG.

[0085] In this example, each rectangle represents one electrode that constitutes the electrode array portion of the MEA chip 21. Furthermore, for example, cell C11 is present on electrodes E11 and E12, while no cell is present on electrode E13. In other words, electrodes E11 and E12 are in contact with cell C11, while electrode E13 is not in contact with any cell.

[0086] When the cells (cell sample) are cultured on the MEA chip 21 in the well 26 in this way, the measurement process shown in Fig. 4 is started. Fig. 4 is a flowchart showing the measurement process performed by the measurement system 11.

[0087] First, in the measurement system 11, a compound such as a drug whose cytotoxicity is to be evaluated is administered to the cultured cells on the MEA chip 21. Then, the cells in the wells 26 begin to react to the compound.

[0088] In step S11, the control unit 31 of the measurement PC 24 outputs a control signal to cause action potential measurement to be performed on all electrodes constituting the electrode array portion of the MEA chip 21.

[0089] Specifically, the control unit 31 designates all electrodes of the electrode array unit (MEA chip 21) as regions to be measured for the cellular action potential, i.e., as electrodes, and generates a control signal instructing the execution of action potential measurement at those designated electrodes. The control unit 31 then supplies the generated control signal to the MEA chip 21 via the FPGA 23 and readout circuit board 22, causing the MEA chip 21 to measure the action potential at all electrodes.

[0090] The MEA chip 21 performs measurements at all electrodes in response to the supplied control signal and outputs the resulting measurement data, i.e., measurement data (measurement information) indicating the measurement results of the action potential at each electrode. The measurement data for each electrode output by the MEA chip 21 is supplied to the control unit 31 via the readout circuit board 22 and FPGA 23.

[0091] Such measurement of action potentials at all electrodes is performed continuously (continuously) for a period of time, for example, after administration of a compound, until the cell response (reaction) stabilizes, specifically, for about 10 minutes, etc. Therefore, measurement data for a predetermined period, such as 10 minutes, is supplied to the control unit 31 for each electrode.

[0092] In step S12, the control unit 31 supplies the measurement data for all electrodes supplied from the MEA chip 21 via the FPGA 23 and the readout circuit board 22 to the storage 32, where it is stored.

[0093] As a result, for example, as shown by arrow Q22 in FIG. 3, measurements are performed on all electrodes, and the resulting measurement data for all electrodes is temporarily saved (recorded) in storage 32.

[0094] Thereafter, in the wells 26, a staining and fixing process is carried out to stain and fix the cells (cultured cells) on the MEA chip 21 so that the cells can be observed under a microscope.

[0095] The control unit 31 outputs a control signal to the microscope 25 as appropriate to observe the cells, and the microscope 25 performs processing for microscopic observation of the cells, such as irradiating the cells with excitation light in response to the control signal from the control unit 31. The microscope 25 then observes the cells on the MEA chip 21, for example, as indicated by arrow 23 in Figure 3. Figure 3 shows the state when the cells are fluorescently observed using a reflective optical microscope as the microscope 25, and it can be seen that cells are present in region R11, for example.

[0096] Furthermore, the control unit 31 instructs the microscope 25 to capture an image of the cell by supplying a control signal to the microscope 25. That is, the control unit 31 causes the microscope 25 to capture a microscopic image (microscopically captured image) of the stained cell as the subject after measurement at all electrodes in step S11.

[0097] In step S13, the microscope 25 photographs the cells in the well 26 in accordance with instructions from the control unit 31. That is, the microscope 25 photographs the area inside the well 26 as the subject. This results in a microscopic image that includes the cells in the well 26 as the subject. This microscopic image is an example of information related to the microscopic image of the cells, and is used as "other measurement information."

[0098] When photographing, the entire area of ​​the electrode array section of the MEA chip 21 in the well 26, i.e., the absolute coordinates (positions) of the cells in the area of ​​all electrodes, is photographed in one go, or photographed in multiple locations.

[0099] The microscope 25 supplies the microscopic image obtained by photographing to the control unit 31. The microscopic image is image information obtained by photographing a certain region of the electrode array part of the MEA chip 21 in the well 26, and therefore can be said to be measurement information obtained by measuring (photographing) the certain region of the electrode array part.

[0100] In step S14, the control unit 31 identifies the coordinates (area) where the cell exists on the microscopic image based on the microscopic image supplied from the microscope 25, that is, the position of the cell in the microscopic image.

[0101] For example, the control unit 31 identifies an area in the microscopic image where cells are located by performing binarization processing, image recognition processing, etc. on the microscopic image. For example, in fluorescent observation, the luminance value of the area of ​​cells is high and the luminance value of the area other than the cells is low, so it is possible to identify the area of ​​cells by performing binarization processing.

[0102] Alternatively, for example, the control unit 31 may display a microscopic image on a display (not shown) of the measurement PC 24, and identify the area (position) of the cell by receiving an input operation from the user, i.e., an operation to specify the area of ​​the cell on the microscopic image.

[0103] In addition, when the entire area of ​​the electrode array section of the MEA chip 21 is divided into multiple locations and photographed, a stitching process may be performed as appropriate based on the microscopic images obtained by photographing each location, and a single microscopic image corresponding to the entire area of ​​the electrode array section may be generated. In such a case, the single microscopic image obtained by the stitching process is used to identify the area (position) of the cell.

[0104] In step S15, based on the result of identifying the region (coordinates) where cells exist in step S14, the control unit 31 identifies the electrodes on which cells are placed (in contact with the cells) among the electrodes constituting the electrode array unit of the MEA chip 21. In other words, the control unit 31 identifies the ROI (region of interest) by determining whether a cell (at least one of a cell body or an axon) exists on each electrode constituting the electrode array unit.

[0105] For example, in the measurement system 11, the correspondence between each position within the observation field of the microscope 25, i.e., each position on the microscopic image, and the position of each electrode constituting the electrode array part of the MEA chip 21 at the bottom of the well 26 is known. Therefore, the control unit 31 can identify the electrode corresponding to the region of a cell from the result of identifying that region in the microscopic image.

[0106] Within the entire area of ​​the electrode array section of the MEA chip 21, the area consisting of the electrodes on which the cells are mounted is the ROI (region of interest) for measuring the electrical activity of the cells. In other words, the area of ​​the entire electrode array section occupied by the cells (at least either the cell body or the axon) is the ROI.

[0107] In step S16, the control unit 31 analyzes the measurement data for all electrodes stored in storage 32 in step S12 based on the results of identifying electrodes on which cells are present in step S15, and deletes the measurement data for electrodes on which no cells are present.

[0108] That is, based on the result of identifying the electrode on which the cell is located, the control unit 31 identifies measurement data other than the measurement data of the electrode on which the cell is located from the measurement data of all electrodes, in other words, measurement data of electrodes other than those included in the ROI, and deletes this identified measurement data.

[0109] As a result, for example, the state in which measurement data for all electrodes is saved, as shown by arrow Q22 in Figure 3, changes to a state in which only measurement data for electrodes with cells on them is saved, as shown by arrow Q24. In other words, unnecessary data, that is, measurement data for electrodes without cells on them (electrodes outside the ROI), is deleted from the measurement data for all electrodes.

[0110] The processing of steps S14 to S16 can be said to be a process of identifying the area consisting of electrodes where cells exist (are mounted) within the entire area of ​​the electrode array section of the MEA chip 21 as the ROI (region of interest), and extracting and saving (recording) only the measurement data in the identified ROI.

[0111] In particular, these processes can be said to be processes that identify an ROI (region of interest) based on at least the microscopic image, which is the above-mentioned "other measurement information," and delete measurement data of regions (positions) other than the ROI from the measurement data of the entire region. In other words, the measurement data of the ROI is identified by analysis based on the microscopic image and the measurement data of the entire region.

[0112] When using a microscopic image to identify the ROI, it is possible to identify it without the need for long-term measurements in advance. In other words, it is possible to identify the ROI in a short time. Moreover, by performing image processing such as binarization on the microscopic image, it is possible to accurately identify the electrode (ROI) on which the cell is placed.

[0113] When the process of step S16 is performed and measurement data measured in the ROI (region of interest), that is, measurement data only of the electrodes where cells are present, is saved, the measurement process ends.

[0114] In this way, the measurement system 11 acquires measurement data and microscopic images of all electrodes by measurement, identifies the electrodes on which cells are located, and extracts and saves only the measurement data of the electrodes on which cells are located. In this way, by using the microscopic images, it is possible to identify the ROI (region of interest) with higher accuracy and in a shorter time.

[0115] Second Embodiment Explanation of Model Construction Process In the measurement system 11 shown in FIG. 2, a cell estimation model can be constructed in advance by learning using a microscopic image (information related to a microscopic image of a cell) as the above-mentioned "other measurement information," and an ROI (region of interest) can be identified using the cell estimation model.

[0116] Hereinafter, an embodiment using such a cell estimation model will be described, namely, a method using machine learning using microscopic images of cells as training data.

[0117] In this embodiment, a cell estimation model is constructed as a preliminary step in the measurement system 11. The cell estimation model is formed, for example, by a convolutional neural network (CNN) for cell estimation.

[0118] First, construction of a cell estimation model will be described with reference to Figures 5 and 6. In Figure 5, parts corresponding to those in Figure 1 are given the same reference numerals, and their description will be omitted as appropriate.

[0119] In the measurement system 11, cells are cultured in the well 26, for example, as shown by the arrow Q41 in Fig. 5. In this example, cells are cultured in the same manner as shown by the arrow Q21 in Fig. 3.

[0120] However, here, cells (cell sample) of the same type and quality as the cells that will ultimately be measured by the MEA chip 21 are cultured. That is, in this example, the cells C21 to be cultured are cells of the same type and quality as the cells that will ultimately be measured.

[0121] When the cells (cell sample) are cultured on the MEA chip 21 in the well 26 in this way, the model construction process shown in Fig. 6 is started. Fig. 6 is a flowchart showing the model construction process performed by the measurement system 11 as advance preparation.

[0122] In step S41, the control unit 31 of the measurement PC 24 outputs a control signal to cause action potential measurement to be performed on all electrodes constituting the electrode array portion of the MEA chip 21.

[0123] After the action potentials at all the electrodes have been measured, the MEA chip 21 supplies the control unit 31 with the measurement data for each electrode obtained by the measurement via the readout circuit board 22 and FPGA 23 .

[0124] In step S42, the control unit 31 supplies the measurement data for all electrodes supplied from the MEA chip 21 via the FPGA 23 and the readout circuit board 22 to the storage 32, where it is stored.

[0125] The processes in steps S41 and S42 are similar to those in steps S11 and S12 in FIG. 4, and therefore will not be described further.

[0126] However, in step S11 described above, measurements are performed in a state where a compound has been administered to the cells, whereas in step S41, measurements are performed in a state where no compound has been administered to the well 26. Furthermore, in step S41, unstained cells are the measurement targets.

[0127] In step S41, measurements are taken continuously (successively) for all electrodes for, for example, about 10 seconds. Therefore, as shown by arrow Q42 in Figure 5, measurement data for a relatively short period, such as 10 seconds, is saved for each electrode in storage 32.

[0128] After the measurement data is obtained, a staining and fixing process is performed in the wells 26 to stain and fix the cells (cultured cells) on the MEA chip 21 so that the cells can be observed under a microscope.

[0129] After the staining and fixing process is performed, the cells are photographed in step S43. Note that the process in step S43 is similar to the process in step S13 in FIG. 4, and therefore a description thereof will be omitted.

[0130] When the process of step S43 is performed, the microscope 25 supplies the microscopic image obtained by photographing to the control unit 31. This microscopic image is information about the microscopic image of the stained cells after measurement.

[0131] In step S43, too, the entire area of ​​the electrode array section of the MEA chip 21 in the well 26 (area of ​​all electrodes) is photographed in one go, or photographed in multiple locations so that the absolute coordinates (positions) of the cells in the entire area of ​​the electrode array section (area of ​​all electrodes) can be identified.

[0132] In step S44, the control unit 31 generates a still image from the measurement data obtained in step S42 and stored in the storage 32.

[0133] For example, the control unit 31 outputs 10 seconds of measurement data from all electrodes as moving image data of the action potentials of all electrodes, and generates still images at each time by dividing the 10 seconds of moving image data into 100 parts in the time direction. In this case, an image at a given time (frame) of the moving image data is constructed from the measurement data from all electrodes at a given time. Then, one still image showing the measurement results of the action potentials at all electrodes at one time is generated from images of one or more consecutive frames of the moving image data.

[0134] As a result, for example, as indicated by arrow Q43 in FIG. 5, 100 still images for 100 time periods including the still image P11 at the predetermined time period are obtained, that is, 100 still images.

[0135] In step S45, the control unit 31 generates one cell image based on the microscopic image of the cell obtained in step S43.

[0136] For example, when the entire area of ​​the electrode array section of the MEA chip 21 is divided into multiple locations and photographed, the control unit 31 performs stitching or the like as appropriate so that all electrode areas are included as subjects in a single image, and the resulting single image is used as a cell image. Also, when a single photograph is taken of the entire electrode area, the resulting single microscopic image is used as a single cell image.

[0137] As a result, a single still image is obtained as cell image P12, in which the position of each cell within the region consisting of all the electrodes can be identified, as shown by arrow Q44 in Figure 5. In this example, cells are included as subjects in regions R21 and R22 on cell image P12, for example.

[0138] In step S46, the control unit 31 performs binarization processing on the cell image obtained in step S45, using a threshold value for the brightness value of the cell, thereby generating a binary image from the cell image.

[0139] In a cell image, the brightness value of an area with cells is high and the brightness value of an area without cells is low, so by performing binarization processing, it is possible to separate cells (at least either cell bodies or axons) from areas other than cells. In other words, the binarized image becomes an image that shows the areas of cells.

[0140] In step S47, the control unit 31 generates learning still images from the still images obtained in step S44, i.e., the still images based on the measurement data, by dividing the images into regions.

[0141] As an example, for each of a total of 100 still images at each time, the entire area of ​​the still image is divided into 10 x 10 areas in the surface direction, and 100 learning still images are generated from one still image.

[0142] That is, one still image is divided into 100 regions (10 x 10), and the image in each divided region is used as a learning still image. Therefore, for example, from the 100 still images obtained in step S44, a total of 10,000 learning still images (100 regions x 100 times) are obtained.

[0143] For example, in the portion indicated by the arrow Q43 in FIG. 5, an image of one divided area on one still image P11 is set as one learning still image P21.

[0144] In step S48, the control unit 31 divides the entire area of ​​the binary image obtained in step S46 into 10 x 10 areas in the surface direction, particularly in the surface direction of the electrode array portion of the MEA chip 21, thereby generating teacher data (teacher images) for 100 areas from one binary image.

[0145] For example, in the portion indicated by the arrow Q44 in FIG. 5, the cell image P12 is divided into a plurality of regions such as region R31, and training data is generated.

[0146] As a result, images of the areas occupied by cells (at least one of the cell body and axons) can be obtained as training data. For example, label information indicating whether or not the training data includes a cell as a subject can be added to the training data as needed.

[0147] Through the above processing, pairs of 10,000 training still images and cell teacher data corresponding to each of these training still images (each region of the microscopic image) were obtained for use in machine learning of the cell estimation model.

[0148] In step S49, the control unit 31 generates a cell estimation model by performing machine learning based on a pair of the learning still image obtained in step S47 and the teacher data (teacher image) obtained in step S48. For example, in this case, the cell estimation model is generated (constructed) by deep learning using CNN.

[0149] This allows us to obtain, as a cell estimation model, an image showing the action potential measurement results, i.e., a trained CNN optimized for estimating the cell position (area) based on the measurement data from each electrode, as shown by arrow Q45 in Figure 5.

[0150] The cell estimation model obtained in this way is a model for identifying the electrode where a cell is present, i.e., the ROI (region of interest), from the measurement data at each electrode.

[0151] The cell estimation model receives, as input, for example, measurement data from one or more electrodes, more specifically, still images made up of the measurement data from each electrode. In particular, in this example, still images taken at multiple different times are used as input to the cell estimation model.

[0152] The output of the cell estimation model is the determination result of whether or not a cell exists in each region of the input still image, in other words, whether or not a cell is placed (exists) on each electrode corresponding to each measurement data. This determination result can be said to indicate the ROI identification result.

[0153] The control unit 31 supplies the cell estimation model obtained by machine learning to a memory such as the storage 32 and records (stores) it, and the model construction process ends.

[0154] In this way, the measurement system 11 performs machine learning based on the measurement data of each electrode and the microscopic images as the above-mentioned "other measurement information," and generates a cell estimation model that can identify the measurement data of the electrode on which the cell is placed, in other words, the ROI (region of interest).

[0155] If such a cell estimation model is constructed (generated) in advance as a preliminary step, then when measuring the electrical activity of the cells, the cell estimation model can be used to identify the ROI (region of interest) with greater accuracy and in a shorter time.

[0156] <Explanation of Measurement Processing> Once the cell estimation model described with reference to FIG. 6 has been constructed, the toxicity of a cell to a compound can be evaluated at any timing thereafter using the cell estimation model.

[0157] The measurement process performed by the measurement system 11 will be described below with reference to FIGS.

[0158] In the measurement system 11, cells are first cultured in the wells 26. In this case, the cells to be cultured are of the same type and quality as those in the construction of the cell estimation model described above, i.e., in the advance preparation. Furthermore, at this stage, no compounds such as drugs are administered.

[0159] When cells are cultured, for example, as shown by arrow Q51 in Figure 7, each electrode that makes up the electrode array portion of the MEA chip 21 will have electrodes on which cells (at least either cell bodies or axons) are placed and electrodes on which no cells are placed.

[0160] In the portion indicated by the arrow Q51, one square represents one electrode that constitutes the electrode array portion of the MEA chip 21.

[0161] In particular, in this example, hatched electrodes, such as electrode E31, represent electrodes that have cells on them (are in contact with cells), whereas unhatched electrodes, such as electrode E32, represent electrodes that do not have cells on them (are not in contact with cells).

[0162] After the cells are cultured, the measurement process shown in Fig. 8 is started. Fig. 8 is a flowchart showing the measurement process performed by the measurement system 11.

[0163] In step S81, the control unit 31 of the measurement PC 24 outputs a control signal to perform pre-measurement of action potentials at all electrodes constituting the electrode array part of the MEA chip 21. This pre-measurement is performed in a state where no compound is administered to the wells 26.

[0164] When the action potential is measured in advance using all the electrodes, the measurement data of each electrode obtained by the measurement is supplied from the MEA chip 21 to the control unit 31 via the readout circuit board 22 and FPGA 23 .

[0165] In addition, since the processing in step S81 is the same as the processing in step S41 in FIG. 6, a detailed description thereof will be omitted.

[0166] For example, in step S81, measurements are made continuously (successively) at all electrodes for about 10 seconds without any compound being administered to well 26. Therefore, measurement data for a relatively short period, such as 10 seconds, is supplied to control unit 31 for each electrode.

[0167] In the measurement system 11, by using a cell estimation model, it is possible to identify with sufficiently high accuracy the region where a cell exists, i.e., the electrode on which the cell (at least one of the cell body or axon) is located, even from measurement data of a short time. Therefore, it is sufficient to perform preliminary measurement for a short time, such as about 10 seconds. In other words, the time for preliminary measurement is set to be shorter than the time for the main measurement that is performed thereafter.

[0168] In step S82, the control unit 31 generates still images showing the measurement results of the action potential at each of a plurality of times during the period in which the preliminary measurement was performed, based on the measurement data of all the electrodes obtained in step S81.

[0169] In step S82, the same process as in step S44 in Fig. 6 is carried out. As a result, 100 still images for 100 time periods, i.e., 100 still images, are generated from the measurement data of all the electrodes for, for example, 10 seconds.

[0170] In step S83, the control unit 31 identifies the electrode where the cell is located (the electrode on which the cell is located) based on the multiple still images obtained at each time point in step S82 and the cell estimation model that was generated in advance by the model construction process of Figure 6 and recorded in storage 32, etc.

[0171] For example, as shown by arrow Q52 in Figure 7, the control unit 31 inputs the still images obtained at each time point in step S82, i.e., the measurement data obtained by the preliminary measurement, into the cell estimation model, performs calculations, and obtains, as an output of the cell estimation model, a determination result as to the presence or absence of a cell for each region (position). Based on this determination result, the control unit 31 determines whether or not a cell is present on each electrode constituting the electrode array part of the MEA chip 21.

[0172] This identifies the electrodes to be measured, as shown by arrow Q53 in Figure 7. In other words, the ROI (region of interest) consisting of the electrodes on which cells are located is identified. In this example, the hatched electrodes, such as electrode E31, represent the electrodes on which cells are located.

[0173] Once the electrodes on which cells are placed, in other words, the ROI (region of interest) is identified, a compound such as a drug whose cytotoxicity is to be evaluated is administered to the cells on the MEA chip 21 in the cultured well 26, as shown by arrow Q54 in Figure 7. Then, with the compound administered to the well 26, only the electrodes identified as having cells placed on them are used to perform a new measurement of the action potential (main measurement).

[0174] That is, in step S84, the control unit 31 outputs a control signal to cause actual measurement of action potentials to be performed only on the electrodes that constitute the electrode array portion of the MEA chip 21 and that were identified in the processing of step S83.

[0175] Specifically, the control unit 31 designates only the electrodes on which the cells are located, as identified in the processing of step S83, as electrodes included in the region of interest (ROI) for the main measurement of the cellular action potential, i.e., as electrodes for the main measurement, and generates a control signal instructing the execution of action potential measurement at these designated electrodes. The control unit 31 then supplies the generated control signal to the MEA chip 21 via the FPGA 23 and readout circuit board 22, causing the MEA chip 21 to newly measure the action potential using only the designated electrodes.

[0176] In response to the supplied control signal, the MEA chip 21 performs measurements at the electrodes specified by the control unit 31 and outputs the resulting measurement data, i.e., measurement data indicating the measurement results of the action potential at each electrode. The measurement data for each electrode output by the MEA chip 21 is supplied to the control unit 31 via the readout circuit board 22 and FPGA 23.

[0177] Measurement of action potentials at such designated electrodes, that is, electrodes included in the ROI, is carried out continuously (continuously) for, for example, about 10 minutes after administration of the compound.

[0178] In step S85, the control unit 31 supplies the measurement data of each specified electrode, which is supplied from the MEA chip 21 via the FPGA 23 and the readout circuit board 22, to the storage 32 and stores it.

[0179] As a result, for example, as shown by arrow Q55 in Figure 7, action potentials are measured only at the electrodes on which cells are placed, and the measurement data for the electrodes on which measurements were taken are saved (recorded) in storage 32.

[0180] Once the measurement data is stored in the storage 32, the measurement process ends.

[0181] In this way, the measurement system 11 identifies the electrodes on which the cells are located, i.e., ROI (region of interest), based on the measurement data for all electrodes obtained in the preliminary measurement and the constructed cell estimation model, and performs the actual measurement using only those identified electrodes.

[0182] This shortens the measurement time for pre-measurement and enables the electrode (ROI) on which the cell is located to be identified with high accuracy using a cell estimation model.

[0183] In other words, by using a cell estimation model constructed based on microscopic images as "other measurement information," it is possible to identify the ROI (region of interest) with greater accuracy and in a shorter time.

[0184] Third Embodiment Example of Configuration of Measurement System A case will be described in which measurement data (measurement information) obtained by measurement at other electrodes in the vicinity of the electrode of interest is used as the above-mentioned "other measurement information."

[0185] In such a case, the measurement system 11 may be configured as shown in Fig. 9. In Fig. 9, the same reference numerals are used to designate parts corresponding to those in Fig. 2, and the description thereof will be omitted as appropriate.

[0186] The measurement system 11 shown in Fig. 9 is composed of an MEA chip 21 to a measurement PC 24. That is, the configuration of the measurement system 11 shown in Fig. 9 is a configuration in which the microscope 25 in the measurement system 11 shown in Fig. 2 is not provided.

[0187] In this example, whether or not a cell is present on the electrode of interest is determined based on the measurement data of the electrode of interest and the measurement data of another electrode different from the electrode of interest.

[0188] <Description of Measurement Processing> The operation of the measurement system 11 shown in Fig. 9 will be described with reference to Fig. 10 and Fig. 11. Fig. 10 shows a flowchart illustrating the measurement processing performed by the measurement system 11.

[0189] In the measurement system 11, cells are cultured in the wells 26. After the cells are cultured, preliminary measurement for measuring the electrical activity of the cells is started at an arbitrary timing.

[0190] That is, in step S111, the control unit 31 of the measurement PC 24 outputs a control signal to perform pre-measurement of action potentials at all electrodes constituting the electrode array part of the MEA chip 21. This pre-measurement is performed in a state where no compound is administered to the wells 26.

[0191] After the action potentials at all electrodes have been measured in advance, the MEA chip 21 supplies the control unit 31 with the measurement data for each electrode obtained by the measurement via the readout circuit board 22 and FPGA 23. The control unit 31 then supplies the measurement data to the storage 32 as appropriate to store it.

[0192] In addition, the process of step S111 is the same as the process of step S11 in Fig. 4, and therefore a detailed description thereof will be omitted. In this case as well, the time for the preliminary measurement is set to be shorter than the time for the main measurement that is performed thereafter.

[0193] For example, in step S111, measurements are continuously (successively) taken at all electrodes for about 10 seconds without any compound being administered to well 26. Therefore, control unit 31 acquires measurement data for each electrode for a relatively short period, such as 10 seconds.

[0194] In step S112, the control unit 31 calculates, for each electrode, the cosine similarity between the electrode and other adjacent electrodes based on the measurement data obtained in step S111.

[0195] For example, as shown by arrow Q61 in Figure 11, let us focus on a specific region R61 in the electrode array section of the MEA chip 21. In the portion indicated by arrow Q61, each square represents one electrode. In addition, in region R61, cells are placed on some of the electrodes.

[0196] In region R61, there are a total of nine electrodes, arranged in a 3 x 3 pattern with electrode E61 at the center, as indicated by arrow Q62. Focusing on electrode E61 located in the center, in step S112, the cosine similarity between electrode E61 and each of the eight electrodes adjacent to electrode E61 is calculated as the cosine similarity for electrode E61.

[0197] As a specific example, the calculation of the cosine similarity between electrode E61 and electrode E62 adjacent to the right side of electrode E61 will be described.

[0198] The measurement data for n frames obtained by measurement at electrode E61 is denoted as x, and the measurement data for n frames obtained by measurement at electrode E62 is denoted as y.

[0199] Furthermore, data of a predetermined frame k (where k=1, 2, . . . , n) constituting the measurement data x is referred to as measurement data x ｋ and the data of a predetermined frame k constituting the measurement data y is expressed as measurement data y ｋ It will be written as follows.

[0200] In this case, the control unit 31 calculates the cosine similarity between the electrodes E61 and E62, i.e., the cosine similarity cos(x, y) between the measurement data x and the measurement data y, by calculating the following equation (1) based on the measurement data x and the measurement data y.

[0201]

[0202] The control unit 31 calculates the cosine similarity between each electrode provided in the electrode array unit of the MEA chip 21 and all of the electrodes adjacent to that electrode. Calculating the cosine similarity between two electrodes in this way can be said to be comparing and analyzing the measurement data of the two electrodes.

[0203] For example, if there is a correlation between the measurement data of two electrodes, that is, if the measurement data of the two electrodes are similar, it can be assumed that the similarity is due to the same factor. Specifically, if the same cell is placed on two electrodes, the measurement data of the two electrodes will be similar, and the cosine similarity will be high.

[0204] For example, the noise observed (measured) at the two electrodes is similar not only when cells on the two electrodes are electrically active and an action potential is generated, but also when the cells are not electrically active.

[0205] In contrast, when one of the two electrodes has a cell on it and the other does not, the noise levels of the electrodes are different.Furthermore, the noise levels (random noise) observed between the two electrodes without cells are also different.

[0206] Therefore, based on the cosine similarity between two electrodes, it is possible to determine whether a cell is located on an electrode even when the cell is not firing (not performing electrical activity). In other words, it is possible to identify the cell located on the electrode.

[0207] This identification method using cosine similarity can be said to be a method of evaluating the similarity of the potential waveforms (measurement data) measured by the electrodes using cosine similarity to identify whether or not a cell is located on the electrode of interest.

[0208] In particular, when this technique is used, at the time when a cell on the electrode of interest is firing, it is possible to identify that a cell is located on the electrode of interest based on the similarity of the potential waveform when the cell is electrically active, i.e., the waveform of the action potential, between the electrode of interest and other electrodes.

[0209] In contrast, at times when the cell on the electrode of interest is not firing, the presence of a cell on the electrode of interest is identified based on the similarity of the potential waveform, i.e., the noise waveform, between the electrode of interest and other electrodes when the cell is not electrically active.

[0210] Although an example in which cosine similarity is used as the similarity of the measurement data will be described here, the present invention is not limited to this, and any other similarity may be used.

[0211] Returning to the explanation of the flowchart in Figure 10, in step S113, the control unit 31 identifies electrodes among the electrodes constituting the electrode array portion of the MEA chip 21 in which cells exist (on which cells are placed) based on the cosine similarity calculated for each electrode in step S112.

[0212] For example, the control unit 31 performs threshold processing to compare the cosine similarity calculated for the electrode of interest with a predetermined threshold, thereby determining whether or not a cell is present on the electrode of interest.

[0213] Specifically, for example, assume that there are eight adjacent electrodes for a target electrode, and the cosine similarities between these adjacent electrodes are calculated, i.e., eight cosine similarities are calculated for the target electrode.

[0214] In such a case, for example, the control unit 31 determines that a cell is present on the electrode of interest if, among the eight cosine similarities calculated for the electrode of interest, a predetermined number of cosine similarities whose values ​​are equal to or greater than a predetermined threshold value are equal to or greater than 1. Here, the predetermined threshold value can be, for example, 0.3.

[0215] The determination (identification) of whether or not a cell is present on the electrode of interest may be performed in any manner, for example, by determining that a cell is present on the electrode of interest if the average, median, maximum, etc. of the eight cosine similarities is equal to or greater than a predetermined threshold value.

[0216] The control unit 31 performs the above-mentioned threshold processing for each electrode on all electrodes provided in the electrode array unit of the MEA chip 21, and determines whether or not a cell is present on each electrode, i.e., whether or not the electrode is the target of measurement and recording. This identifies the ROI (region of interest) consisting of the electrodes on which cells are present.

[0217] Once the electrode on which the cells are placed, in other words the ROI (region of interest), has been identified, a compound whose cytotoxicity is to be evaluated is then administered to the cells on the MEA chip 21 in the cultured well 26.

[0218] Thereafter, the processes of steps S114 and S115 are performed while the compound is being administered, and the measurement process ends, but these processes are similar to the processes of steps S84 and S85 in Fig. 8, and therefore a description thereof will be omitted. For example, in step S114, the main measurement is performed continuously (successively) for about 10 minutes after the compound is administered.

[0219] Here, an example has been described in which an ROI (region of interest) is specified based on the measurement results of the preliminary measurement, and then the main measurement is performed using only the electrodes within the ROI based on the result of the specification.

[0220] However, this is not limiting, and the main measurement may be performed on all electrodes, and an ROI (region of interest) may be specified based on the measurement results of the main measurement by the processes of steps S112 and S113. In such a case, the measurement data of all electrodes obtained in the main measurement except for the electrodes within the specified ROI is deleted, so that only the measurement data of the electrodes within the ROI is ultimately saved.

[0221] In this way, the measurement system 11 compares (analyzes) the measurement data of adjacent electrodes based on the measurement data of all electrodes obtained in the preliminary measurement, thereby identifying the electrode on which the cell is located, i.e., the ROI (region of interest), and performs the actual measurement using only the identified electrode.

[0222] This allows the electrode (ROI) on which the cell is located to be identified with high accuracy even with a short preliminary measurement. In other words, by using the measurement data (measurement information) of electrodes near the electrode of interest as "other measurement information," the ROI (region of interest) can be identified with higher accuracy and in a shorter time.

[0223] As described in the first to third embodiments, according to the present technology described above, it is possible to specify an ROI (region of interest) with higher accuracy and in a shorter time.

[0224] For example, with this technology, measurement data from all electrodes within the action potential measurement area is analyzed to identify electrodes on which cells are located, and unnecessary data generated from electrodes on which no cells are located can be extracted and deleted, thereby reducing the amount of measurement data that is ultimately stored.

[0225] Furthermore, for example, with this technology, electrodes on which cells are present can be identified through prior measurement and data analysis, and measurements can be performed using only the identified electrodes in the actual measurement, thereby suppressing the generation of unnecessary data from electrodes on which cells are not present, and reducing the amount of measurement data that is ultimately stored.

[0226] <Fourth embodiment> <Regarding reduction of data volume> Incidentally, although CMOS-MEA devices are capable of measuring a wide range with single-neuron level resolution by using a large number of high-density electrodes, it is known that the volume of data output from an MEA system having a CMOS-MEA device becomes enormous.

[0227] For example, if 16-bit data is sampled at 10 kHz on a CMOS-MEA device with 230,000 electrodes, the data volume will be approximately 4.4 GB (gigabytes) per second.

[0228] As the amount of data increases, not only does it become difficult to record the data output from the MEA system on storage devices such as PCs, but it also places a heavy burden on the storage and management of the recorded data and the analysis of the acquired data. Therefore, in MEA systems using CMOS-MEA devices, data handling - how to record, store, and analyze the increasing amount of data - becomes extremely important.

[0229] Therefore, in this technology, only data useful to the user is extracted from data output from an MEA device such as a CMOS-MEA device, and the data is further converted into an optimal data format and compressed, thereby reducing the amount of data. In this embodiment, extracting useful data (hereinafter referred to as "valid data") can also be said to identify a region of interest at a certain timing. In other words, according to this embodiment, it is possible to identify a region of interest spatially and temporally with higher accuracy and in a shorter time.

[0230] The valid data is data that is useful (significant) to the user, that is, data that includes signals of interest to the user, such as data that includes signals resulting from the potential activity of cells.

[0231] To determine whether data is valid, specific calculations are performed on a specific piece of data (measurement data) using multiple related measurement data, and the results of these calculations determine whether the measurement data is valid. Related measurement data can be, for example, measurement data from the same electrode at a different time (time) or measurement data from other nearby electrodes. Furthermore, data compression reconstructs the original measurement data, which was a list of sampled data (sampled values), into a data format consisting only of the extracted valid data. This reduces the amount of data output from the MEA system.

[0232] MEA devices are primarily used to measure cellular action potentials, but action potentials are not always generated by cells. When focusing on a certain period of time, the proportion of the period during which action potentials are generated is relatively small compared to the period during which no action potentials are generated. Also, action potentials are not observed at all electrodes in an MEA device; action potentials are not observed at electrodes where there are no active cells (not in contact) or at electrodes with low contact with cells.

[0233] Therefore, measurement data acquired from MEA devices is generally a set of spatially and temporally sparse data, characterized by a large proportion of noise data. Therefore, it is expected that the amount of data can be significantly reduced by extracting data that is useful to users and reducing noise data.

[0234] In MEA devices such as CMOS-MEA devices that can record data (measurement data) with sufficiently high spatial resolution, cells contact multiple electrodes, resulting in similar data groups being observed at multiple adjacent electrodes. Similarly, in MEA devices such as CMOS-MEA devices that can record data (measurement data) with sufficiently high temporal resolution, action potentials generated by a cell are recorded as multiple related data groups over a certain time interval. For example, data from a time interval consisting of multiple samples is recorded as data for a single spike portion corresponding to the cell's activity.

[0235] Therefore, it is considered that determining whether a certain measurement data of interest is valid data can be realized by performing a specific calculation process using one or more measurement data that are spatially or temporally different from the measurement data. Based on the determination result, it becomes possible to extract valid data and discard noise data. Spatially or temporally related measurement data includes, for example, measurement data from electrodes that are spatially close (neighboring), such as adjacent electrodes, or measurement data measured by the same electrode but in different time intervals, i.e., data in different time intervals within the measurement data.

[0236] An example of measurement data obtained from the electrodes that make up the MEA device is shown in Figure 12. In Figure 12, the horizontal axis represents time, and the vertical axis represents the amplitude value (voltage value) of the measurement data.

[0237] The upper part of Figure 12 shows the measurement data measured (recorded) at one electrode constituting the MEA device, i.e., the time-series waveform data of the cell's action potential. In this example, spike-shaped action potentials are recorded (observed) three times, and the data for the other periods is noise.

[0238] The middle section of the figure shows the results of valid data determination for the measurement data shown in the upper section of the figure. In this example, the section marked with the character "Yes" indicates a section of valid data, and the section marked with the character "Invalid" indicates a section of invalid data that is not valid data, i.e., noise. For example, of the measurement data, the period (section) of the cell's action potential is determined to be valid data, and the period other than the period determined to be valid data is determined to be invalid data (noise).

[0239] 12, the lower part shows an example of data obtained by deleting invalid data from the measurement data (waveform) based on the result of valid data determination, more specifically, data in which invalid data portions in the measurement data have been replaced with 0 data (zero data). In other words, an example of the result of valid data extraction is shown. In this example, data compression is achieved by replacing invalid data portions of the measurement data with 0 data.

[0240] In addition to the determination of whether the data is valid (hereinafter also referred to as valid data determination) and the extraction of valid data, the data is converted into a data format that effectively holds only the extracted valid data. For example, if a data group for a predetermined time period is extracted as valid data for a certain electrode, all that is required is information on the electrode coordinates, timestamp, and the extracted series of data group, and the data can be handled by converting it into a data format that includes only this information.

[0241] The flow of the above processing is shown in FIG. 13 as an outline of the data flow.

[0242] In the example shown in FIG. 13, a measurement system for measuring the action potential of cells includes an MEA device 101 such as a CMOS-MEA device, a valid data determination and extraction circuit 102, and a data format conversion circuit 103.

[0243] The MEA device 101 is provided with a plurality of electrodes, and outputs measurement data obtained by measurement using each electrode to a valid data determination and extraction circuit 102 .

[0244] The valid data determination and extraction circuit 102 performs the above-mentioned valid data determination on the measurement data for each electrode supplied from the MEA device 101 and extracts valid data based on the determination results, and supplies the valid data, or more specifically, the measurement data in which invalid data has been replaced with 0 data, to the data format conversion circuit 103.

[0245] The data format conversion circuit 103 converts the valid data (measurement data) supplied from the valid data determination and extraction circuit 102 into an appropriate data format and outputs the converted valid data (measurement data) to a subsequent stage. The valid data output from the data format conversion circuit 103 is recorded in storage at the subsequent stage or displayed on a monitor for user confirmation.

[0246] <Regarding Valid Data Determination> (Determination Method 1) A specific example of valid data determination will be described below.

[0247] First, as a method for determining valid data, a method for determining whether data is valid by focusing on spatial data correlation (hereinafter also referred to as determination method 1) will be described.

[0248] When a cell emits action potentials across multiple electrodes, it is highly likely that measurement data with similar waveforms will be acquired between adjacent electrodes. Therefore, by evaluating spatial data correlation, it is possible to determine whether the measurement data from a certain electrode is valid data derived from cell activity.

[0249] The spatial data correlation can be calculated using the cosine similarity shown in the following equation (2), for example: That is, equation (2) is a definition equation of the cosine similarity.

[0250]

[0251] In equation (2), a represents measurement data for a period consisting of multiple samples of an electrode of interest (hereinafter also referred to as the electrode of interest), and b represents measurement data for a period consisting of multiple samples of an electrode in the vicinity of the electrode of interest (hereinafter also referred to as the reference electrode). For example, the reference electrode is an electrode adjacent to the electrode of interest.

[0252] Furthermore, the measurement data a is data consisting of sampling data (sample values) of n consecutive samples, and in equation (2), a ｉ indicates the i-th (i=1, 2, ..., n) sampling data constituting the measurement data a. Similarly, the measurement data b is data consisting of sampling data of n samples, and in equation (2), b ｉ indicates the i-th (i=1, 2, . . . , n) sampling data constituting the measurement data b.

[0253] In equation (2), the inner product of measurement data a, which is a group of sampling data of the target electrode, and measurement data b, which is a group of sampling data of the reference electrode, is calculated, and this inner product is normalized by the magnitude of the measurement data. Therefore, a value between 0 and 1 is obtained as the cosine similarity. The valid data determination and extraction circuit 102 calculates equation (2) (calculates the inner product) to find the cosine similarity between measurement data a of the target electrode and measurement data b.

[0254] When the waveforms of the measurement data a of the target electrode and the measurement data b of the comparison electrode are similar, the inner product of these measurement data becomes large. Therefore, in determining valid data, a predetermined determination threshold is set in advance, and when the cosine similarity obtained by calculating Equation (2) is equal to or greater than the determination threshold, the measurement data a of the target electrode can be determined to be valid data, i.e., data containing valid data.

[0255] An example of the calculation results of the cosine similarity for each electrode, i.e., a heat map of the cosine similarity, is shown in Fig. 14. Note that in Fig. 14, each square represents an electrode, and the shading of each electrode indicates the magnitude of the cosine similarity calculated for the measurement data of that electrode.

[0256] In this example, when a certain electrode is designated as a target electrode, eight electrodes adjacent to the target electrode are designated as comparison electrodes, and the cosine similarity of the target electrode is calculated for each comparison electrode. The final cosine similarity of the target electrode is then calculated based on these eight cosine similarities. For example, the maximum value, average value, or median of the eight cosine similarities may be used as the final cosine similarity of the target electrode.

[0257] 14, the cosine similarity values ​​are high for eight electrodes in region R101, and it can be assumed that the measurement data for these electrodes contain signals from the same cell. In this case, the measurement data for each of these eight electrodes can be regarded as valid data and retained, while the measurement data for the other electrodes can be regarded as noise data and discarded.

[0258] (Determination Method 2) As a method for determining valid data, a method (hereinafter also referred to as determination method 2) that uses a difference calculation with measurement data of adjacent electrodes to evaluate data correlation in the spatial direction can also be considered.

[0259] In the determination method 2, the difference between the measurement data of the electrode of interest and the measurement data of the electrode adjacent to the electrode of interest is evaluated, and when the difference is small, the measurement data of the electrode of interest is determined to be invalid data, and when the difference is large, the measurement data of the electrode of interest is determined to be valid data. For example, the difference in the measurement data is calculated for each sample.

[0260] Specifically, for example, a method can be considered in which the difference values ​​between the measurement data of the electrode of interest and the measurement data of each of the eight electrodes adjacent to the electrode of interest are calculated, and the median (center value) of these eight difference values ​​is taken. In this case, the median value of the measurement data between the adjacent electrodes serves as an index indicating whether the measurement data of the electrode of interest is valid, and when the median value is equal to or greater than a predetermined judgment threshold, the measurement data of the electrode of interest is considered to be valid.

[0261] In this type of determination method 2, spatially steep data changes are determined to be valid data, and spatially gradual data changes not caused by cellular activity are determined to be noise data. This determination method 2 is based on the concept of an epsilon filter, which has the property of removing small-amplitude noise without impairing sudden signal changes.

[0262] Fig. 15 shows an example of a specific determination result by determination method 2. In Fig. 15, each square represents an electrode.

[0263] The left side of the figure shows the signal value of each electrode at a specific time (hour), i.e., the value of one sampled data constituting the measurement data of each electrode. In particular, the number inside the square representing the electrode indicates the value of the sampled data of that electrode, i.e., the potential (voltage value) observed at the electrode.

[0264] Based on the measurement data for each electrode, the difference between each of the eight adjacent electrodes is calculated, and the median of these difference values ​​is calculated. The result shown in the center of Figure 15 is obtained. In the center of the figure, the number inside the square representing the electrode indicates the median value calculated for that electrode. Furthermore, the square representing the electrode is drawn with a density corresponding to the magnitude of the median value of that electrode. In other words, the shading of each square represents the median value of the electrode. In this example, the median values ​​of the electrodes in the center of the 8 x 8 electrodes are large, indicating that the data changes rapidly near those electrodes.

[0265] The right side of the figure shows the results of threshold processing using a decision threshold for the median value obtained for each electrode, i.e., the result of valid data determination. In other words, the classification results of the measurement data for each electrode as to whether or not there is a signal due to cellular activity are shown.

[0266] On the right side of the figure, the numbers in the squares representing the electrodes indicate the results of the valid data determination for that electrode. In particular, a number "1" indicates valid data, and a number "0" indicates invalid data.

[0267] In this example, it can be seen that the measurement data of the electrode with the large median value in the central part of the figure is determined to be valid data.

[0268] As described above, the measurement data of an electrode that has a large output difference from an electrode that is spatially close (nearby) to the electrode of interest can be regarded as valid data, and other measurement data can be regarded as noise data (invalid data) and discarded.

[0269] (Determination Method 3) As a method for determining valid data, a method (hereinafter also referred to as determination method 3) for determining whether data is valid by focusing on temporal characteristics of the data may be considered.

[0270] Fig. 16 shows an example of a group of sampling data in the time direction when focusing on a certain electrode, i.e., an example of measurement data for a target electrode. In Fig. 16, the vertical axis represents the amplitude value of the measurement data, i.e., the value of each sampling data (sample value), and the horizontal axis represents time. In particular, in Fig. 16, one circle on the polygonal line represents one sample.

[0271] For example, in determination method 3, an evaluation is made as to whether a signal (measurement data) is random or not in a time interval T101 consisting of multiple samples including a sample SP101 that is to be determined as valid data. In this example, the time interval T101 is an interval of seven samples in the measurement data, and the evaluation is made for this time interval T101.

[0272] If the measurement data is noise data, the measurement data is random data, but if the measurement data contains some signal component, the measurement data is no longer random data. Therefore, by calculating the degree of randomness (degree of randomness) of a time interval of interest in the measurement data of the electrode of interest using some calculation method, it is possible to determine whether the time interval of interest contains a signal component, i.e., whether the sampling data of interest is valid data.

[0273] The degree of randomness of the measurement data can be evaluated by calculating the average value of the absolute values ​​of the sampling data in the time interval of interest, for example, as shown in the following equation (3).

[0274]

[0275] In the formula (3), Ra represents a random value indicating the degree of randomness of the sampling data (time interval) of interest, and a ｉ indicates the i-th sampling data in the time interval of interest. Note that, hereinafter, the random value Ra will also be referred to as the Ra value.

[0276] If the Ra value (random value Ra) obtained by formula (3) is high, the degree of randomness of the target time interval is low and some kind of signal is included in that time interval, so the data of the time interval used to calculate the Ra value is determined to be valid data. More specifically, the sampled data of interest included in that time interval is determined to be valid data.

[0277] On the other hand, if the Ra value (random value Ra) is low, the degree of randomness of the time interval is high and no signal is included in that time interval, so the data of the time interval used to calculate the Ra value is regarded as noise data and is determined to be invalid data. More specifically, the sampling data of interest included in that time interval is determined to be invalid data.

[0278] In the valid data determination and extraction circuit 102, one or more pieces of sampled data are treated as a processing unit, and for each processing unit, the random value Ra calculated by equation (3) is compared with a predetermined determination threshold value. If the random value Ra is equal to or greater than the determination threshold value, the sampled data of the processing unit is determined to be valid data.

[0279] It is desirable that the time interval for calculating the Ra value (random value Ra) be an interval equivalent to the time width of the action potential. For example, if the sampling rate of the measurement data is 10 kHz (sampling every 100 μsec), a 1 msec spike of the action potential will be recorded as approximately 10 sampling data points. In such a case, the Ra value can be calculated with high accuracy by setting the time interval width to 1 msec.

[0280] Fig. 17 shows a specific example of valid data determination and valid data extraction by determination method 3. In Fig. 17, the vertical axis represents the amplitude value or Ra value of the measurement data, and the horizontal axis represents time.

[0281] The upper part of FIG. 17 shows an example of measurement data obtained from one electrode, that is, waveform data consisting of sampling data at each time.

[0282] 17 shows the Ra values ​​(random values ​​Ra) at each time point, i.e., the calculation results of the Ra values, obtained by calculating Equation (3) for each piece of sampling data constituting the measurement data shown in the upper part. Also, in the middle part, the dotted line represents the determination threshold for determining valid data, and the time intervals (sampling data) in which the Ra value is equal to or greater than the determination threshold among the measurement data are determined to be valid data.

[0283] The result of valid data extraction based on the result of valid data determination is shown in the lower part of Fig. 17. Here, only the valid data in the measurement data is retained, and invalid data portions in the measurement data are deleted, more specifically, the waveform of data in which the invalid data portions have been replaced with zero data is shown as the result of valid data extraction.

[0284] In the example of Figure 17, the measurement data contains three spike-like action potentials, and it can be seen that the Ra values ​​of those parts are large. It can also be seen that the sampling data of those spike-like parts are extracted as valid data.

[0285] In particular, in this example, the Ra value is calculated for a predetermined time interval including the sampling data of interest. Therefore, for example, in valid data determination, a time interval including the sampling data before and after the spike-like portion is determined to be valid data. In other words, rather than extracting only the strictly useful portion, a time leeway is provided so that the useful data portion of the measurement data and the portions before and after that data portion are extracted as valid data. Furthermore, by performing valid data determination based on the Ra value, for example, it is possible to extract as valid data portions that appear to be noise but are actually useful data.

[0286] <Data Format Conversion> (Conversion Method 1) A specific example of data format conversion performed after valid data extraction will be described.

[0287] As a data format conversion method, we will explain a method (hereinafter also referred to as conversion method 1) that generates data consisting of valid data, information that can identify the electrode from which the valid data was obtained, and information that indicates the time when the valid data was observed.

[0288] In the following, the data obtained by replacing the invalid data portion of the measurement data with 0 data (zero data) will also be referred to as valid measurement data, and the data obtained by converting the data format of valid measurement data will also be referred to as output measurement data.

[0289] In the following, a time interval consisting of sampling data arranged consecutively in time in the measurement data and which is determined to be valid data will also be referred to as a valid interval.Furthermore, in the following, a time interval consisting of sampling data arranged consecutively in time in the measurement data and which is determined to be invalid data will also be referred to as an invalid interval.

[0290] For example, as shown on the left side of FIG. 18, it is assumed that valid measurement data is obtained for each electrode that constitutes the MEA device 101.

[0291] In this example, the valid measurement data of one electrode is time-series data in a data format in which the values ​​of sampling data (sample values) at each time from time t1 to time tN are arranged (arranged) in time series. In this case, the value of sampling data that is determined to be valid data in the valid measurement data is the amplitude value observed at the electrode, and the value of sampling data that is determined to be invalid data in the valid measurement data is 0.

[0292] The data format conversion circuit 103 converts such a data group consisting of the effective measurement data of each electrode into one piece of output measurement data as shown on the right side of the figure. For example, as shown in Figure 12, when only the section (period) of effective data including the action potential of the cell is extracted, the data can be converted into a data format in which the electrode coordinates, time stamp, and data group of the effective data are arranged as one unit.

[0293] In this example, the output measurement data includes data for all valid sections extracted from the valid measurement data of each electrode. For example, data for one valid section in the valid measurement data is referred to as one set of data. In this example, one row of data in the right part of the figure is one set of data.

[0294] Each set of data includes electrode coordinates indicating which electrode the data corresponds to, a timestamp indicating the start time of the valid section, i.e., the time of valid data, and the values ​​of each sampled data item constituting the valid section. That is, each set of data includes sampled data (valid data) measured at the electrode identified by the electrode coordinates at each time point in the valid section beginning at the time indicated by the timestamp. For example, the electrode coordinates are coordinate information indicating the placement position of the electrode in the MEA device 101. While an example of using electrode coordinates as information identifying an electrode will be described here, the information identifying an electrode is not limited to this, and may be any information, such as ID information specific to the electrode.

[0295] In the output measurement data, one set of data (one data set) is data for one effective section (effective period) for one electrode.

[0296] Therefore, for example, if the measurement data for one electrode includes data for three effective intervals, the output measurement data will include three sets of data for that one electrode, i.e., data sets for three different effective intervals (time intervals) for the same electrode.Furthermore, for example, the output measurement data may include data sets for the effective intervals of multiple different electrodes.

[0297] The data bandwidth of output measurement data in this data format, i.e., the output bit rate of the output measurement data, is variable depending on the observation frequency of valid data, etc. However, by setting the data frame of the entire output measurement data, i.e., the data size of the output measurement data, to a fixed size, it is also possible to set the data bandwidth of the output measurement data to a fixed data bandwidth (constant bit rate).

[0298] In such cases, the output measurement data can be made into data of a fixed data size by inserting invalid data into parts of the output measurement data that have no actual data (remaining parts), such as by padding the output measurement data with zeros.

[0299] Specifically, for example, the data of each set is concatenated to form output measurement data, and then zero-padding is performed to add zero data to the end of the output measurement data, thereby obtaining output measurement data of a desired data size. Furthermore, since the data size of one set of data varies depending on the number of samples that make up the valid interval, the data size of each set of data may be made fixed by padding the end of one set of data with zeros, for example.

[0300] As described above, by making the output measurement data data of a fixed size, it becomes easier to handle the output measurement data in the subsequent stages.

[0301] (Conversion Method 2) As a data format conversion method, we will explain a method (hereinafter also referred to as conversion method 2) that generates data having a data section in which actual data, etc. is stored as data for each sample, and a determination bit (determination data) indicating whether the data is valid or not.

[0302] For example, it is assumed that the data shown in FIG. 19 is obtained as valid measurement data for one electrode.

[0303] In this example, one column in the data string indicates one sample of data that constitutes valid measurement data, i.e., one piece of sampling data (actual data), and the numerical value written in that column indicates the value of the sampling data. Note that a column marked with the word "invalid" indicates that the value of the sampling data is 0, i.e., the sampling data is invalid. Also, in this example, one piece of sampling data is 16 bits of data.

[0304] The valid measurement data shown in FIG. 19 includes a valid section consisting of five samples (sampling data), followed by an invalid section consisting of nine samples, and then another valid section consisting of five samples.

[0305] In conversion method 2, the valid measurement data shown in FIG. 19 is converted into data (output measurement data) in the data format shown in FIG.

[0306] 20, the data string portion, consisting of two vertically aligned columns, i.e., the "Decision Bit" column and the "Actual Data" column, represents the data portion of one sample that constitutes the output measurement data. That is, the output measurement data has, as data for each sample, a 1-bit decision bit and a 15-bit data portion in which actual data (sampling data) and the like are stored.

[0307] The determination bit is data (flag information) that indicates whether the data portion of the sample having the determination bit is actual data, i.e., sampling data that has been determined to be valid data, in other words, whether the data portion is data that indicates the number of invalid data. That is, the determination bit is data that indicates whether the data portion of the sample is valid data or data that indicates the width of a section (number of invalid data) that has been determined to be invalid data in the measurement data.

[0308] For example, a judgment bit "0" indicates that the data portion of the sample having that judgment bit is actual data. When the value of the judgment bit is 0, the value of the sampling data of the sample corresponding to the valid measurement data is stored as actual data in the data portion of the sample having that judgment bit. In this case, the actual data is sampling data that has been determined to be valid data.

[0309] For example, in the first sample of the output measurement data shown in FIG. 20, the determination bit is 0, so the data portion of that first sample is "13", the same as the value of the sampling data of the first sample of the valid measurement data shown in FIG. 19.

[0310] Furthermore, a judgment bit of "1" indicates that the data portion of the sample having that judgment bit is not actual data but data indicating the number of invalid data. When the value of the judgment bit is 1, the data portion of the sample having that judgment bit stores data indicating the number of samples in the invalid section starting from the corresponding sample of valid measurement data, that is, the number of consecutive invalid data.

[0311] For example, in the sixth sample from the beginning of the output measurement data shown in Fig. 20, the determination bit is 1. Also, the section from the sixth sample from the beginning to the fourteenth sample of the valid measurement data shown in Fig. 19 is an invalid section. Therefore, in the sixth sample from the beginning of the output measurement data shown in Fig. 20, the value of the data section is "9", which indicates the length of the invalid section of the corresponding valid measurement data, i.e., the number of invalid data consecutively arranged in the time direction.

[0312] In this way, by embedding data indicating the length (width) of the invalid section in the data portion of the output measurement data, the amount of data in the output measurement data can be compressed. Also, by providing a determination bit, it is possible to determine whether the data portion indicates actual data or the number of invalid data.

[0313] (Conversion Method 3) As a data format conversion method, we will explain a method (hereinafter also referred to as conversion method 3) for generating data for each sample that has actual data (valid data) and data indicating the number of subsequent invalid data (hereinafter also referred to as invalid section width data).

[0314] For example, suppose that the data shown in Fig. 19 is obtained as effective measurement data for one electrode. In such a case, conversion method 3 converts the effective measurement data shown in Fig. 19 into data (output measurement data) in the data format shown in Fig. 21.

[0315] 21, the data string portion, consisting of two columns aligned vertically in the figure, namely the "Number of Invalid Data" column and the "Actual Data" column, represents the data portion of one sample that constitutes the output measurement data. That is, the output measurement data has, as data for each sample, 4-bit invalid section width data and 12-bit actual data (sampling data).

[0316] The actual data of one sample (hereinafter also referred to as the target sample) in the output measurement data is the sampling data of the sample of valid measurement data corresponding to that target sample.In this case, the actual data is always sampling data that is considered to be valid data.

[0317] Furthermore, the invalid section width data of a target sample in the output measurement data indicates the width of a section (invalid section) that is set as invalid data and begins with the sample next to the valid measurement data sample corresponding to the target sample. In other words, the invalid section width data indicates the number of samples that make up the invalid section immediately following the valid measurement data sample corresponding to the target sample.

[0318] Therefore, for example, in the fifth sample from the beginning of the output measurement data shown in Fig. 21, the actual data is "35" and the invalid section width data is "9." In the valid measurement data shown in Fig. 19, the value of the sampling data for the fifth sample from the beginning is "35," and the section consisting of the sixth to fourteenth samples from the beginning is an invalid section. From these facts, it can be seen that the valid measurement data in Fig. 19 corresponds to the output measurement data in Fig. 21, that is, it is possible to reproduce the valid measurement data from the output measurement data.

[0319] In this way, by predetermining the bit allocation in the output measurement data, it is possible to embed the actual data and the invalid section width data in the same data without using a determination bit.

[0320] <Configuration Example of Measurement System> Fig. 22 shows a configuration example of a measurement system in the fourth embodiment. In Fig. 22, parts corresponding to those in 13 are given the same reference numerals, and their explanation will be omitted as appropriate.

[0321] The measurement system 151 shown in FIG. 22 includes an MEA system 161 and a PC 162 .

[0322] The MEA system 161 has an MEA device 101 and an FPGA 171. In the MEA system 161, the MEA device 101 measures the action potential of cells, and the resulting measurement data is supplied to the FPGA 171. The FPGA 171 determines whether the measurement data is valid and converts the data format, and outputs the resulting output measurement data to the PC 162.

[0323] The MEA device 101 is a device corresponding to the above-mentioned MEA chip 21 , and includes an electrode sensor 181 , an ADC (Analog to Digital Converter) 182 , and a sensor I / F (Interface) 183 .

[0324] The electrode sensor 181 is disposed, for example, on the bottom surface of a well provided in the MEA device 101, and has a plurality of electrodes arranged in an array. In the MEA device 101, cells to be measured are cultured on the electrodes that make up the electrode sensor 181 in the well, and the action potential of the cells is measured by measuring the potential (voltage value) of each electrode of the electrode sensor 181, i.e., the potential waveform. The electrode sensor 181 sequentially supplies the measurement data of each electrode obtained by measurement to the ADC 182.

[0325] The ADC 182 performs AD conversion on the measurement data supplied from each electrode of the electrode sensor 181, and supplies the resulting digital measurement data to the sensor I / F 183. The sensor I / F 183 supplies the measurement data supplied from the ADC 182 to the FPGA 171.

[0326] For simplicity of explanation, the MEA device 101 will be described as a device with a single well configuration, but the MEA device 101 may also be a device with a multi-well configuration. In such a case, an electrode sensor 181 will be provided for each of the multiple wells.

[0327] The FPGA 171 includes a sensor I / F 184 , a buffer 185 , a valid data determination and extraction circuit 102 , a data format conversion circuit 103 , and a PC I / F 186 .

[0328] The sensor I / F 184 supplies the measurement data of each electrode supplied from the sensor I / F 183 to the buffer 185. The buffer 185 temporarily holds the measurement data supplied from the sensor I / F 184. The valid data determination and extraction circuit 102 reads out the measurement data held in the buffer 185, determines whether the measurement data is valid, and extracts the valid data, and supplies the resulting valid measurement data to the data format conversion circuit 103.

[0329] The data format conversion circuit 103 converts the data format of the valid measurement data supplied from the valid data determination and extraction circuit 102, and supplies the resulting output measurement data to the PC I / F 186. The PC I / F 186 outputs (supplies) the output measurement data supplied from the data format conversion circuit 103 to the PC 162.

[0330] The PC 162 has a PC I / F 187, which acquires the output measurement data output from the PC I / F 186. The output measurement data acquired by the PC I / F 187 is supplied to and recorded in a storage 189 as appropriate. The output measurement data may be supplied to a display (not shown) to display the output measurement data.

[0331] As described above, the measurement system 151 is made up of an MEA system 161 made up of the MEA device 101 and FPGA 171, and a PC 162 that controls the MEA system 161 and records output measurement data.

[0332] The analog measurement data output from the electrode sensor 181 is converted into a digital signal by the ADC 182 in the MEA device 101 and output to the FPGA 171 via the sensor I / F 183 in the MEA device 101. Within the FPGA 171, the valid data determination and extraction circuit 102 determines whether the data is valid and extracts the valid data, and then the data format conversion circuit 103 converts the valid measurement data into data in an optimal data format, i.e., output measurement data. The obtained output measurement data is then output to the PC 162 via the PC I / F 186 in the FPGA 171 and recorded in the storage 189 in the PC 162.

[0333] <Description of Measurement Processing> The operation of the measurement system 151 shown in Fig. 22 will be described below. That is, the measurement processing by the measurement system 151 will be described below with reference to the flowchart in Fig. 23.

[0334] In step S201, the MEA device 101 acquires measurement data for each electrode.

[0335] Specifically, the electrode sensor 181 measures the potential at each electrode and sequentially supplies the resulting measurement data for each electrode to the ADC 182. The ADC 182 also performs AD conversion of the measurement data supplied from the electrode sensor 181. The digital measurement data obtained by the AD conversion is supplied from the ADC 182 to the buffer 185 via the sensor I / F 183 and the sensor I / F 184, and is temporarily stored in the buffer 185.

[0336] In step S202, the valid data determination and extraction circuit 102 reads out the measurement data stored in the buffer 185 and performs a valid data determination on the read measurement data. More specifically, in step S202, for each electrode, valid data determination is performed for each predetermined time interval consisting of one or more samples that make up the measurement data. Also, in step S202, valid data determination is performed based on mutually related measurement data.

[0337] For example, when the valid data determination / extraction circuit 102 performs valid data determination using the above-described determination method 1, it selects each electrode in turn as an electrode of interest and calculates the cosine similarity for each electrode.

[0338] That is, the valid data determination and extraction circuit 102 calculates the cosine similarity of the measurement data of the target electrode based on the above-mentioned formula (2) based on the measurement data of the target electrode and the measurement data of the comparison electrode adjacent to the target electrode. The valid data determination and extraction circuit 102 also calculates the final cosine similarity of the target electrode based on the cosine similarities between the target electrode and each of the multiple comparison electrodes adjacent to the target electrode.

[0339] The valid data determination and extraction circuit 102 then compares the cosine similarity calculated for the electrode of interest with a predetermined determination threshold, and determines that the measurement data for the electrode of interest is valid if the cosine similarity is equal to or greater than the determination threshold. Note that the valid data determination may be based on the similarity of mutually related measurement data calculated by any method, not limited to the cosine similarity.

[0340] Furthermore, for example, when the valid data determination and extraction circuit 102 determines valid data using the above-mentioned determination method 2, it calculates the median of the difference values ​​of the measurement data between each electrode and adjacent electrodes, and compares the median with a predetermined determination threshold value.

[0341] That is, the valid data determination and extraction circuit 102 determines a predetermined electrode as a target electrode, calculates the difference between the measurement data of the target electrode and the measurement data of each of the eight electrodes adjacent to the target electrode, and calculates the median of the obtained eight difference values. If the calculated median is equal to or greater than a predetermined determination threshold, the valid data determination and extraction circuit 102 determines that the measurement data of the target electrode is valid data.

[0342] Furthermore, for example, when valid data determination is performed using the above-described determination method 3, the valid data determination and extraction circuit 102 calculates a random value Ra for each sample of the measurement data of each electrode and compares it with a predetermined determination threshold value.

[0343] That is, the valid data determination and extraction circuit 102 calculates the random value Ra by performing the calculation of the above-mentioned formula (3) based on the sampling data of each sample in the electrode measurement data during a predetermined time interval including the sample to be determined. In other words, the circuit calculates the degree of randomness of the measurement data during a predetermined time interval including sampling data (measurement data) obtained by measuring the same electrode at multiple different timings (times). If the random value Ra is equal to or greater than a predetermined determination threshold, the valid data determination and extraction circuit 102 determines that the sampling data of the sample to be determined in the measurement data is valid data.

[0344] The valid data determination and extraction circuit 102 may determine valid data using a method other than the above-described determination methods 1 to 3.

[0345] The valid data determination and extraction circuit 102's determination of the valid data of the measurement data of each electrode can be said to be the identification of a region of interest spatially and temporally, that is, the identification of a region of interest at a certain time on the electrode array made up of the multiple electrodes that make up the electrode sensor 181. The valid data determination and extraction circuit 102 can identify a region of interest with high accuracy and in a short time using determination methods 1 to 3, that is, can determine whether the data is valid.

[0346] In step S203, the valid data determination and extraction circuit 102 extracts valid data from the measurement data of each electrode based on the result of the valid data determination in step S202.

[0347] For example, the valid data determination and extraction circuit 102 leaves the data (sampling data) of the sections of the measurement data for each electrode that were determined to be valid in step S202 as is, and replaces the data of the sections that were determined to be invalid in step S202 with zero data. This results in valid measurement data that includes only valid data and in which the invalid sections are set to zero data. The valid data determination and extraction circuit 102 supplies the obtained valid measurement data for each electrode to the data format conversion circuit 103.

[0348] In step S204, the data format conversion circuit 103 performs data format conversion on the valid measurement data of each electrode supplied from the valid data determination and extraction circuit 102. That is, in step S204, the valid measurement data, i.e., the measurement data obtained by measurement, is converted into output measurement data in a data format different from that of the valid measurement data (measurement data).

[0349] For example, the data format conversion circuit 103 converts the effective measurement data of each electrode into output measurement data of a fixed data size using the above-mentioned conversion method 1.

[0350] That is, the data format conversion circuit 103 extracts sampling data of one or more samples constituting the valid interval from the valid measurement data for each electrode, and adds electrode coordinates indicating the arrangement position of the electrode and a timestamp indicating the start time of the valid interval to the extracted sampling data group to create one set of data.The data format conversion circuit 103 combines all sets of data obtained from the valid measurement data of each electrode and pads the combined data with zeros as appropriate to create one output measurement data of a fixed data size.

[0351] Furthermore, for example, the data format conversion circuit 103 converts the effective measurement data of each electrode into output measurement data for each electrode using the conversion method 2 described above.

[0352] That is, for the valid measurement data of each electrode, the data format conversion circuit 103 identifies an invalid section consisting of invalid data in the valid measurement data and the length (width) of the invalid section. The data format conversion circuit 103 adds a determination bit with a value of 0 to sampling data (actual data) that is valid data in the valid measurement data. The data format conversion circuit 103 also replaces the sampling data that is invalid data in the valid measurement data, more specifically, the portion of the sampling data of one or more samples that constitute the invalid section, with a determination bit with a value of 1 and data indicating the length of the invalid section (the number of invalid data). This results in output measurement data in a data format having a determination bit and a data portion as data for each sample, as shown in FIG. 20, for example.

[0353] Furthermore, for example, the data format conversion circuit 103 converts the effective measurement data of each electrode into output measurement data for each electrode using the conversion method 3 described above.

[0354] That is, the data format conversion circuit 103 identifies, for the valid measurement data of each electrode, an invalid section consisting of invalid data in the valid measurement data and the length (width) of the invalid section, in the same manner as in conversion method 2.

[0355] When the sampling data of a sample (hereinafter also referred to as a target sample) that constitutes valid measurement data is valid data, the data format conversion circuit 103 determines whether the next sample (hereinafter also referred to as the next sample) of the target sample is a sample in the valid section, i.e., a sample of valid data.

[0356] When the next sample is a sample in the valid section, the data format conversion circuit 103 uses the sampling data of the target sample as the sampling data of the sample of the output measurement data as is, and adds invalid section width data whose value is 0 to the sampling data.

[0357] On the other hand, when the next sample is a sample in an invalid section, the data format conversion circuit 103 uses the sampling data of the target sample as is for the sample of the output measurement data, and adds to the sampling data invalid section width data having a value indicating the length (width) of the invalid section starting from the next sample. At this time, the data format conversion circuit 103 deletes the data of the invalid section starting from the next sample (sampling data of each sample) in the valid measurement data.

[0358] As a result, output measurement data is obtained in a data format having invalid section width data as data for each sample and sampling data (actual data) that is set as valid data, as shown in FIG.

[0359] The data format conversion circuit 103 may perform data format conversion using a method other than the above-described conversion methods 1 to 3. The data format conversion circuit 103 supplies the output measurement data obtained by the data format conversion to the PC I / F 186.

[0360] In step S205, the PC I / F 186 outputs the output measurement data sequentially supplied from the data format conversion circuit 103 to the PC 162. In the PC 162, the output measurement data supplied from the PC I / F 186 is recorded in the storage 189 or displayed on a display or the like. When the output measurement data is output, the measurement process ends.

[0361] In this way, the measurement system 151 determines whether the measurement data from each electrode is valid, extracts valid data, and converts the valid measurement data obtained as a result of the extraction into output measurement data in a different data format. In this way, the amount of data can be easily reduced.

[0362] The measurement process described with reference to FIG. 23 is basically performed simultaneously with the measurement of the action potential of the cell, that is, during the measurement.

[0363] However, the timing of the measurement process is not limited to this, and may be any timing. For example, the action potential of a cell may be measured, the measurement data of each electrode as the measurement result may be temporarily recorded, the measurement process of Fig. 23 may be performed on the recorded measurement data, and the resulting output measurement data may be recorded as the final measurement result.

[0364] 23 may be performed as a preliminary measurement before measuring the action potential of a cell (main measurement) to identify electrodes from which valid data has been extracted, and then in the main measurement, measurement or recording of measurement data may be performed only for the identified electrodes.In addition, any of the first to third embodiments may be combined with the fourth embodiment to perform measurement and record output measurement data.

[0365] <Other Configuration Examples of Measurement System> The configuration of the measurement system 151 is not limited to the configuration shown in Fig. 22, and may be the configurations shown in Fig. 24 or 25. In Fig. 24 or 25, parts corresponding to those in Fig. 22 are denoted by the same reference numerals, and descriptions thereof will be omitted as appropriate.

[0366] The measurement system 151 shown in FIG. 24 includes an MEA system 161 and a PC 162 , and the MEA system 161 includes an MEA device 101 and an FPGA 171 .

[0367] The MEA device 101 also has an electrode sensor 181, an ADC 182, a valid data determination and extraction circuit 102, a data format conversion circuit 103, and a sensor I / F 183. In particular, in the MEA device 101, the valid data determination and extraction circuit 102 and the data format conversion circuit 103 are provided between the ADC 182 and the sensor I / F 183. Furthermore, the FPGA 171 has a sensor I / F 184, a buffer 185, and a PC I / F 186, and the PC 162 has a PC I / F 187 and storage 189.

[0368] 22, the determination of valid data, extraction of valid data, and conversion of data format are performed within the FPGA 171, but in the example shown in Fig. 24, these determination of valid data, extraction of valid data, and conversion of data format are performed within the MEA device 101. In this way, it is possible to reduce the amount of data at an earlier stage in the data flow, and it is possible to reduce the data processing load in subsequent stages.

[0369] The measurement system 151 shown in FIG. 25 includes an MEA system 161 and a PC 162 , and the MEA system 161 includes an MEA device 101 and an FPGA 171 .

[0370] The MEA device 101 also has an electrode sensor 181, an ADC 182, and a sensor I / F 183, and the FPGA 171 has a sensor I / F 184, a buffer 185, and a PC I / F 186. The PC 162 also has a PC I / F 187, a valid data determination and extraction circuit 102, a data format conversion circuit 103, and storage 189. In particular, the PC 162 has the valid data determination and extraction circuit 102 and the data format conversion circuit 103 between the PC I / F 187 and storage 189.

[0371] In the example of Figure 22, valid data determination, valid data extraction, and data format conversion were performed within FPGA 171, but in the example shown in Figure 25, these valid data determination, valid data extraction, and data format conversion are performed within PC 162.

[0372] For example, a processing circuit such as a GPU (Graphics Processing Unit) in the PC 162 can be made to function as the valid data determination and extraction circuit 102 or the data format conversion circuit 103 to perform valid data determination, valid data extraction, and data format conversion. This allows the processing content (data processing method) such as valid data determination to be flexibly changed by operating the PC 162. Therefore, the processing content can be flexibly changed depending on, for example, the type of cell, the data analysis method, the application, etc.

[0373] According to the fourth embodiment described above, in an MEA device such as a CMOS-MEA device having a high density and a large number of electrodes, it is possible to reduce the amount of data output by extracting only data that is valid (useful) for the user from the large amount of data output. This makes it easy to record, store, and manage data. It also makes it easy to analyze the recorded data.

[0374] <Example of Computer Configuration> The above-described series of processes can be executed by hardware or software. When the series of processes is executed by software, the programs constituting the software are installed on a computer. Here, the computer includes a computer built into dedicated hardware, and a general-purpose personal computer, for example, that can execute various functions by installing various programs.

[0375] FIG. 26 is a block diagram showing an example of the hardware configuration of a computer that executes the above-described series of processes by a program.

[0376] In the computer, a CPU 501 , a ROM (Read Only Memory) 502 , and a RAM (Random Access Memory) 503 are interconnected by a bus 504 .

[0377] An input / output interface 505 is further connected to the bus 504. An input unit 506, an output unit 507, a recording unit 508, a communication unit 509, and a drive 510 are connected to the input / output interface 505.

[0378] The input unit 506 includes a keyboard, a mouse, a microphone, an image sensor, etc. The output unit 507 includes a display, a speaker, etc. The recording unit 508 includes a hard disk, a non-volatile memory, etc. The communication unit 509 includes a network interface, etc. The drive 510 drives a removable recording medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory.

[0379] In a computer configured as described above, the CPU 501 loads, for example, a program recorded in the recording unit 508 into the RAM 503 via the input / output interface 505 and the bus 504, and executes the program, thereby performing the above-described series of processes.

[0380] The program executed by the computer (CPU 501) can be provided by being recorded on a removable recording medium 511 such as a package medium, for example. The program can also be provided via a wired or wireless transmission medium such as a local area network, the Internet, or digital satellite broadcasting.

[0381] In a computer, a program can be installed in the recording unit 508 via the input / output interface 505 by inserting a removable recording medium 511 into the drive 510. The program can also be received by the communication unit 509 via a wired or wireless transmission medium and installed in the recording unit 508. Alternatively, the program can be installed in the ROM 502 or the recording unit 508 in advance.

[0382] The program executed by the computer may be a program that processes in chronological order according to the order described in this specification, or may be a program that processes in parallel or at the required timing, such as when called.

[0383] Furthermore, the embodiments of the present technology are not limited to the above-described embodiments, and various modifications are possible within the scope of the gist of the present technology.

[0384] For example, the present technology can be configured as a cloud computing system in which a single function is shared and processed collaboratively by a plurality of devices via a network.

[0385] Furthermore, each step described in the above flowchart can be executed by one device, or can be shared and executed by a plurality of devices.

[0386] Furthermore, when one step includes multiple processes, the multiple processes included in that one step can be executed by one device or can be shared and executed by multiple devices.

[0387] Furthermore, the present technology can also be configured as follows.

[0388] (1) A measurement system including a control unit that identifies a region of interest for each of a plurality of electrodes of an MEA placed in a well in which cells are cultured by performing processing based on measurement information other than measurement information obtained by measurement of the electrode itself, and performs a new measurement using only the electrodes included in the region of interest, or deletes the measurement information of electrodes other than those included in the region of interest from the measurement information of the plurality of electrodes. (2) The measurement system described in (1), in which the region of interest is an area occupied by the cell. (3) The measurement system described in (1) or (2), in which the other measurement information is information related to a microscopic image of the cell. (4) The measurement system described in (3), in which the control unit identifies the region of interest by determining whether the cell is present on the electrode based on the position of the cell in the microscopic image. (5) The measurement system according to (4), wherein the control unit causes measurement to be performed using a plurality of the electrodes, causes the microscope image to be taken of the cell stained after measurement as a subject, and deletes the measurement information of the electrodes other than those included in the region of interest from the measurement information of the plurality of electrodes. (6) The measurement system according to (3), wherein the control unit identifies the region of interest based on a model for identifying the region of interest from the measurement information, the model being generated in advance by learning based on the measurement information of the plurality of electrodes and information related to the microscope image of the cell. (7) The measurement system according to (6), wherein the control unit causes measurement to be performed using a plurality of the electrodes as preliminary measurement, identifies the region of interest based on the measurement information of the plurality of electrodes obtained by the preliminary measurement and the model, and causes main measurement to be performed using only the electrodes included in the region of interest. (8) The measurement system according to (7), wherein the time period for the preliminary measurement is shorter than the time period for the main measurement. (9) The measurement system according to (7) or (8), wherein the model is generated by the learning based on the measurement information of the plurality of electrodes obtained by measurement in a state where no compound is administered to the well and information on a microscopic image of the cell stained after measurement.(10) The measurement system according to (9), wherein the preliminary measurement is performed in a state where the compound has not been administered to the well, and the main measurement is performed in a state where the compound has been administered to the well. (11) The measurement system according to any one of (6) to (10), wherein the control unit performs the learning and generates the model. (12) The measurement system according to (1) or (2), wherein the processing is processing of comparing the measurement information of the electrode with the measurement information of another electrode different from the electrode. (13) The measurement system according to (12), wherein the processing is processing of identifying whether the cell is present on the electrode based on similarity of potential waveforms when the cell is electrically active. (14) The measurement system according to (12) or (13), wherein the processing is processing of identifying whether the cell is present on the electrode based on similarity of noise waveforms when the cell is not electrically active. (15) The measurement system according to any one of (12) to (14), wherein the control unit determines whether the cell is present on the electrode based on a similarity between the measurement information of the electrode and the measurement information of the other electrode. (16) The measurement system according to (15), wherein the similarity is cosine similarity. (17) The control unit causes measurements to be performed using a plurality of the electrodes as preliminary measurements, determines whether the cell is present on the electrode based on the measurement information of the electrode and the measurement information of the other electrode obtained by the preliminary measurements, thereby specifying the region of interest, and causes a new main measurement to be performed using only the electrodes included in the region of interest. (18) The measurement system according to (17), wherein the time period for the preliminary measurements is shorter than the time period for the main measurement. (19) The measurement system according to (17) or (18), wherein the preliminary measurement is performed in a state where no compound is administered to the well, and the main measurement is performed in a state where the compound is administered to the well.(20) A measurement method in which a measurement system identifies a region of interest by performing processing based on measurement information other than measurement information obtained by measurement of each of multiple electrodes of an MEA placed in a well in which cells are cultured, and either performs a new measurement using only the electrodes included in the region of interest, or deletes the measurement information of electrodes other than those included in the region of interest from the measurement information of the multiple electrodes. (21) A measurement system comprising a determination unit that determines whether first measurement data obtained by measurement of a first electrode among multiple electrodes constituting an MEA and one or more second measurement data related to the first measurement data are valid data. (22) The measurement system described in (21), in which the determination unit makes the determination based on a similarity between the first measurement data of the first electrode and the second measurement data of a second electrode located near the first electrode. (23) The measurement system described in (22), in which the similarity is cosine similarity. (24) The measurement system according to (21), wherein the determination unit calculates a difference value between the first measurement data of the first electrode and the second measurement data of the second electrode for each of a plurality of second electrodes adjacent to the first electrode, and makes the determination based on the difference value between the plurality of second electrodes. (25) The measurement system according to (21), wherein the determination unit makes the determination based on a degree of randomness of measurement data for a predetermined time interval including the first measurement data of the first electrode and the second measurement data of the first electrode obtained by measurement at a different timing than the first measurement data. (26) The measurement system according to any one of (21) to (25), further comprising a conversion unit that converts the first measurement data into output measurement data having a data format different from that of the first measurement data based on the result of the determination. (27) The measurement system according to (26), wherein the conversion unit converts valid measurement data obtained by extracting valid data from the first measurement data into the output measurement data.(28) The measurement system according to (26) or (27), wherein the output measurement data includes information identifying the first electrode at which the valid data was measured, a timestamp indicating the time of the valid data, and the valid data. (29) The measurement system according to (28), wherein the output measurement data is a data set consisting of information identifying the first electrode, the timestamp, and the valid data, and includes data sets for a plurality of the first electrodes, or data sets for a plurality of different time intervals for the same first electrode. (30) The measurement system according to (28) or (29), wherein the output measurement data is data of a fixed data size. (31) The measurement system according to (26) or (27), wherein the output measurement data includes a sample data portion and determination data indicating whether the sample data portion is the valid data or data indicating the width of a section determined to be invalid data in the first measurement data. (32) The measurement system according to (26) or (27), wherein the output measurement data includes the valid data of a sample and data indicating a width of a section of the first measurement data that is determined to be invalid, the section starting from the sample next to the sample. (33) An information processing method including: a measurement system determining whether the first measurement data is valid data based on first measurement data obtained by measurement at a first electrode among multiple electrodes that make up an MEA, and one or more second measurement data related to the first measurement data.

[0389] 11 Measurement system, 21 MEA chip, 22 Readout circuit board, 23 FPGA, 24 Measurement PC, 25 Microscope, 26 Well, 31 Control unit, 32 Storage, 102 Valid data determination and extraction circuit, 103 Data format conversion circuit, 151 Measurement system

Claims

1. A measurement system comprising a control unit that identifies a region of interest by performing processing based on measurement information other than that obtained by measurement using each of a plurality of electrodes of an MEA placed in a well in which cells are cultured, and either performs a new measurement using only the electrodes included in the region of interest, or deletes the measurement information of electrodes other than those included in the region of interest from the measurement information of the plurality of electrodes.

2. The measurement system according to claim 1, wherein the region of interest is an area occupied by the cell.

3. The measurement system according to claim 1, wherein the other measurement information is information relating to a microscopic image of the cell.

4. The measurement system according to claim 3, wherein the control unit identifies the region of interest by determining whether the cell is present on the electrode based on the position of the cell in the microscope image.

5. The measurement system according to claim 4, wherein the control unit performs measurements using a plurality of the electrodes, takes a microscopic image of the stained cells as a subject after the measurements, and deletes the measurement information of the electrodes other than those included within the region of interest from the measurement information of the plurality of the electrodes.

6. The measurement system of claim 3, wherein the control unit identifies the region of interest based on a model for identifying the region of interest from the measurement information, which model has been generated in advance by learning based on the measurement information of the plurality of electrodes and information related to the microscopic image of the cell, and on the measurement information of the plurality of electrodes.

7. The measurement system according to claim 6, wherein the control unit performs a measurement using a plurality of the electrodes as a preliminary measurement, identifies the region of interest based on the measurement information of the plurality of electrodes obtained by the preliminary measurement and the model, and performs the main measurement using only the electrodes included in the region of interest.

8. The measurement system according to claim 7, wherein the time for the preliminary measurement is shorter than the time for the main measurement.

9. The measurement system of claim 7, wherein the model is generated by learning based on the measurement information of the plurality of electrodes obtained by measurement when no compound is administered to the well and information on microscopic images of the cells stained after measurement.

10. The measurement system according to claim 9, wherein the preliminary measurement is performed in a state where the compound has not been administered to the well, and the main measurement is performed in a state where the compound has been administered to the well.

11. The measurement system according to claim 6, wherein the control unit performs the learning and generates the model.

12. The measurement system according to claim 1, wherein the processing is processing for comparing the measurement information of the electrode with the measurement information of another electrode different from the electrode.

13. The measurement system according to claim 12, wherein the processing is a processing for identifying whether the cell is present on the electrode based on the similarity of the potential waveform when the cell is electrically active.

14. The measurement system according to claim 12, wherein the processing identifies whether the cell is present on the electrode based on the similarity of a noise waveform when the cell is not electrically active.

15. The measurement system according to claim 12, wherein the control unit determines whether the cell is present on the electrode based on the similarity between the measurement information of the electrode and the measurement information of the other electrode.

16. The measurement system according to claim 15, wherein the similarity is a cosine similarity.

17. The measurement system of claim 12, wherein the control unit performs a measurement using a plurality of the electrodes as a preliminary measurement, identifies the region of interest by determining whether the cells are present on the electrodes based on the measurement information of the electrodes and the measurement information of the other electrodes obtained by the preliminary measurement, and performs a new main measurement using only the electrodes included in the region of interest.

18. The measurement system according to claim 17, wherein the time for the preliminary measurement is shorter than the time for the main measurement.

19. The measurement system according to claim 17, wherein the preliminary measurement is performed in a state where no compound is administered to the well, and the main measurement is performed in a state where the compound is administered to the well.

20. A measurement method in which a measurement system identifies a region of interest by performing processing based on measurement information other than that obtained by measurement using the electrodes themselves for each of a plurality of electrodes of an MEA placed in a well in which cells are cultured, and either performs a new measurement using only the electrodes included in the region of interest, or deletes the measurement information of electrodes other than those included in the region of interest from the measurement information of the plurality of electrodes.

21. A measurement system comprising a determination unit that determines whether first measurement data obtained by measurement at a first electrode among multiple electrodes that make up an MEA is valid data based on the first measurement data and one or more second measurement data related to the first measurement data.

22. The measurement system according to claim 21, wherein the determination unit makes the determination based on the similarity between the first measurement data of the first electrode and the second measurement data of a second electrode located in the vicinity of the first electrode.

23. The measurement system according to claim 22, wherein the similarity is a cosine similarity.

24. The measurement system described in claim 21, wherein the determination unit calculates a difference value between the first measurement data of the first electrode and the second measurement data of the second electrode for each of a plurality of second electrodes adjacent to the first electrode, and makes the determination based on the difference value between the plurality of second electrodes.

25. The measurement system described in claim 21, wherein the judgment unit makes the judgment based on the degree of randomness of measurement data in a predetermined time interval including the first measurement data of the first electrode and the second measurement data of the first electrode obtained by measurement at a different timing than the first measurement data.

26. The measurement system according to claim 21, further comprising a conversion unit that converts the first measurement data into output measurement data having a data format different from that of the first measurement data, based on the result of the determination.

27. The measurement system according to claim 26, wherein the conversion unit converts effective measurement data obtained by extracting the effective data from the first measurement data into the output measurement data.

28. The measurement system of claim 26, wherein the output measurement data includes information identifying the first electrode from which the valid data was measured, a timestamp indicating the time of the valid data, and the valid data.

29. The measurement system of claim 28, wherein the output measurement data is a data set consisting of information identifying the first electrode, the timestamp, and the valid data, and includes a data set for each of a plurality of the first electrodes, or a data set for each of a plurality of different time intervals for the same first electrode.

30. The measurement system of claim 28, wherein the output measurement data is data of a fixed data size.

31. The measurement system according to claim 26, wherein the output measurement data includes a sample data portion and determination data indicating whether the sample data portion is the valid data or data indicating the width of a section determined to be invalid data in the first measurement data.

32. The measurement system according to claim 26, wherein the output measurement data includes the valid data of a sample and data indicating the width of a section of invalid data beginning with the sample next to the valid data in the first measurement data.

33. An information processing method including: a measurement system determining whether first measurement data obtained by measurement at a first electrode among multiple electrodes that make up an MEA is valid data based on the first measurement data and one or more second measurement data related to the first measurement data.

Images