Data processing method

Abstract

This invention provides a data processing method for measurement data from a multi-electrode array, which appropriately compresses digital data from multiple channels that are simultaneously measured at multiple points using a multi-electrode array device. [Solution] A method for obtaining compressed data by performing a decimation step and a compression step on input data, wherein the decimation step includes a peak point extraction step that extracts only the peak points of the input data. The compression step generates compressed data by changing the data format of the decimated data of multiple channels generated in the decimation step so that the offset byte count for each channel and the channel data for each channel are arranged in that order. Each channel data contains, in order, the number of skipped points for all data points of each channel and the difference in data values ​​for all data points.

Description

Technical Field

[0001] The present invention relates to a data processing method for measurement data obtained by a multi-electrode array.

Background Art

[0002] In the field of electrophysiology, analysis of the behavior of cellular ion channels in individual or populations of cells is performed. As ion channel analysis apparatuses, for example, intracellular action potential detection apparatuses using the patch clamp method and extracellular action potential detection apparatuses using a multi-electrode array (MEA: Multi Electrode Array) device are known. A multi-electrode array device is disclosed in, for example, Patent Document 1.

Prior Art Literature

Patent Literature

[0003]

Patent Document 1

Summary of the Invention

Problem to be Solved by the Invention

[0004] Using an MEA device, current values or voltage values can be simultaneously measured for each of a large number of electrodes. By utilizing this property, for example, multi-point simultaneous measurement of extracellular potentials caused by the activity of nerve cells, cardiomyocytes, and the like can be performed.

[0005] In recent years, with the improvement of device miniaturization technology, new MEA devices called CMOS-MEA, which have more electrodes than conventional MEA devices or have 1000 units of ultra-fine electrodes, have been developed. As described above, when the number of electrodes increases, when simultaneous measurement data is obtained through multiple channels, the amount of data becomes enormous, which causes the problem that an enormous data capacity is required for data storage.

[0006] To address these problems, a common method to reduce data size is to use a low-pass filter to thin the data. However, thinning the data to a sufficient level may result in the loss of important signal elements. While there are lossless compression methods such as differential methods that compress data while preserving important elements, they can only reduce the size by about 50% at most. Therefore, a new data compression method specifically tailored to electrical signals acquired by MEA devices is needed.

[0007] The objective of the present invention is to provide a technology that can appropriately compress digital data from multiple channels measured simultaneously at multiple points using a multi-electrode array device. [Means for solving the problem]

[0008] To solve the above problems, the first invention of the present application is a data processing method for time-series digital data of multiple channels measured simultaneously at multiple points, which is measurement data of a multi-electrode array, comprising: a) a decimation step of performing a decimation process on each of the input data for all channels to create decimated data; b) a compression step of changing the format of the decimated data to generate compressed data; and c) a storage step of storing the compressed data in a storage unit, wherein step a) includes a1) a peak extraction step of extracting only the peak points of the time-series digital data, and in step b), the compressed data is arranged in the order of offset byte count indicating the data start point for all the channels and channel data for all the channels, and the channel data is arranged in the order of skipped points from the previous data point for all the data points and the difference in data value between all the data points and the previous data point.

[0009] The second invention of the present application is a data processing method of the first invention, wherein step a) further includes a2) a continuous point thinning step of thinning out locations in the data generated in step a1) where multiple peak points appear within a predetermined period.

[0010] The third invention of this application is a data processing method of the second invention, wherein in step a2), if the peak points to be thinned out are consecutive within a predetermined period, the consecutive peak points are not thinned out.

[0011] The fourth invention of this application is a data processing method of the third invention, wherein step a) further includes a3) a near-zero decimation step of decimating the data generated in step a2) such that the peak points have an absolute value less than or equal to a predetermined threshold.

[0012] The fifth invention of this application is a data processing method of the fourth invention, wherein the threshold used in step a3) is a constant multiple of the standard deviation of the first predetermined number of data values ​​of the time-series digital data. [Effects of the Invention]

[0013] According to the first to fifth inventions of this application, digital data from multiple channels measured simultaneously at multiple points using a multi-electrode array device can be appropriately compressed without losing important signal elements.

[0014] In particular, according to the second invention, data can be appropriately and further compressed.

[0015] In particular, according to the third invention, it is possible to suppress the loss of important signal elements of the data.

[0016] In particular, according to the fourth and fifth inventions, data can be appropriately and further compressed. [Brief explanation of the drawing]

[0017] [Figure 1] This figure schematically shows the configuration of a measuring device and a data processing device according to one embodiment. [Figure 2] This is a functional block diagram of a data processing device according to one embodiment. [Figure 3] This is a flowchart illustrating the data processing flow according to one embodiment. [Figure 4] It is a flowchart showing a flow of a thinning process according to an embodiment. [Figure 5] It is a diagram showing an example of first data, second data and thinned data according to an embodiment. [Figure 6] It is a diagram showing an example of first data, second data and thinned data according to an embodiment. [Figure 7] It is a diagram showing an example of a data format of measurement data according to an embodiment. [Figure 8] It is a diagram showing an example of a data format of first data according to an embodiment. [Figure 9] It is a diagram showing an example of a data format of second data and thinned data according to an embodiment. [Figure 10] It is a diagram showing an example of a data format of compressed data according to an embodiment. [Figure 11] It is a functional block diagram of a data processing apparatus according to a modification. [Figure 12] It is a flowchart showing a flow of a thinning process according to a modification. DETAILED DESCRIPTION OF THE INVENTION

[0018] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. It should be noted that the constituent elements described in this embodiment are merely illustrative, and are not intended to limit the scope of the present invention only thereto. In the drawings, for ease of understanding, the dimensions and numbers of respective parts may be exaggerated or simplified as necessary.

[0019] <1. Measurement apparatus using MEA devices> First, as an example of a measurement apparatus for acquiring time-series digital data subjected to multipoint simultaneous measurement which is a target to which the data processing method according to an embodiment is applied, the configuration of a measurement apparatus 1 will be described with reference to FIG. 1. FIG. 1 is a diagram schematically showing the configuration of the measurement apparatus 1. As shown in FIG. 1, the measurement apparatus 1 includes a measurement container 10 and a measurement unit 20.

[0020] The measurement container 10 is a container for measuring the extracellular potential, extracellular voltage, or impedance of the target object, which is cell 9. A cell suspension containing cell 9 is dropped into the measurement container 10. As shown in Figure 1, the measurement container 10 has a bottom 11 and side walls 13. The bottom 11 spreads out in a disc shape along the horizontal plane. The side walls 13 extend upward in a cylindrical shape from the periphery of the bottom 11. The upper surface of the bottom 11 is the bottom surface inside the measurement container 10 into which the cell suspension is dropped. Note that the object to be measured in the measurement container 10 is not limited to cell 9, but may also be a slice of biological tissue or the like.

[0021] The measuring container 10 has a plurality of first electrodes 31, a second electrode 32, and a third electrode 33. The plurality of first electrodes 31 and the third electrode 33 are located on the upper surface of the bottom 11 (the bottom surface of the measuring container 10). Wiring connecting each first electrode 31 and third electrode 33 to an electrode that can be connected to the outside is also located on the upper surface of the bottom 11 (the bottom surface of the measuring container 10). These wires are covered with an insulator (for example, photosensitive polyimide). Each first electrode 31, third electrode 33, and the above-mentioned wiring are formed, for example, by photolithography.

[0022] Multiple first electrodes 31 are arranged in an array. The measuring container 10 is a multi-electrode array (or micro-electrode array) device in which multiple minute first electrodes 31 are arranged in an array. In the example shown in Figure 1, 16 first electrodes 31 are arranged in a 4x4 matrix. The number and arrangement of the first electrodes 31 can be set arbitrarily. Preferably, 10 or more first electrodes 31 are arranged. The shape of the first electrode 31 is square when viewed from above. However, the shape of the first electrode 31 may be a polygon or a circle other than a square.

[0023] The second electrode 32 is located inside the measuring container 10. The second electrode 32 is, for example, stick-shaped, and at least a portion of it (for example, the lower end) is immersed in the liquid 91 injected into the measuring container 10.

[0024] The third electrode 33 is located radially outward from the plurality of first electrodes 31. The plurality of first electrodes 31 and the third electrode 33 are insulated from each other. In the examples shown in Figures 1 and 2, the third electrode 33 is square in top view. However, the shape of the third electrode 33 may be a polygon or a circle other than a square.

[0025] The measurement unit 20 has a plurality of first wires 21 that are electrically connected to each of the first electrodes 31, and a second wire 22 that is electrically connected to the third electrode 33. The measurement unit 20 can obtain electrical measurement values ​​such as current value, voltage value, or impedance for each first electrode 31 as time-series digital data from the plurality of first wires 21 and second wires 22. In the example in Figure 1, 16 simultaneous measurement data, input data D1 to D16, which are time-series digital data, are obtained for each of the 16 first electrodes 31.

[0026] <2. Data Processing Devices> Next, a data processing device 5 according to one embodiment will be described with reference to Figures 1 and 2. Figure 2 is a functional block diagram of the data processing device 5.

[0027] As shown in Figure 1, the data processing device 5 comprises a computer main unit 51, a display unit 52, and an input unit 53. The computer main unit 51 includes a processor 511 such as a CPU, memory 512 such as RAM, and a storage unit 513 such as a hard disk drive. The display unit 52 displays graphs, images, etc., output from the computer main unit 51. The display unit 52 is, for example, a liquid crystal display such as a PC monitor. The input unit 53 can receive commands to the computer main unit 51. The input unit 53 is, for example, a keyboard and mouse. The display unit 52 and the input unit 53 may be an integrated unit such as a touch panel.

[0028] The memory 512 and storage unit 513 are connected to the processor 511 via bus wiring (not shown). The storage unit 513 stores the computer program P. The processor 511 loads the computer program P stored in the storage unit 513 into the memory 512 and executes the code contained in the computer program P sequentially. In this way, the data processing device 5 performs data processing on multiple time-series digital data inputs.

[0029] Computer program P is application software that causes the computer main unit 51 to execute various processes related to the image processing described above. Computer program P is read from a storage medium M such as a CD or DVD and installed on the computer main unit 51. However, computer program P may also be downloaded to the computer main unit 51 via a network N such as the internet.

[0030] Multiple time-series digital data, D1 to D16, measured by the measuring device 1, are input to the data processing device 5. The data processing device 5 then performs a compression process on the input data D1 to D16 to generate compressed data C.

[0031] As shown in Figure 2, the computer body 51 has a data processing unit 60 as a processing unit implemented in software. The data processing unit 60 has a decimation processing unit 70 and a compression processing unit 80.

[0032] In the following explanation, we will assume that each of the 16 input data channels D1 to D16 is digital data consisting of 10,000 values ​​acquired over a period of 500ms at a sampling rate of 20,000Hz. Note that the number of channels, i.e., the number of data measured simultaneously, is not limited to 16 channels but can be any number. For example, the number of channels could be 384, 5000, or even 30,000 or more.

[0033] The decimation processing unit 70 performs decimation on each of the input data D1 to D16 to generate processed decimated data T1 to T16.

[0034] The decimation processing unit 70 includes an initial processing unit 71, a peak extraction unit 72, and a continuous point decimation unit 73. The initial processing unit 71 performs numbering on each data point of the input data D1 to D16 to generate first data Da1 to Da16. The peak extraction unit 72 performs a peak extraction process on the first data Da1 to Da16 to extract only peak points and generates second data Db1 to Db16. The continuous point decimation unit 73 performs a continuous point decimation process on the second data Db1 to Db16 to decimate consecutive peak points and generate decimated data T1 to T16.

[0035] The compression processing unit 80 generates compressed data C by changing the data format of the decimated data T1 to T16, and stores the compressed data C in the data storage unit 600. The data storage unit 600 may be the storage unit 513 described above, or it may be a separately provided storage unit. Alternatively, the data storage unit 600 may be an external storage medium connected to the computer main unit 51.

[0036] Details of the processes performed by the thinning processing unit 70 and the compression processing unit 80 will be described later.

[0037] <3. Data Processing Flow> Next, the data processing flow performed by the data processing device 5 will be explained with reference to Figure 3. Figure 3 is a flowchart showing the data processing flow. In the following, for cases where the same processing is performed on all 16 input data D1 to D16, only input data D1 will be explained as a representative example.

[0038] As shown in Figure 3, first, the data processing unit 60 acquires input data D1 to D16, which are time-series digital data, as input data to be processed (step S1). Specifically, the data processing unit 60 receives input data D1 to D16 from the measuring device 1. At this time, the acquired input data D1 to D16 may be temporarily stored in the data storage unit 600, or the data may be processed directly by the decimation processing unit 70.

[0039] Then, the decimation processing unit 70 performs decimation on each of the input data D1 to D16 (Step S2: Decimation Process). Figure 4 is a flowchart showing the detailed flow of the decimation process S2. In the decimation process S2, as shown in Figure 4, the initial processing unit 71 of the decimation processing unit 70 first performs initial processing on each of the input data D1 to D16 to generate the first data Da1 to Da16 (Step S201: Initial Processing Process).

[0040] Specifically, if the input data D1 is data consisting only of measurement data representing measured values, without a time or a value corresponding to time, the initial processing unit 71 first assigns a number corresponding to time, 1, 2, 3...10,000, to each measurement value. Therefore, if each input data D1 to D16 is digital data consisting of 10,000 values, the first data Da1 after numbering will each consist of 2 × 10,000 values. Hereafter, the newly assigned numbering value will be referred to as the "number," the value of the original input data D1 as the "data value," and the corresponding number and data value together will be referred to as the "data set."

[0041] Next, the peak extraction unit 72 performs peak point extraction processing on each of the first data Da1 to Da16 to generate second data Db1 to Db2 (Step S202: Peak point extraction step).

[0042] Specifically, in the first data set Da1, only the data sets where the data value is the peak value are extracted and designated as the second data set Db1. Note that "peak value" means that the data values ​​immediately before and after the data value are all smaller than that data value, or that the data values ​​immediately before and after the data value are all larger than that data value.

[0043] Figure 5 shows examples of the first data set Da1, the second data set Db1, and the decimated data set T1. The top row (a) of Figure 5 shows an example of the first data set Da1, the middle row (b) shows an example of the second data set Db1, and the bottom row (c) shows an example of the decimated data set T1. In all of Figures 5 (a) to (c), the horizontal axis is the number and the vertical axis is the data value. Each data point that makes up one dataset is shown as a single point (circular marker). Comparing the first data set Da1 shown in Figure 5 (a) with the second data set Db1 shown in Figure 5 (b), it can be seen that although the second data set Db1 has fewer data points than the first data set Da1, it retains the waveform characteristics of the first data set Da1.

[0044] Next, the continuous point thinning unit 73 thins out the second data Db1 to Db16 where multiple peak points appear within a predetermined period, thereby generating thinned data T1 to T16 (Step S203: Continuous point thinning process).

[0045] Specifically, the consecutive point thinning unit 73 first extracts locations in the second data Db1 where peak points appear consecutively in a short period of time as consecutive points. That is, it extracts locations where multiple peak points appear within a predetermined period as consecutive points. For example, if the predetermined period is set to 2 sampling periods, it extracts locations where the next peak point appears within 2 sampling periods after a certain peak point, i.e., locations where the difference between the numbers of two adjacent peak points is 2 or less as consecutive points.

[0046] Here, Figure 6 shows an example of the first data Da1, the second data Db1, and the data obtained by thinning out all consecutive points from the second data Db1, in a case where consecutive points continue in the second data Db1. The top row (a) of Figure 6 shows the first data Da1, the middle row (b) shows the second data Db1, and the bottom row (c) shows the example obtained by thinning out all consecutive points from the second data Db1. In all of Figures 6 (a) to (c), the horizontal axis is the number and the vertical axis is the data value.

[0047] As shown in Figures 6(a) to (c), when comparing the first data Da1 shown in Figure 6(a) with the second data Db1 shown in Figure 6(b), in the example in Figure 6, the second data Db1, which was generated by extracting peak points from the first data Da1, retains the waveform characteristics of the first data Da1. However, when comparing Figures 6(a) and (b) with the data shown in Figure 6(c), the data obtained by downsampling all continuous points from the second data Db1 in Figure 6(c) does not retain any of the waveform characteristics of the first data Da1.

[0048] Thus, simply thinning out consecutive points may cause characteristic peak points to disappear if the consecutive points continue due to the influence of noise or other factors.

[0049] Therefore, in the continuous point thinning process S203, the continuous point thinning unit 73 does not thin out continuous points if there are a predetermined number (for example, 3 points) or more consecutive points. On the other hand, the continuous point thinning unit 73 thins out continuous points if there are not a predetermined number or more consecutive points. In this way, thinned data T1 to T16 are generated.

[0050] Once the thinning process S2 is completed, the compression processing unit 80 then compresses the thinned data T1 to T16, compressing all of the thinned data T1 to T16 to generate a single compressed data C (step S3: compression process). The compression processing unit 80 then stores the compressed data C in the data storage unit 600 (step S4: storage process).

[0051] Here, Figures 7 to 10 show examples of the data formats for each data set. Figure 7 is an example of the data format for the raw measurement data, input data D1 to D16. Figure 8 is an example of the data format for the first data sets Da1 and Da2. Figure 9 is an example of the data format for the second data set Db1 and the decimated data T1. Figure 10 is an example of the data format for the compressed data C. In the examples in Figures 7 to 10, the rectangles that make up the matrices represent data fragments, and an arrow indicating the order of the data is written at the beginning of each row. The data in the examples in Figures 7 to 10 is a single set of data in which the data fragments are stored in order from the top row to the bottom row, in the order of the arrows.

[0052] In Figures 7 to 9, the data fragments written in the format Ch[XX]-[Y,YYY] represent the data value of the [XX]th channel at time [Y,YYY] (not the actual time, but a "number" indicating the data order). Similarly, the data fragments written in the format Nm-[Y,YYY] represent the "number" value itself, indicating time [Y,YYY].

[0053] The input data D1 to D16, which are measurement data, acquires 16 channels of data simultaneously at each sampling time. Therefore, as shown in Figure 7, the data initially stored is the 16 data segments in the first row of Figure 7, specifically the measured values ​​of the first channel Ch01-0,000, the second channel Ch02-0,000, the third channel Ch03-0,000, ... (omitted) ..., and the 16th channel Ch16-0,000 at time 0,000. Next, the 16 data segments of each of the 16 channels at sampling time 0,001, Ch01-0,001, Ch02-0,001, ... (omitted) ..., and Ch16-0,001, are stored. In this way, the input data D1 to D16 are acquired as data blocks containing the same number of data segments as the number of sampling times, with 16 data segments as one set.

[0054] In the initial processing step S201, as shown in Figure 8, the data blocks of input data D1 to D16 are each divided into 16 first data sets Da1 to Da16, and each measured value is numbered. In the example in Figure 8, each dataset, which includes both the number and the data value, is stored sequentially. Specifically, in the first data set Da1 of the first channel, the first two data points listed in the first row are the number Nm-0,000 at time 0,000 and the measured value Ch01-0,000 for the first channel at time 0,000. The next two data points listed in the second row are the number Nm-0,001 at time 0,001 and the measured value Ch01-0,001 for the first channel at time 0,001. The same applies to the first data set Da2 of the second channel.

[0055] In the peak extraction process S202, points other than the extracted peak points are removed, thus reducing the dataset shown as rows in Figure 8. As a result, the number of rows (column length) of the data shown in Figure 8 becomes smaller, as shown in Figure 9. Specifically, in the second data Db1, the points at times 0,001, 0,002, and 9,997 are removed compared to the first data Da1.

[0056] Furthermore, in the example in Figure 9, in the decimated data T1, in which data points have been further reduced in the continuous point decimation process S203, time points 0,000 and 9,998 have been reduced compared to the second data Db1.

[0057] In the examples in Figures 8 and 9, the first data Da1-Da16, the second data Db1-Db16, and the decimated data T1-T16 were stored row by row from top to bottom, and from left to right within each row. That is, the data was stored in the order of number, data value, number, data value... However, the data could also be stored from top to bottom in the left column, followed by top to bottom in the right column, that is, all the numbers in order first, and then all the data values ​​in order.

[0058] When the resulting decimated data T1 to T16 are compressed to obtain compressed data C, the data format shown in Figure 10 is adopted in this embodiment.

[0059] Here, in Figure 10, the data fragments written in the format Ch[XX]-OS are so-called offset bytes, which represent the distance in bytes from the data point where the data for the first channel begins to the data point where the data for the [XX]th channel begins. In the example in Figure 10, the data for the first channel begins at the beginning of the second line. This position is set to 0. Therefore, the value of Ch01-OS is 0. Then, the data for the second channel begins at the beginning of the fourth line. The value of Ch02-OS, which indicates this position, is the number of bytes used in the data for the second and third lines of the first channel. Similarly, the value of Ch03-OS is the total number of bytes used in the data for the second and third lines of the first channel and the data for the fourth and fifth lines of the second channel.

[0060] Furthermore, in Figure 10, there are data fragments shown in the format Ch[XX]-d[Z,ZZZ] and Ch[XX]-V[Z,ZZZ]. [XX] indicates the [XX]th channel, as described above, and [Z,ZZZ] is a number indicating the order in the decimated data T1 to T16. That is, in the decimated data T1 in Figure 9, the data at time 0:003 is [Z,ZZZ]=0th, and the data at time 0:004 is [Z,ZZZ]=1st.

[0061] Ch[XX]-d[Z,ZZZ] indicates the number of points skipped from the previous data point. In other words, it is a value that indicates how many data points after the time of the previous data point the time of that data point is. For the first data point, it shows the difference from time 0,000. In the example of the decimated data T1 in Figure 9, the first data point is at time 0,003, so the value of Ch01-d0,000 is "0,003". The time of the second data point is 0,004, so the difference from the time of the first data point (0,003) is "0,001", which is the value of Ch01-d0,001.

[0062] Furthermore, Ch[XX]-V[Z,ZZZ] indicates the difference in data value from the previous data point. Note that the first data point contains the data value itself. In the example of the decimated data T1 in Figure 9, Ch01-V0,000 contains the data value itself from Ch01-0,003. Then, Ch02-V0,001 contains the difference value obtained by subtracting the data value of Ch01-0,003 from the data value of Ch01-0,004.

[0063] Thus, in the compressed data C shown in Figure 10, the offset byte counts indicating the data start points for all channels, followed by the channel data for all channels, are arranged in order. That is, in the compressed data, the offset byte counts for the 1st channel, the 2nd channel, ... (omitted) ..., the 16th channel, and so on, are arranged, followed by the channel data for the 1st channel, the 2nd channel, ... (omitted) ..., the 16th channel, and so on, for all 16 channels.

[0064] Each channel's channel data contains, in order, the number of points skipped from the previous data point for all data points, and the difference in data value between each data point and the previous data point. Specifically, the first channel data contains the starting position of the first data point of the decimated data T1, the number of points skipped for the second data point, the number of points skipped for the third data point, ... (omitted) ..., the number of points skipped for the last data point, followed by the data value of the first data point of the decimated data T1, the difference in data value of the second data point, the difference in data value of the third data point, ... (omitted) ..., and the difference in data value of the last data point.

[0065] By creating compressed data C in this format, the data size can be significantly reduced. For example, in the measurement data including input data D1 to D6 shown in Figure 7, the size of each data segment Ch[XX]-[Y,YYY] is 2 bytes. In this case, the data size of the measurement data is 2 × 16 × 10,000 = 320,000 bytes.

[0066] In contrast, in compressed data C, the offset byte Ch[XX]-OS can be 4 bytes, the skipped point Ch[XX]-d[Z,ZZZ] can be 0.5 bytes (i.e., 1 byte in pairs), and the data value difference Ch[XX]-V[Z,ZZZ] can be 1 byte. Note that depending on the data waveform, there may be parts where the data value difference Ch[XX]-V[Z,ZZZ] becomes 2 bytes, but these are few in number.

[0067] If the data difference Ch[XX]-V[Z,ZZZ] is 2 bytes for all data values, and the average number of data points in the downsampled data T1~T16 is 4,000, which is 40% of the original input data D1~D16, then the data capacity of compressed data C will be 4 × 16 + (0.5 + 2) × 4,000 × 16 = 10,064 bytes. In other words, in this case, the data capacity of compressed data C can be compressed to about 32% of the data capacity of the original input data D1~D16.

[0068] In the explanation above, we used the assumption of 16 channels and 10,000 samples, but the number of channels and samples can be any number.

[0069] In this way, by extracting peak points and deleting consecutive peak points, the digital data of electrical signals measured using a multi-electrode array device can be appropriately compressed while suppressing the loss of important signal elements.

[0070] Furthermore, by constructing compressed data C in the above data format, digital data from multiple channels measured simultaneously at multiple points using a multi-electrode array device can be compressed significantly and reversibly.

[0071] <4. Variation> Although embodiments have been described above, the present invention is not limited to the embodiments described above.

[0072] Figure 11 is a functional block diagram of the data processing unit 60A in a modified example. Figure 12 is a flowchart showing the flow of the decimation process performed by the data processing unit 60A.

[0073] As shown in Figure 11, the data processing unit 60A has a decimation processing unit 70A which further has a filtering unit 74A. In this data processing unit 60A, as shown in Figure 12, in the decimation process, first the initial processing unit 71A generates first data Da1 to Da16 and calculates the standard deviations σ1 to σ16 of the input data D1 to D16 (step S201A: initial processing step).

[0074] When calculating the standard deviation σ1 of the input data D1, the initial processing unit 71A calculates the standard deviation σ1 not from all the data values ​​of the input data D1, but from the first n data values ​​to the nth data value. The predetermined number n is a natural number that is set in advance, for example, 500.

[0075] Next, similar to the embodiment described above, the peak extraction unit 72 performs peak point extraction processing on each of the first data Da1 to Da16 to generate second data Db1 to Db2 (step S202: peak point extraction step).

[0076] Next, similar to the embodiment described above, the continuous point thinning unit 73 thins out the second data Db1 to Db16 where multiple peak points appear within a predetermined period to generate thinned data T1 to T16 (step S203: continuous point thinning step).

[0077] Subsequently, the filtering unit 74A performs filtering on the decimated data T1 to T16 to generate second decimated data T'1 to T'16 (step S204A: filtering step).

[0078] Specifically, the filtering unit 74A calculates thresholds α*σ1 to α*σ16 by multiplying the calculated standard deviations σ1 to σ16 by a constant α. Here, the constant α is a predetermined positive value, for example, 1. Note that the constant α may be a value less than 1, such as 0.3 or 0.7, or a value greater than 1, such as 1.2 or 1.6, depending on the waveform characteristics of the acquired input data D1 to D16.

[0079] The filtering unit 74A then removes data points from the downsampled data T1 to T16 whose absolute values ​​are smaller than the thresholds α*σ1 to α*σ16, generating the second downsampled data T'1 to T'16. Subsequently, the compression processing unit 80 compresses the second downsampled data T'1 to T'16 to generate compressed data C'. At this time, the values ​​of the standard deviations σ1 to σ16 may be stored at the end of the compressed data C' and used for subsequent data analysis.

[0080] Thus, in the examples shown in Figures 11 and 12, when analyzing electrical signals measured using a multi-electrode array device, the data size can be appropriately reduced while retaining feature points by downsampling values ​​near zero that are less likely to be feature-rich.

[0081] Although this invention has been described in detail, the above description is illustrative in all respects, and the invention is not limited thereto. It is understood that countless variations not illustrated can be conceived without falling outside the scope of this invention. The components described in each of the above embodiments and variations can be combined or omitted as appropriate, as long as they do not contradict each other. [Explanation of Symbols]

[0082] 1. Measuring device 5. Data Processing Devices 51 Computer main unit 60, 60A Data Processing Unit 70,70A Thinning Processing Unit 71,71A Initial Processing Unit 72 Peak Extraction Section 73. Continuous point thinning section 74A Filtering section 80 Compression Processing Unit 600 Data Storage Unit C,C' compressed data D1~D16 Input Data Da1~Da16 First Data Db1~Db16 Second Data T1~T16 thinned data T'1~T'16 Second Thinning Data α constant σ1~σ16 Standard deviation

Claims

1. A data processing method for time-series digital data of multiple channels measured simultaneously at multiple points, which is measurement data of a multi-electrode array, a) A decimation process that performs decimation on each of the input data for all channels to create decimated data, b) A compression step of generating compressed data by changing the format of the thinned data, c) A storage step of storing the compressed data in the storage unit, It has, The aforementioned step a) is, a1) A peak extraction process that extracts only the peak points of the time-series digital data. Includes, In step b) above, the compressed data is The number of offset bytes indicating the data start point for all of the aforementioned channels, Channel data for all of the aforementioned channels, They are arranged in that order. The channel data mentioned above are, respectively, The number of points skipped from the previous data point for all data points, The difference in data value between all the aforementioned data points and the previous aforementioned data point, The data processing methods are listed in the following order.

2. A data processing method according to claim 1, The aforementioned step a) is, a2) A continuous point thinning process, in which the data generated in step a1) is thinned out in locations where multiple peak points appear within a predetermined period. A data processing method that further includes this.

3. A data processing method according to claim 2, A data processing method in which, in step a2), if the peak points to be thinned out are consecutive within a predetermined period, the consecutive peak points are not thinned out.

4. A data processing method according to claim 3, The aforementioned step a) is, a3) A near-zero decimation step in which the peak points whose absolute value is less than or equal to a predetermined threshold are decimated from the data generated in step a2). A data processing method that further includes this.

5. A data processing method according to claim 4, A data processing method wherein the threshold used in step a3) is a constant multiple of the standard deviation of the first predetermined number of data values ​​of the time-series digital data.

Images