Light detection device and signal processing method

Abstract

This light detection device 500 comprises: an exchangeable capillary array 501 in the interior of which a plurality of types of light-emitting body emit light; an optical sensor 504 which has a plurality of detection channels for detecting the light emissions from the plurality of types of light-emitting body in a respective plurality of wavelength bands; an optical fiber array 510 which fixes the incidence positions of the light emissions from the plurality of types of light-emitting body relative to the optical sensor 504; and a computer 550. The computer 550 has crosstalk information therein, reduces crosstalk existing between the detection channels of the optical sensor 504 by means of a calculation with respect to a plurality of light emission points emitted by the plurality of kinds of light-emitting bodies using the crosstalk information as well as the outputs from the optical sensor 504 corresponding to the (respective) plurality of wavelength bands, derives the concentration ratio or signal amount ratio of each of the plurality of types of light-emitting body for each of a plurality of capillaries, and uses the crosstalk information to perform a calculation even when the capillary array 501 is replaced with another.

Description

[Technical Field]

[0001] The present invention relates to a photodetector and a signal processing method, and more particularly to a photodetector and a signal processing method that reduces crosstalk that occurs when multiple types of light emitters emit light at multiple light-emitting points. [Background technology]

[0002] There are optical measurement methods that detect the location and concentration ratio of multiple types of light emitters present at multiple spatial locations using a photodetector that separates and detects the wavelength of light for each spatial location. One example of such a method is the photodetector used in a capillary electrophoresis apparatus equipped with multiple analytical capillaries to analyze samples labeled with multiple types of fluorescent dyes.

[0003] In capillary electrophoresis analysis, the analyte sample is injected into a capillary filled with a separation medium, and a voltage is applied to both ends to separate the analytes based on the difference in their mobility. For detecting the analytes, for example, fluorescence detection is used. The analytes are labeled with a fluorescent dye, and the separated analytes are detected by detecting the fluorescence generated when irradiated with excitation light.

[0004] One example is a method in which DNA labeled with a fluorescent dye is electrophoresed in a polymer-filled capillary to separate it by chain length. Excitation light is shone onto a detection site on the capillary, and the resulting fluorescence is detected. DNA molecules in the sample move through the capillary, passing through the detection site at different times depending on their chain length. As a result, the chain length distribution of DNA molecules in the sample is obtained as a fluorescence intensity waveform. When multiple types of DNA molecules are present in a sample and need to be distinguished and analyzed, multiple types of fluorescent dyes may be used to label the DNA. The photodetector separates and detects the fluorescence wavelength using a grating or similar device, and identifies the type of fluorescent dye from the shape of the resulting spectrum.

[0005] In capillary electrophoresis systems, multiple capillaries are sometimes installed to improve measurement throughput. In such cases, for example, the separation capillaries are arranged in a row, and excitation light is shone on all capillaries simultaneously. The generated fluorescence is acquired by an image sensor. The signal intensity is calculated from the fluorescence image of each capillary to obtain the electrophoretic waveform of the sample.

[0006] As mentioned above, when using multiple types of fluorescent dyes and multiple capillaries, a signal other than the one that should be detected may be detected, causing the measured concentration ratio of the object to deviate from the true value. For example, when fluorescent dye A is emitting light in a certain capillary, it may be mistakenly identified as fluorescent dye B emitting light, or when a certain i-th capillary is emitting light, it may be mistakenly identified as a different j-th capillary emitting light. Here, the signal output due to misidentification is called crosstalk, and the former, caused by misidentification of the dye, is called spectral crosstalk, and the latter, caused by misidentification of the emitting capillary, is called spatial crosstalk.

[0007] There are various possible causes of crosstalk. For example, spectral crosstalk can occur due to the width of the emission spectrum of fluorescent dyes. Generally, fluorescent dyes have emission spectra with wavelengths of several tens of nanometers. When multiple dyes are used, the emission spectra of dye A and dye B may overlap. If there is overlap between the emission spectra of dye A and dye B, even if dye A is emitting light, the portion of dye A's emission spectrum that falls within the detection wavelength band of dye B may be detected as an emission signal from dye B.

[0008] As an example, spatial crosstalk can occur due to surface reflection from adjacent capillaries. Suppose a device has multiple capillaries arranged in a row, and a detector is positioned facing this capillary array. When a fluorescent dye emits light from the i-th capillary (hereinafter referred to as capillary i), the fluorescence is emitted in all directions, and some of it is introduced into the detector. However, some of the light hits the nearby j-th capillary (hereinafter referred to as capillary j), and some of that light is reflected back towards the detector. In such a case, the detector may perceive capillary j, which has reflected the fluorescence, as emitting fluorescence, and thus detect it as an emission signal from capillary j.

[0009] Crosstalk has various adverse effects in capillary electrophoresis analysis. For example, if spatial crosstalk causes the fluorescence of capillary i to be mistakenly identified as the fluorescence of capillary j, it may be possible to mistakenly believe that components of the sample being analyzed with capillary i were also present in the sample being analyzed with capillary j. Furthermore, spectral crosstalk may cause a misinterpretation, where component b, labeled with fluorescent dye B, is present in the sample even though only component a, labeled with fluorescent dye A, is actually present.

[0010] Crosstalk can be reduced by optimizing measurement conditions and equipment configuration. For example, spectral crosstalk can be reduced by widening the interval between the emission wavelengths of the fluorescent dyes used, and spatial crosstalk can be eliminated by preparing a completely independent measurement optical system for each capillary. However, since there are upper and lower limits to the detection wavelengths that a detector can detect, widening the emission spectral interval of the dyes reduces the number of fluorescent dyes that can be used. Also, while spatial crosstalk can be eliminated by making the optical system for each capillary independent, it requires a light source and detector for each capillary, which is disadvantageous in terms of equipment cost and size.

[0011] On the other hand, a method for reducing crosstalk through data processing is also known (Patent Document 1). In this method, the crosstalk when a certain fluorescent dye emits light through a certain capillary is comprehensively acquired in advance, and the inverse matrix of the matrix containing the crosstalk components is generated. By applying this inverse matrix to the actually acquired signal, the crosstalk is reduced. With this method, it is possible to reduce crosstalk without changing the device configuration or measurement conditions. [Prior art documents] [Patent Documents]

[0012] [Patent Document 1] Patent No. 7282880 [Patent Document 2] International Publication No. 2018 / 151843 [Patent Document 3] International Publication No. 2023 / 276078 [Non-patent literature]

[0013] [Non-Patent Document 1] SeqStudioTMGenetic Analyzer Instrument and Software USER GUIDEhttps: / / assets.thermofisher.com / TFS-Assets / LSG / manuals / MAN0018646_SeqStudioInstSW_UG.pdf [Overview of the Initiative] [Problems that the invention aims to solve]

[0014] The method described in Patent Document 1 assumes that the ratio of spectra and spatial crosstalk between each dye and between capillaries remains unchanged between the time the matrix used for crosstalk reduction processing is generated and the time the sample is analyzed. This condition is basically met when the same capillary array is used for generating the matrix used for crosstalk reduction processing and for analyzing the sample. However, in capillary electrophoresis, the capillary array is a consumable item and needs to be replaced after a certain number of uses.

[0015] Replacing a capillary array alters the ratio of spectral crosstalk and spatial crosstalk. While capillaries are arranged in a single row, each capillary has a positional error of several micrometers to over ten micrometers, meaning the light reflection pattern on the capillary surface can vary from array to array. Furthermore, the position and tilt of the center of the capillary array relative to the detector can also change before and after replacement. These factors alter the ratio of spatial crosstalk.

[0016] Since the emission spectrum of the dye does not change with capillary replacement, the change in spectral crosstalk associated with capillary array replacement is expected to be smaller than that of spatial crosstalk. However, spectral crosstalk can also change with capillary array replacement. For example, if crosstalk occurs in a path where multiple reflections occur within the detector's optical system and the signal of dye A enters the detection channel of dye B, then a change in the position of the capillary array can change the state of multiple reflections, and as a result, change the spectral crosstalk.

[0017] Therefore, when adopting the method described in Patent Document 1, it is necessary to generate a matrix for crosstalk reduction processing for the new array after replacing the capillary array. In generating the matrix for crosstalk reduction processing, it is necessary to determine the crosstalk ratio for all dyes and capillaries. The crosstalk ratio when dye A emits light in a certain capillary i can be determined by injecting only dye A into capillary i and performing electrophoresis. Therefore, for example, when using a capillary array with four types of dyes and eight capillaries, generating the matrix for crosstalk reduction processing will require 4 × 8 = 32 electrophoresis cycles. Considering that one electrophoresis cycle takes several tens of minutes or more, including preparatory operations such as polymer replacement and preliminary electrophoresis to remove unwanted ions, generating the matrix for crosstalk reduction processing will take a long time. This time increases with the number of dyes and capillaries.

[0018] Patent Document 1 also describes a method to reduce the effort and time required for matrix generation used in crosstalk reduction processing. Specifically, it describes a method in which the timing of sample injection into each capillary is staggered to stagger the timing of dye emission, so that when one capillary is emitting light, other capillaries are not. It also describes a method in which fluorescently labeled DNA with different chain lengths is introduced into each capillary, similarly preventing other capillaries from emitting light when one capillary is emitting light.

[0019] However, the former requires the instrument to be equipped with a different operating sequence than that used for normal analysis, and the latter has hurdles such as the need to prepare reagents containing DNA with controlled chain lengths for generating the matrix used in the crosstalk reduction process. In addition, if the dye used is changed, the matrix used in the crosstalk reduction process must be regenerated even if the capillary array is not replaced.

[0020] Patent Document 2 describes a method for reducing spectral crosstalk (equivalent to the process of determining the ratio of multiple dyes with overlapping spectra from fluorescence spectra) without performing electrophoresis for calibration to obtain information on the matrix used to reduce spectral crosstalk. Instead, it optimizes the matrix used to reduce spectral crosstalk based on information obtained from the analysis of the sample during the analysis. This method optimizes the matrix used for reducing spectral crosstalk by repeatedly modifying the matrix based on the correlation information between any two estimated dye concentrations, evaluating the correlation of dye concentrations, and deciding whether or not to update the matrix, all based on the initial matrix.

[0021] In this method, outliers must be removed in order to correctly obtain the correlation between two dyes due to spectral crosstalk. That is, it is necessary to distinguish between spectral crosstalk with dye B when dye A is present and when dye B is actually present in the sample. It is generally expected that this distinction can be made with respect to spectral crosstalk. Generally, when multiple dyes are used, the purpose is to distinguish between different analytes, so dyes A and B label different analytes (molecules). Therefore, when electrophoresis is performed on a sample containing both analytes labeled with dye A and analytes labeled with dye B, and time-series data of the fluorescence spectrum is obtained, it can be expected that dyes A and B will often emit light at different times. On the other hand, spectral crosstalk with dye B caused by dye A occurs at the same time as the emission of dye A. Since it is rare for analytes labeled with dye A and analytes labeled with dye B to emit light by chance at the same time, when calculating the correlation information between the estimated dye concentrations of dyes A and B, it is sufficient to remove such rare values ​​that differ in trend from the majority of other data as outliers.

[0022] However, even if we attempt to extend the method of Patent Document 2 to spatial crosstalk, the above prerequisites regarding outlier removal are not met. When the method of Patent Document 2 is extended to spatial crosstalk, it is necessary to distinguish between the spatial crosstalk of dye A in capillary j when dye A emits light in capillary i and the signal actually emitted by dye A in capillary j. As in the previous case, the fluorescence of dye A due to spatial crosstalk in capillary j occurs at the same timing as when dye A emits light in capillary i. Here, we assume that dye A labels the same analyte α in the sample analyzed in capillary i and the sample analyzed in capillary j. In this case, the timing of the emission of dye A actually contained in the sample in capillary i and capillary j is the same. Thus, there is no guarantee that the spatial crosstalk that occurs with respect to dye A in capillary j when dye A emits light in capillary i can always be distinguished from the signal actually generated by the emission of dye A in capillary j.

[0023] This poses a significant obstacle in situations such as the following. For example, consider analytes α labeled with dye A and β labeled with dye B. Sample 1 contains only analyte α. Sample 2 contains analyte β, and we want to determine whether a trace amount of analyte α is present in Sample 2. This is similar to a case where Sample 1 is a pure substance, and we analyze the trace components (analyte α) present in Sample 2 based on the data from Sample 1.

[0024] Here, sample 1 is analyzed with capillary tube i, and sample 2 is analyzed with capillary tube j. At this time, a strong signal of dye A is obtained with capillary tube i, and a strong signal of dye B is obtained with capillary tube j. We want to determine whether a weak signal of dye A is generated with capillary tube j. However, since dye A is labeling the analyte α in both capillary tubes i and j, the fluorescence of dye A will be obtained at almost the same time whether it is due to spatial crosstalk or the actual signal of the analyte α. Therefore, even if a weak signal of dye A is obtained with capillary tube j, it is not possible to distinguish from the data trend alone whether it is caused by spatial crosstalk or whether a trace amount of analyte α labeled with dye A is actually present.

[0025] Thus, by combining the method of Patent Document 1 with the method of Patent Document 2, it is not possible to optimize the matrix to reduce both spectral and spatial crosstalk during analysis, thereby eliminating the need to reacquire the crosstalk reduction matrix when replacing the capillary array.

[0026] Furthermore, Non-Patent Document 1 describes a capillary electrophoresis apparatus that offers the option of using a pre-configured matrix as the spectral crosstalk reduction matrix (Factory Calibration, p. 217). However, this matrix is ​​not optimized for each individual instrument and can only be used in applications where residual spectral crosstalk due to incomplete information is acceptable.

[0027] Therefore, the present invention has been made in view of these circumstances, and aims to provide a photodetector and a signal processing method that can perform crosstalk reduction processing using crosstalk information used for crosstalk reduction processing that has been generated once, even if multiple light-emitting element holders (e.g., capillary arrays) are replaced. [Means for solving the problem]

[0028] To solve the above problems, the present invention provides a photodetector comprising: a plurality of interchangeable light-emitting holders in which multiple types of light-emitting elements emit light internally; a photosensor having a plurality of detection channels for detecting light emitted from multiple types of light-emitting elements in multiple wavelength bands; a spatial filter for fixing the incident position of light emitted from multiple types of light-emitting elements to the photosensor; and a computer for processing signals output from the photosensor. The computer internally holds crosstalk information for reducing crosstalk between the multiple detection channels, and by performing calculations on the output from the photosensor corresponding to multiple wavelength bands for each of the multiple light-emitting points in which multiple types of light-emitting elements emit light using the crosstalk information, it reduces crosstalk between the detection channels of the photosensor, derives the respective concentration ratio or signal amount ratio of the multiple types of light-emitting elements for each of the multiple light-emitting holders, and performs calculations using the crosstalk information even when a light-emitting holder is replaced with another light-emitting holder.

[0029] Furthermore, the signal processing method of the present invention includes preparing a photodetector comprising: a plurality of interchangeable light-emitting holders in which a plurality of types of light-emitting elements emit light internally; a photosensor having a plurality of detection channels for detecting the light emitted from the plurality of types of light-emitting elements in a plurality of wavelength bands; a spatial filter for fixing the incident position of the light emitted from the plurality of types of light-emitting elements to the photosensor; and a computer for processing the signal output from the photosensor; measuring the light emitted from the plurality of types of light-emitting elements using the photosensor; using crosstalk information stored in the computer to reduce crosstalk between the plurality of detection channels to perform calculations on the output from the photosensor corresponding to the plurality of wavelength bands of each of the plurality of light-emitting points in which the plurality of types of light-emitting elements emit light; reducing the crosstalk between the detection channels of the photosensor; deriving the respective concentration ratio or signal quantity ratio of the plurality of types of light-emitting elements for each of the plurality of light-emitting holders; and performing calculations using the crosstalk information even when the light-emitting holders are replaced. [Effects of the Invention]

[0030] According to the photodetector and signal processing method of the present invention, even if multiple light-emitting element holders are replaced, crosstalk reduction processing can be performed using crosstalk information that has been generated once. Therefore, the effort and time required to regenerate the crosstalk information used for crosstalk reduction processing when multiple light-emitting element holders are replaced can be reduced. Other issues, configurations, and effects not mentioned above will be clarified by the following description of the embodiments. [Brief explanation of the drawing]

[0031] [Figure 1A] This is a diagram of a photodetector that may be subject to the crosstalk reduction method described in Patent Document 1. [Figure 1B] Figure 1A is a schematic diagram of the spectral image on the optical image sensor of the photodetector. [Figure 2A] This is a schematic diagram of a crosstalk simulation model. [Figure 2B] This figure shows a flowchart of the simulation performed using the simulation model in Figure 2A. [Figure 3A] This is a diagram showing the results of a crosstalk simulation. [Figure 3B] This figure shows the results when the crosstalk reduction process described in Patent Document 1 is applied to the simulation results in Figure 3A. [Figure 4A] This figure shows the results of a crosstalk simulation assuming the replacement of a capillary array. [Figure 4B] This figure shows the results when the crosstalk reduction process described in Patent Document 1 is applied to the simulation results in Figure 4A. [Figure 5] This is a schematic diagram of the light detection device according to the first embodiment. [Figure 6] This is a schematic diagram showing the crosstalk generation path that occurs on the capillary side of the optical fiber. [Figure 7A] This figure shows a simulation model of a photodetector to which the crosstalk reduction process according to the first embodiment is applied. [Figure 7B] Figure 7A shows the simulation results of crosstalk in the simulation model. [Figure 7C] Figure 7B shows the results after applying crosstalk reduction processing to the simulation results. [Figure 8A] This is a schematic diagram showing a method for connecting a calibration light source to obtain a matrix used in the crosstalk reduction process according to the first embodiment. [Figure 8B] Figure 8A is a schematic diagram showing an example of the structure of a calibration light source. [Figure 9] This figure shows the effect of detector saturation on the crosstalk reduction process according to the first embodiment. [Figure 10] This diagram shows the components of matrices G and F. [Figure 11] This is a schematic diagram showing an example of the structure of a calibration light source for obtaining a matrix used in the crosstalk reduction process according to the second embodiment. [Figure 12] This flowchart shows the operation of the optical detection device when acquiring the matrix used for crosstalk reduction processing according to the second embodiment. [Figure 13] This is a schematic diagram of a modified photodetector. [Modes for carrying out the invention]

[0032] In the following embodiments, when necessary for convenience, the description will be divided into multiple sections or embodiments. Unless otherwise specified, these are not unrelated, and one may be a modification, detail, or supplementary explanation of part or all of the other. Furthermore, in the following embodiments, when referring to the number of elements (including number, numerical value, quantity, range, etc.), unless otherwise specified or clearly limited to a specific number in principle, it is not limited to that specific number, and may be greater than or less than that number.

[0033] Furthermore, in the following embodiments, it goes without saying that the components (including element steps, etc.) are not necessarily essential unless specifically stated or considered to be clearly essential in principle. Similarly, in the following embodiments, when referring to the shape, positional relationship, etc., of the components, etc., it shall include those that substantially approximate or resemble their shape, etc., unless specifically stated or considered to be not in principle. The same applies to the numerical values ​​and ranges mentioned above.

[0034] In addition, in all the drawings used to explain the embodiments, the same reference numerals are generally used for identical components, and repeated explanations of them are omitted.

[0035] In the following embodiments, the use of the photodetector of the present invention in a capillary electrophoresis apparatus will be described as a typical application example. This is to explain the configuration and effects of the present invention in more detail, and the photodetector of the present invention is not limited to a capillary electrophoresis apparatus.

[0036] (Overview of the capillary electrophoresis apparatus in this embodiment) In the configuration of the capillary electrophoresis apparatus of this embodiment, as an example, a spatial filter is installed in the light intake section of a spectrometer that spectrally separates fluorescence from a fluorescent dye to acquire a spectrum, thereby fixing the position of light intake to the spectrometer. As an example, the spectrometer has a first lens that collimates the incident light, a grating that separates the light by wavelength, a second lens that images the light that has passed through the grating onto a photosensor, and a photonic image sensor that detects the light. As an example, the spatial filter is a multimode optical fiber, with one end of the optical fiber fixed to the light incident position of the spectrometer. The other end of the optical fiber is placed near the capillary to capture the fluorescence emitted from inside the capillary. As an example, the photonic image sensor is a CCD or CMOS image sensor.

[0037] Crosstalk can occur due to multiple factors, but by adopting the aforementioned configuration, the ratio of crosstalk generated on the spectrometer side via the optical fiber can be fixed without being affected by the replacement of the light emitter holder (capillary array). Fluorescence generated within the capillary is captured by the optical fiber and guided to the spectrometer. Possible factors causing crosstalk inside the spectrometer include the spatial broadening of the fluorescence spectrum profile, multiple reflections between element surfaces such as lenses, and grating anomalies. If there were no optical fiber and the capillary was directly installed at the light introduction position of the spectrometer, the positional shift of each capillary due to the replacement of the capillary array could affect the spatial profile of fluorescence, the occurrence of multiple reflections and anomalies, and thus the ratio of crosstalk could change.

[0038] By fixing the incident position of the optical fiber into the spectrometer, the spatial profile of fluorescence and the occurrence of multiple reflections and anomalies are fixed. A change in the position of the capillary relative to the other end of the optical fiber opposite the spectrometer affects the amount of light incident on the optical fiber, but does not affect the ratio of crosstalk generated inside the spectrometer. Because the change in the position of the capillary does not change the ratio of crosstalk, even if the capillary array is replaced, the same information as before the replacement can be used for crosstalk reduction processing.

[0039] On the other hand, crosstalk generated on the capillary side of the optical fiber can still be altered by replacing the capillary array. For example, crosstalk generated by the reflection of fluorescence on the capillary surface is affected by the arrangement error of the capillaries constituting the capillary array, and therefore changes when the capillary array is replaced. Such crosstalk cannot be completely eliminated simply by providing the optical fiber described above. It is possible to sufficiently reduce the crosstalk generated on the capillary side and which fluctuates when the capillary array is replaced. This can be achieved, for example, by separating each capillary with a light-shielding wall to make them independent. Alternatively, the spacing between capillaries can be made large enough that the crosstalk is negligible. A method of limiting the incident angle of the detected light by placing a pinhole at the exit of the optical fiber (Patent Document 3) is also known.

[0040] While optical fibers alone cannot eliminate crosstalk generated on the capillary side and fluctuating with capillary array replacement, this does not diminish the effectiveness of the present invention. Even if crosstalk generated on the capillary side of the optical fiber is not completely eliminated, it is sufficient if the remaining components are low enough to be acceptable for analytical purposes.

[0041] (Regarding the prior art described in Patent Document 1) Prior to a detailed description of the embodiments, the prior art and its problems described in Patent Document 1, which form the background of the present invention, will be summarized. Figure 1A is a diagram of the configuration of a photodetector 100 to which the technology described in Patent Document 1 may be applied. The photodetector 100 comprises a capillary array 101, two lenses 102, a grating 103, and an optical image sensor 104. The photodetector 100 spectrally analyzes and detects fluorescence emitted from the capillary array 101. The arrows in the figure indicate the irradiation direction of the excitation laser light L that excites the fluorescence. The laser light L is irradiated from the side of the capillary array 101 so as to pierce all the capillaries of the capillary array, exciting the dye inside all the capillaries.

[0042] In the photodetector 100, the wavelengths are dispersed perpendicular to the plane of the paper by the grating 103. As shown in Figure 1B, spectral images 105 corresponding to each capillary are arranged on the optical image sensor 104. The optical image sensor 104 outputs the light intensity measurement result for each pixel. The light intensity measurement results are summed up for each wavelength division of each capillary. A wavelength division refers to, for example, if the measurement wavelength range is from 500 nm to 700 nm, and the light intensity values ​​for each wavelength are obtained as 20 data points with a width of 10 nm, then the divisions are 500 nm to 510 nm, 510 nm to 520 nm, ... 690 nm to 700 nm. In other words, a set of values ​​obtained by summing the outputs from pixels in each section (channel 106) separated by the dotted lines in Figure 1B is obtained as spectral information for each capillary.

[0043] The conventional crosstalk reduction method described in Patent Document 1 will be summarized with the photodetector 100 in mind. As previously defined, crosstalk includes two types: spectral crosstalk, which is the misidentification of the signal of dye A as the signal of dye B, and spatial crosstalk, which is the misidentification of the signal emitted by capillary i as the signal emitted by capillary j. Patent Document 1 and the crosstalk reduction method of the present invention assume that this crosstalk is linear with respect to the signal vector (a vector obtained by the optical image sensor 104 by arranging the signals corresponding to each capillary and each wavelength division). This is usually satisfied because the photodetector 100 does not include any elements that exert a nonlinear effect on the light to be detected, and is not used under conditions where a nonlinear effect would appear (generally requiring very strong light intensity).

[0044] The above content is the signal when dye A emits light in capillary i, S Ai Let S be the crosstalk detected in the region where dye B should be detected by capillary j when dye A emits light in capillary i. CT Bj←Ai In this case, the relationship between the two can be expressed by the following equation (1) using the constant c. S CT Bj←Ai =cS Ai (1)

[0045] In the conventional crosstalk reduction method described in Patent Document 1, first, a process of obtaining a constant c is performed. Assume a situation where dye A is injected into capillary i, and nothing is injected into other capillaries. In this situation, the signal S when dye A emits light in capillary i Ai is obtained, and when dye A emits light in capillary i, the crosstalk S detected in the region where dye B should be detected in capillary j CT Bj←Ai is obtained. From the ratio of these two values, the constant c is obtained by the following formula (2). c=S CT Bj←Ai / S Ai (2)

[0046] Next, a sample to be analyzed is measured. A sample labeled with dye A is injected into capillary i. A sample labeled with dye B is injected into capillary j. The signal acquired in capillary i is S Ai It is desirable that the signal S when dye B emits light can be obtained in capillary j Bj , however, since crosstalk S CT Bj←Ai is superimposed on the signal, the actually measured signal S act Bj is calculated by the following formula (3). S act Bj =S Bj +S CT Bj←Ai (3)

[0047] Here, the crosstalk component in formula (3) is represented by formula (1). In formula (1), the constant c is obtained by formula (2). In addition, S Ai is obtained as a signal acquired in capillary i. Therefore, using formulas (1) and (3), the true signal S that should be obtained when there is no crosstalk can be obtained by the following formula (4) Bj . S Bj =S act Bj -S CT Bj←Ai =Sact Bj -cS Ai (4)

[0048] The above description is a simplified version of the method described in Patent Document 1. In reality, crosstalk may also include components generated from capillaries other than capillaries i and j, and from dyes other than dyes A and B. In the above description, fluorescence generated from dyes A and B is referred to as a signal. In reality, the optical image sensor 104 outputs the intensity of light in each wavelength range for each capillary as a signal.

[0049] Based on the above, the method described in Patent Document 1 is summarized below. For example, suppose an L-colored fluorescent dye is used for measurement, there are N capillaries to be measured, and the optical image sensor 104 measures M wavelength intervals. When the wavelength interval is, for example, M=20 and the detection wavelength range is 500~700nm, the optical image sensor 104 outputs 20 signals for each capillary: a fluorescence signal at wavelengths of 500~510nm, a fluorescence signal at wavelengths of 510~520nm, ... a fluorescence signal at wavelengths of 690~700nm. Below, when the l-th dye is introduced into capillary j, the signal (or crosstalk) generated for the wavelength interval k of capillary i is defined as S. ik←jl This is how it is written.

[0050] Here, S ik←jl Let C be a matrix whose components are S. ik←jl It has four types of subscripts, but the ζ and η components of matrix C are C ζη For this, we arrange it in two dimensions such that ζ=Mi+k and η=Mj+l. ik←jl The specific value can be obtained by sequentially passing one type of dye used only in capillary j through the light image sensor 104 and recording it as the output. ik←jl The value of may be appropriately normalized by the signal when the dye used is at a reference concentration. Matrix C is a matrix in which the spectra of each dye used are arranged in the positions corresponding to each capillary in the matrix. Here, prior to measuring the sample to be measured, the generalized inverse matrix C of matrix C is obtained. - This is obtained from equation (6) below. C- =[C t C] -1 C t (6)

[0051] Next, let's consider a scenario where the sample to be analyzed is being analyzed. In this case, the signal of the wavelength segment k of capillary i is S ik Let's assume that the signal vector s is S ik It is constructed by arranging these. Specifically, the zeta component s of the signal vector s ζ For S such that ζ = Mi + k ik Arrange them. At this time, a vector d whose components are the ratio of each pigment can be obtained as shown in equation (7) below. d=C - s (7)

[0052] Equation (7) extends the explanation for removing a single crosstalk component described above to cases where spatial crosstalk exists between multiple capillaries and spectral crosstalk exists between multiple dyes. The above is the conventional crosstalk reduction method described in Patent Document 1. In addition, in analysis by capillary electrophoresis, a vector s is output from the optical image sensor at each time point, so by performing the calculation of equation (7) at each time point, the signal waveform of each dye in each capillary can be obtained.

[0053] When applying the aforementioned crosstalk reduction method, it is necessary that matrix C does not change between the time matrix C is obtained and the time the signal is actually measured. Patent Document 1 shows that this condition is met when the same capillary array is used for obtaining the crosstalk matrix C and for measuring the actual sample. However, in capillary electrophoresis, the capillary array is a consumable item and needs to be replaced with a new array after a certain number of uses. Patent Document 1 does not describe whether the above preconditions are met and whether the crosstalk reduction method functions in this case.

[0054] (simulation) If all parameters, such as the diameter of each capillary, as well as the installation position and angle of each capillary, are exactly the same before and after the replacement of the capillary array, then the aforementioned conditions are met. However, it is practically impossible to satisfy these conditions, and therefore the aforementioned conditions do not hold when the capillary array is replaced. This will be illustrated in detail below using a simulation. The following simulation evaluates the effect on crosstalk when the relative position error of the capillaries in each capillary array changes due to the replacement of the capillary array. The relative position error of the capillaries refers to the deviation from the ideal position of the capillaries, which should ideally be arranged at equal intervals on the same plane. Capillary arrays are manufactured by arranging and fixing capillaries on a substrate, but due to individual differences in the substrates and variations in the fixing work, the spacing between capillaries and the distance of each capillary from the substrate are not exactly the same for each capillary array.

[0055] Figure 2A shows the simulation model. For simplicity of explanation, this simulation does not perform wavelength separation and only considers spatial crosstalk. This model includes a capillary array 201, two lenses 202, and an optical image sensor 203. The inner diameter of the capillary array 201 is set to 50 μm, and the outer diameter to 343 μm, with each capillary arranged at a pitch of 370 μm. A light-emitting region of 44 μm in length is set inside each capillary. The image from the capillary array 201 is formed on the optical image sensor 203 by the two lenses 202. The number of capillaries in the capillary array 201 is set to 8.

[0056] Figure 2B shows the simulation flow. When crosstalk reduction processing is performed (S200: Yes), first, the aforementioned matrix C - Calculate the matrix C. -Assuming a capillary array to be determined, a relative position error is set for the capillary array (S201). After setting the capillary positions, the fluorescence image on the optical image sensor 203 is calculated using the ray tracing method (S202). In this step, fluorescence is generated from only one of the capillaries 1 to 8 and ray tracing is performed. This step is repeated eight times with different emitting capillaries to calculate the output of the optical image sensor 203 when capillaries 1 to 8 emit light. Next, the point image distribution function is convolved onto the calculation result of ray tracing (S203). This point image distribution function is actually measured with a spectrometer composed of a lens, grating, and optical image sensor, and represents the "blur" of the optical system with respect to the capillary array direction (X direction in Figure 2A). In the ray tracing using the model in Figure 2A, lens 202 is calculated as an ideal thin-walled lens, and the deviation from the ideal situation in the actual optical system is expressed by the point image distribution function. Next, the output of the optical image sensor 203 obtained as a result of the calculation is integrated to obtain the crosstalk value by integrating the signals in the regions corresponding to each capillary. For the matrix C generated by arranging the crosstalk values, the generalized inverse matrix C is obtained. - The following is calculated (S204). In this model, wavelength separation is not performed, so matrix C is a square matrix in which the spatial crosstalk between each capillary is arranged in two dimensions.

[0057] Next, the crosstalk is calculated assuming a capillary array to be actually used for sample analysis. If crosstalk reduction processing is not performed (S200: No), the calculation starts from this step. Setting the relative position error to the capillary array (S205), calculation by ray tracing (S206), and convolution of the point image distribution function (S207) are the same as in S201, S202, and S203 described above. After that, the signal vector s is calculated from the output of the obtained optical image sensor 104 (S208). If crosstalk reduction processing is performed (S209: Yes), crosstalk reduction processing is performed according to equation (7) (S210). If crosstalk reduction processing is not performed (S209: No), this step (S210) is skipped. After the calculation is completed, the result is output (S211). Vector d is output if crosstalk reduction processing is performed, and vector s is output if processing is not performed. When there are multiple types of dyes and the dye concentration ratio is calculated from the spectrum, vectors d and s contain different types of information (the ratio of dyes in each capillary and the light intensity of each wavelength segment). However, if the separation of light in the wavelength direction is not considered, they become the same type of information (the light intensity of each capillary directly corresponds to the dye concentration ratio), so the effect of crosstalk processing can be evaluated by comparing the two.

[0058] (Result of crosstalk reduction processing when there is no relative position error) Figure 3A shows the crosstalk without crosstalk reduction processing, obtained by the simulation method described above, and Figure 3B shows the crosstalk with crosstalk reduction processing. In Figures 3A and 3B, it is assumed that there is no relative position error in the capillary array, and matrix C in the same capillary array - The plan is to generate and analyze the samples. Figure 3A shows the horizontal axis representing the luminescent capillaries. Each bar with different hatching indicates the crosstalk value detected in the region (referred to as channels 1-8) where capillaries 1-8 are detected.

[0059] As shown in Figure 3A, in this model, crosstalk is maximized in the channel that detects adjacent capillaries, with a crosstalk of slightly over 1% occurring. The cause of this crosstalk is mainly the reflection of light from the capillary surface and the leakage of fluorescence into adjacent channels due to blurring of the imaging optical system (expressed by the point image distribution function). In reality, other factors that can cause spatial crosstalk include multiple reflections between elements constituting the spectroscopic system, grating anomalies, and scattering due to dust, but this model reflects only the two factors mentioned above.

[0060] As shown in Figure 3B, when the capillary array is the same, it can be seen that crosstalk is reduced by the conventional crosstalk reduction process. The crosstalk, which was slightly over 1%, is reduced to less than 0.01% by the crosstalk reduction process. Generally, the dynamic range of optical image sensors is about 3 to 4 orders of magnitude, and a crosstalk of less than 0.01% means that the crosstalk has fallen below the detection limit. In this case, matrix C - The models used for generating the data and calculating crosstalk are identical, and the reason the crosstalk value in Figure 3B is not zero is due to calculation errors in the ray tracing method.

[0061] (Result of crosstalk reduction processing when relative position error is present) Figures 4A and 4B, like Figures 3A and 3B, show the crosstalk without crosstalk reduction processing and the crosstalk with crosstalk reduction processing. However, in this calculation, the error in the XY direction shown in Figure 2A is given to each capillary using a Gaussian distribution with a standard deviation of 10 μm. Matrix C - Assuming that different capillary arrays are used for matrix generation and sample analysis, matrix C -The relative position error value of the capillaries set in the capillary array differs between the generation of the data and the calculation of crosstalk. Considering that the outer diameter of the capillaries is 343 μm, the above error can be said to be a relatively small relative position error. In addition, although the outer diameter of the capillaries is completely identical in this model, there is generally a tolerance of about ±10 μm for the outer diameter of capillaries. If we assume that changes in the diameter of the capillaries depending on the fixing method will cause a shift in the center position, then the above relative position error is a value that can actually occur.

[0062] Figure 4A shows that even with the introduction of a sequence error, there is still slightly more than 1% crosstalk, similar to the case with zero sequence error. On the other hand, Figure 4B shows that even with conventional crosstalk reduction processing, up to 0.2% of crosstalk remains when a sequence error is present. Furthermore, in some channels, the crosstalk is negative, indicating that excessive crosstalk correction is being performed. The calculation was repeated nine times with different sequence errors, and the average of the maximum residual crosstalk values ​​was taken, resulting in a value of 0.28%. This is because the value of crosstalk fluctuates depending on the sequence error, so matrix C - This indicates that using arrays with different sequence errors during generation and sample analysis may prevent the effectiveness of conventional crosstalk reduction processes from being properly achieved.

[0063] This study demonstrated the impact of array error on crosstalk reduction processing using a model with limited crosstalk-generating factors. In reality, as mentioned above, there are crosstalk-generating factors that are not included in the model. Furthermore, when replacing capillaries, factors other than array error, such as changes in the overall position of the capillary array, changes in the tilt of the capillary array, variations in the inner and outer diameters of the capillaries, and debris on the capillaries, can significantly alter crosstalk. The greater the variation in crosstalk between capillary arrays, the less effective conventional crosstalk reduction processing becomes.

[0064] As the inventors previously stated, conventional crosstalk reduction methods, when the capillary array is replaced, result in matrix C- We have newly discovered that it is necessary to regenerate matrix C. - The crosstalk information used to generate the matrix C is measured by injecting one sample labeled with each dye into each capillary and performing electrophoresis, as described in Patent Document 1, for example. Therefore, if an L-colored fluorescent dye is used and there are N capillaries to be measured, L × N electrophoresis cycles are required. Since one electrophoresis cycle takes several tens of minutes, matrix C - The generation of requires a long time. Patent Document 1 contains matrix C - Methods to reduce the effort and time required for generation are also described. However, using the described methods requires the use of special samples and the operation of special equipment not typically used in normal analysis.

[0065] This invention relates to the exchange of matrix C in a capillary array. - This eliminates the need for regeneration and solves the above-mentioned problems.

[0066] <First Embodiment> Figure 5 is a configuration diagram of the photodetector 500 according to the first embodiment. The photodetector 500 comprises a capillary array 501, an optical fiber array 510 (spatial filter), a spectrometer 505, and a computer 550. The capillary array 501 has a plurality of capillaries 501a, and a plurality of light emitters emit light inside each capillary 501a. The capillary array 501 is replaceable. The optical image sensor 504 constituting the spectrometer 505 has a plurality of detection channels (see Figure 1B) that detect light emitted from the plurality of light emitters in a plurality of wavelength bands. The optical fiber array 510 is a spatial filter that fixes the incident position of the light emitted from the plurality of light emitters to the spectrometer 505. The computer 550 processes the signal output from the optical image sensor 504.

[0067] The photodetector 500 differs from the photodetector 100 in that it has an optical fiber array 510. The optical fiber array 510 is an optical fiber array 510 having multiple optical fibers 510a, each corresponding to the emission point of a plurality of capillaries 501a. The optical fiber array 510 captures the fluorescence emitted from each capillary 501a in the capillary array 501 and guides the fluorescence to a spectrometer 505, which consists of a lens 502, a grating 503, and an optical image sensor 504. The photodetector 500 is intended to be a detection device for a capillary electrophoresis apparatus, and in reality operates as an integrated unit with a temperature control device, a high-voltage application device, an automated stage for samples, a computer for signal processing, etc., to perform analysis of the sample by capillary electrophoresis. However, the explanation of parts other than the photodetector 500 is omitted.

[0068] Furthermore, the crosstalk reduction process of the present invention can be applied even if the spectrometer 505 has a different configuration. Specifically, the method of separating the wavelength of light may be performed using a prism instead of a grating, or it may be performed using multiple dichroic mirrors.

[0069] The crosstalk reduction method in the first embodiment is the same as that described above in equation (7) and related descriptions. However, the conditions for the process require that the fluorescence introduction position to the spectrometer 505 is fixed by the optical fiber array 510, and that the crosstalk generated on the capillary side of the optical fiber array 510 is sufficiently reduced by another method. Note that the fixing of the fluorescence introduction position to the spectrometer 505 may be achieved by a pinhole array or slit array, in addition to the optical fiber array 510.

[0070] The light detection device 500 includes a computer 550. The computer 550 has a processor 551, a main memory unit 552, an auxiliary memory unit 553, and an input / output interface 554 (hereinafter, the interface will be abbreviated as I / F). The processor 551 is a central processing unit that controls the operation of each part of the computer 550. The processor 551 is, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a DSP (Digital Signal Processor), or an ASIC (Application Specific Integrated Circuit). The processor 551 deploys programs stored in the auxiliary memory unit 553 into the working area of ​​the main memory unit 552 in an executable format. The main memory unit 552 stores programs executed by the processor 551, data processed by the processor, etc. The main memory unit 552 is, for example, flash memory, RAM (Random Access Memory), or ROM (Read Only Memory). The auxiliary storage unit 553 stores various programs such as the OS (Operating System) and various data. The auxiliary storage unit 553 is, for example, a solid-state drive (SSD), a hard disk drive (HDD), or a combination thereof. The auxiliary storage unit 553 stores crosstalk information (for example, the matrix C described above) to reduce crosstalk present between multiple detection channels. - The input / output interface 554 is connected to the optical image sensor 504 and the display unit 560 in a way that allows communication.

[0071] Computer 550 uses crosstalk information (for example, the matrix C mentioned above). -By performing calculations on the output from the optical image sensor 504 corresponding to multiple wavelength bands of multiple light-emitting points where multiple light-emitting elements emit light using ) the crosstalk present between the detection channels of the optical image sensor 504 is reduced, and the respective concentration ratio or signal amount ratio of multiple types of light-emitting elements is derived for each of the multiple capillaries. The computer 550 then performs the above calculations using the same crosstalk information even if the capillary array 501 is replaced with another capillary array.

[0072] This crosstalk information is the generalized inverse matrix C of matrix C, which contains information about crosstalk present between multiple detection channels. - or the generalized inverse matrix C - This is equivalent information. Generalized inverse matrix C - The equivalent information is, for example, information that is not in matrix form, and is information for reducing crosstalk present between multiple detection channels. The calculation performed by the computer 550 to derive the density ratio or the signal quantity ratio is performed by a matrix C with respect to the output vector s from the optical image sensor 504. - Apply C from the left - The operation that derives s (see equation (7) above), or an equivalent operation.

[0073] Computer 550 sequentially passes multiple light-emitting elements into each of the multiple capillaries 501a, records the output of the optical image sensor 504 when one of the multiple light-emitting elements is emitting light in one of the multiple capillaries 501a, and generates a matrix C by arranging the outputs of the optical image sensor 504, and the generalized inverse matrix C of matrix C. - to [C t C] -1 C t (See formula (6) above) or an equivalent operation is used to calculate and store the result.

[0074] Crosstalk occurring on the capillary side of the optical fiber array 510 occurs, for example, through paths such as when fluorescence emitted from one capillary directly enters the optical fiber detecting another capillary, as shown in Figure 6, or when fluorescence is reflected off the surface of another capillary and incident on the optical fiber detecting that capillary. Reducing crosstalk occurring through such paths can be achieved, for example, by making the capillary pitch in the capillary array 501 sufficiently large. Alternatively, a known method of reducing crosstalk is to limit the fluorescence acquisition angle for the optical fiber 510a by installing a pinhole slightly smaller than the core diameter of the optical fiber 510a at the end of the optical fiber array 510 on the capillary array 501 side. Another known method involves providing a light-blocking wall between the capillaries of the capillary array 501.

[0075] Referring to Figures 7A to 7C, the simulation results regarding the effectiveness of the crosstalk reduction method of the first embodiment will be explained. Figure 7A shows the model used in the simulation. The model in Figure 7A, like the model in Figure 2A, has a capillary array 701, two lenses 702, and an optical image sensor 703. In Figure 7A, an optical fiber array 710 is added to the model in Figure 2A. The optical fiber array 710 is composed of optical fibers with a core diameter of 200 μm and an NA of 0.5, and eight fibers are arranged at the same intervals as the capillaries of the capillary array 701. In addition, a light-shielding wall 720 is set between the capillaries as a means of eliminating crosstalk that occurs in front of the optical fiber array 710. The simulation flow is the same as that shown in Figure 2B.

[0076] Figures 7B and 7C are graphs showing the results of crosstalk simulations using the model in Figure 7A. Figure 7B is a graph showing the crosstalk without crosstalk reduction processing, and Figure 7C is a graph showing the crosstalk with crosstalk reduction processing. In this simulation, as in Figure 4, matrix C - Assuming that different capillary arrays are used for matrix generation and sample analysis, matrix C -The value of the relative position error of the capillaries set in the capillary array 701 differs between the generation of the data and the calculation of crosstalk.

[0077] As can be seen from Figure 7B, the model in Figure 7A has approximately 0.5% crosstalk. However, when the crosstalk reduction process of equation (7) is applied, the crosstalk is reduced to less than 0.01%, as shown in Figure 7C. The calculation was repeated nine times with different relative position errors, and the average of the maximum values ​​of residual crosstalk was taken, resulting in a value of 0.006%.

[0078] A comparison of the results in Figure 4B and Figure 7C shows that the crosstalk reduction process works correctly even when using capillary arrays with different capillary misalignment values ​​using the method of the first embodiment. This effect is not due to the installation of the light-shielding wall 720. To demonstrate this, the light-shielding wall 720 was added to Figure 2A, and the same simulation was performed. The calculation was repeated nine times with different relative position errors, and the average of the maximum residual crosstalk values ​​was taken. The resulting value was 0.11%, which is more than 10 times the value of the result in Figure 7C.

[0079] As described above, with the crosstalk reduction method of the first embodiment, if the capillary array 501 is replaced, matrix C - The effect of the crosstalk reduction method can be obtained without regenerating the matrix C. This is due to the effect of fixing the fluorescence incidence position to the spectrometer 505 with the optical fiber array 510. Even if the position of each capillary in the capillary array 501 changes, the fluorescence incidence position to the spectrometer 505 is fixed by the optical fiber array 510. Crosstalk that occurs in the spectrometer 505 (for example, caused by multiple reflections or blurring of the spectral image) is not affected by the position of the capillaries because the fluorescence incidence point is fixed. Therefore, the value of the crosstalk does not change by replacing the capillary array, and matrix C - Recreating it becomes unnecessary.

[0080] However, as mentioned above, the method of the present invention requires that the crosstalk generated on the capillary side of the optical fiber array 510 be sufficiently small. Since the method of the first embodiment reduces the invariant component of crosstalk (basically the crosstalk of the spectrometer 505), it cannot remove the component that changes due to the replacement of the capillary array 501. For example, if the amount of crosstalk caused by surface reflection changes due to a change in the placement error of the capillary array, the changed crosstalk will remain even after the crosstalk reduction treatment. To demonstrate this, the light-shielding wall 720 was removed from the model in Figure 7A and the same simulation was performed. The calculation was repeated nine times with different relative position errors, and the average of the maximum value of residual crosstalk was found to be 0.12%. Thus, the optical system to which the method of the present invention is applied must be designed so that the variation in crosstalk generated on the capillary side of the optical fiber array 510 is sufficiently smaller than an acceptable value.

[0081] In the first embodiment of the photodetector 500 and photodetector method, matrix C - Since this matrix C is basically not changed, for example, after the device is manufactured or installed at the place of use, - The matrix generation process should be carried out. Matrix generation can be performed in the same manner as described in Patent Document 1. For example, a sample labeled with a certain dye is injected into a certain capillary, electrophoresis is performed, and the signal of the optical image sensor 504 is recorded. This is repeated for all dyes in all capillaries. Then, the measurement results are arranged to form matrix C. Specifically, the outputs of each wavelength segment of each capillary when the signal of the target dye is maximum in the target capillary are arranged. After that, the general inverse matrix C of matrix C is formed. - Generates.

[0082] During sample analysis, the output of the optical image sensor 504 at each time point is processed according to equation (7) to obtain information on the ratio of each dye at each time point. By arranging the ratios of each dye in a time series, the electrophoretic waveform of the analyte labeled with each dye is obtained.

[0083] matrix C -It basically remains unchanged, but for some reason, matrix C - It is also possible to obtain matrix C. For example, the position of the components of the spectrometer 505 may shift due to relocation of the device or prolonged use. In such cases, the ratio of crosstalk inside the spectrometer may also change, so matrix C may be obtained. - You may regenerate it.

[0084] Furthermore, in order to reduce the effort required to generate matrix C, matrix C may be constructed from partial information. That is, matrix C may consist of multiple matrices C containing only crosstalk information between some of the light-emitting elements or dyes. n This could also be a combination of the components. For example, from the simulation results in Figures 7A to 7C, it can be seen that in this case, spatial crosstalk occurs strongly in the channel measuring adjacent capillaries. In such cases, it is also possible to generate matrix C by ignoring crosstalk from capillaries that are two or more distances apart.

[0085] Let's take the configuration in Figure 7A as an example. Matrix C n Let matrix C be a matrix generated by arranging the signals obtained from the channels measuring each capillary when a dye is injected into capillary n. In generating matrix C, dye is simultaneously injected into each set of capillaries: capillary 1, capillary 4, and capillary 7; capillary 2, capillary 5, and capillary 8; and capillary 3 and capillary 6. Then, matrix C n The matrix C is generated using the signals of the channels corresponding to capillaries n+1, n, and n-1, while the signals of the channels corresponding to the other capillaries are set to zero. That is, when dye is injected into capillaries 1, 4, and 7 simultaneously, C1 is generated using the values ​​of channels 1 and 2, C4 is generated using the values ​​of channels 3, 4, and 5, and C7 is generated using the values ​​of channels 6, 7, and 8. All matrices C n After generating matrix C (C1 to C8 in the above case), nLet matrix C be the result of arranging these elements in order. By generating matrix C in this way, the number of electrophoresis steps required to obtain information can be reduced.

[0086] The above method ignores spatial crosstalk between capillary repairs with small spatial crosstalk values. However, a similar method can be used when a pair of dyes has significantly different emission spectra with little overlap. In other words, for a pair of dyes with non-overlapping fluorescence spectra, matrix C can be generated using data obtained by simultaneously measuring that pair of dyes.

[0087] Furthermore, the generation of matrix C does not necessarily have to be done by injecting the dye into the light-emitting body, i.e., the capillary. Crosstalk information (matrix C) - The matrix C may be obtained by connecting a calibration light source 801 to the spatial filter. The calibration light source 801 sequentially introduces calibration light to each emission acquisition position of the spatial filter. This calibration light is light having the same spectrum as the emission spectra of multiple emitters. Figure 8A shows a diagram of the generation of matrix C using the calibration light source 801. The calibration light source 801 sequentially introduces light having the same spectrum as the emission spectra of each dye into the optical fibers of the optical fiber array 510. Matrix C can be generated from the output of the spectrometer 505 in the same manner as described above.

[0088] Figure 8B shows an example of the structure of the calibration light source 801. The calibration light source 801 has an excitation light source 802 inside. The excitation light source 802 irradiates the channel 803 with fluorescence excitation light. Each dye is sequentially injected into the channel 803 by the dye injection mechanism 804. The fluorescence generated in the channel 803 is collimated by the lens 805, and the excitation light is removed by the wavelength filter 806. The collimated fluorescence is introduced into one of the fibers of the output fiber array 809 by the mirror 807 and lens 808. A part of the mirror 807 is designed to change angle, allowing the fiber into which the fluorescence is introduced to be changed.

[0089] The crosstalk reduction method of the first embodiment assumes a linear relationship between crosstalk and the signal that causes it. Therefore, if this relationship is disrupted due to detector saturation, the method of the first embodiment ceases to function. Figure 9 shows a schematic graph illustrating the effect of detector saturation. In the graph of Figure 9, the horizontal axis represents the concentration of the dye being measured, and the vertical axis represents the amount of crosstalk. In a spectroscopic system with crosstalk, the dye concentration and crosstalk are proportional. In an ideal situation with no crosstalk, the crosstalk is always zero, regardless of the dye concentration.

[0090] According to the crosstalk reduction method of the first embodiment, ideally, the crosstalk is completely zero. Therefore, in a spectroscopic system where crosstalk exists but the crosstalk reduction process of the first embodiment is performed, under ideal conditions, the crosstalk is zero regardless of the dye concentration up to a certain dye concentration (even under realistic conditions where some crosstalk remains after the reduction process, the reduced crosstalk is smaller than the original value). However, if the detector becomes saturated in the capillary measuring the dye to be measured, the output signal remains constant even if the dye concentration increases further. On the other hand, since crosstalk itself is proportional to the fluorescence intensity, it continues to increase in proportion to the concentration. As a result, the actual amount of crosstalk exceeds the amount of crosstalk estimated from the detector output. In this case, the crosstalk reduction method of the first embodiment cannot completely eliminate the crosstalk, and the crosstalk value becomes non-zero. Therefore, the crosstalk reduction method of the first embodiment must be used within a range in which the detector does not become saturated. If one channel of the detector becomes saturated, it is useful to warn the user that the original performance may not be obtained even in other channels that are not saturated.

[0091] Specifically, if the output from the optical image sensor 504 becomes saturated, the computer 550 may display a warning on the display unit 560.

[0092] <Signal Processing Method> Here, a signal processing method using the photodetection apparatus 500 of the first embodiment will be described.

[0093] The signal processing method of the present embodiment includes: preparing the photodetection apparatus 500; measuring light emission from a plurality of light emitters by means of the optical image sensor 504; using crosstalk information (matrix C - ) for reducing crosstalk existing between a plurality of detection channels stored in a computer 550 to perform an arithmetic operation on outputs from the optical image sensor 504 corresponding to a plurality of wavelength bands of each of a plurality of light emission points emitted by the plurality of light emitters, thereby reducing crosstalk existing between detection channels of the optical image sensor 504, and deriving a concentration ratio or a signal amount ratio of each of a plurality of types of light emitters for each of a plurality of light emitter holders (capillaries 501a), and performing the arithmetic operation using the same crosstalk information even when the capillary array 501 is replaced.

[0094] The crosstalk information (matrix C - ) is the generalized inverse matrix C of a matrix C containing crosstalk information - or information equivalent thereto. Further, the operation for deriving the concentration ratio or the signal amount ratio is performed by applying matrix C - from the left to an output vector s from the optical image sensor 504 to obtain C - s, which is an operation (see the above formula (7)) or an operation equivalent thereto.

[0095] <First Embodiment: Summary> The photodetection apparatus 500 according to the first embodiment includes a capillary array 501, an optical fiber array 510, a lens 502, a grating 503, and an optical image sensor 504. The photodetection apparatus also includes a computer 550 that processes outputs from the optical image sensor 504. The computer 550 stores the generalized inverse matrix C of matrix C generated from outputs of the optical image sensor 504 obtained when samples labeled with dyes to be used are sequentially injected into each individual capillary in the capillary array 101 of the photodetection apparatus 500 and measured -is stored internally. When analyzing a measurement object, the computer 550 applies the generalized inverse matrix C to the output of the optical image sensor 504 at each time - to obtain the ratio of each dye. Even when the capillary array 501 is replaced, the same value is used for the generalized inverse matrix C - .

[0096] <Second Embodiment> In the first embodiment, the density ratio of each dye is obtained by collectively processing the output from the optical image sensor 104 with the generalized inverse matrix C - . In the second embodiment, spectrum crosstalk caused by dyes, spectrum crosstalk caused by the spectroscope, and spatial crosstalk are processed separately. In the crosstalk reduction method according to the second embodiment, crosstalk generated inside the spectroscope is reduced using a matrix obtained with a calibration light source, and spectrum crosstalk caused by overlapping fluorescence spectra of dyes is reduced using a matrix obtained by measuring fluorescence spectra from dyes introduced into capillaries. Therefore, the crosstalk reduction processing has a form different from that of formula (7). This processing format allows easy modification and addition of dye types.

[0097] First, crosstalk generated inside the spectroscope 505 is organized. It is assumed that the spectroscope 505 is provided with N optical fibers for introducing light, and the optical image sensor 504 measures M wavelength intervals for each optical fiber. For example, M is 20, and when the detection wavelength range is 500 to 700 nm, the wavelength interval means that the optical image sensor 504 outputs 20 signals for each fiber: a fluorescence signal in the wavelength range 500 to 510 nm, a fluorescence signal in the wavelength range 510 to 520 nm, ..., and a fluorescence signal in the wavelength range 690 to 700 nm. Here, the output of wavelength interval m for fiber n of the optical image sensor 504 is denoted as S mn . A one-dimensionally arrayed version of S mn is used as the output signal vector s of the optical image sensor 104. Note that S mn has two subscripts, and S mn is arranged one-dimensionally so as to become the (Mn+m) component s Mn+m of s.

[0098] On the other hand, consider the light incident on fiber n of spectrometer 505. The power of the light in the wavelength range corresponding to the wavelength interval m is P. mn This is how it is written. This Power P mn Arrange them to create the incident power vector p. mn As before, the Mn + m component of p is p Mn+m They are arranged in one dimension in such a way.

[0099] Spectrometer 505 outputs a signal that is linear with respect to the power of the incident light. Therefore, the relationship between p and s can be expressed by the following equation (8). s = Gp (8)

[0100] Matrix G is a matrix representing the crosstalk occurring in the spectrometer 505. Matrix G is obtained by sequentially introducing monochromatic light corresponding to the wavelength detection category of the optical sensor (e.g., optical image sensor 504) to each emission acquisition position of the spatial filter (e.g., optical fiber array 510). In an ideal spectrometer without crosstalk, matrix G is the identity matrix (assuming that s and p are normalized so that the output signal is 1 when the light power is 1). If there is crosstalk, the triangular portion shown in matrix G in Figure 10 represents the region indicating the value of the crosstalk. Also, if the sensitivity of the spectrometer differs for each emission acquisition position, the diagonal components of matrix G may take values ​​other than 1. For example, if the spectrometer 505 has the configuration shown in Figure 5, the light loss within the spectrometer may increase with increasing emission acquisition positions that are farther from the central axis of the lens 502. In this case, the value of the diagonal component corresponding to that acquisition position will be less than 1.

[0101] Next, we will organize the relationship between the dye in the capillary and the power of the light incident on each fiber. Let's assume that L types of dyes are used for the analysis. Of the N capillaries, let D be the ratio of the l-th dye present in the n-th capillary. ln Let's assume that. D ln Let the one-dimensional arrangement of these be the pigment ratio vector d. ln It has two subscripts, but D ln This is the Ln+l component d of dLn+l They are arranged in one dimension in such a way.

[0102] Here, we assume that the signal generated in the nth capillary is measured in the nth optical fiber. The power P of the light measured in the nth optical fiber. mn The emission spectra of L types of fluorescent dyes in the nth capillary are in ratio D ln This results in the sum of the two vectors. Therefore, the relationship between the incident power vector p and the dye ratio vector d is given by the following equation (9). p=Fd (9) Here, matrix F is a matrix containing the fluorescence spectra of each dye.

[0103] From equations (8) and (9), we obtain the following equation (10). s=GFd (10) Therefore, the inverse matrix of matrix G is G -1 The general inverse of matrix F is F - Then, the following equation (11) is obtained: Generalized inverse matrix F - is...F t F] -1 F t Alternatively, it can be obtained by an equivalent operation. d=F - G -1 s (11) This is G -1 F - If this is determined, it means that the dye ratio d can be determined from the output s of the optical image sensor. -1 The calculation refers to the process of reducing crosstalk that occurs inside the spectrometer, F - This calculation involves calculating the dye ratio while considering the overlap of the dye's fluorescence spectra (reducing spectral crosstalk).

[0104] Figure 10 shows the shapes of matrices G and F. Matrix G is a square matrix of size M × N, with 1s in its diagonal elements. The triangular regions above and below the diagonal elements represent crosstalk. In an ideal spectrometer without crosstalk, all elements in this region are 0. Matrix F is a matrix with M × N rows and L × N columns. Matrix F is a matrix in which N matrices F1, each with M rows and L columns, are arranged in the diagonal, and the rest of the region is 0. The spectra of L dyes are arranged in columns 1 through L of matrix F1 (the rectangles in Figure 10 represent the spectra).

[0105] In the crosstalk reduction method of the second embodiment, the calculation is divided into two stages, and the acquisition of matrices G and F is performed separately. The acquisition of matrix G is performed using a calibration light source 1101 (see Figure 11). The calibration light source 1101 is connected to the optical fiber array 510 of the spectrometer 505 in the same way as the calibration light source 801. The calibration light source 1101 sequentially introduces light of wavelengths corresponding to M wavelength intervals into each of the N fibers. That is, the calibration light source 1101 first introduces light of the wavelength corresponding to wavelength interval 1 into fiber 1. Next, it introduces light of the wavelength corresponding to wavelength interval 2, light of the wavelength corresponding to wavelength interval 3, ... light of the wavelength corresponding to wavelength interval M. Then the same operation is performed for fiber 2. After that, the same operation is performed for fiber 3, fiber 4, ... fiber N.

[0106] Matrix G is generated as follows: When the calibration light source 1101 introduces light of wavelength segment m into fiber n, the output values ​​for fiber k and wavelength segment l of the optical fiber array are A kl←mn Let's assume that the η and ζ components of matrix G are G, where ζ = Mk + l and η = Mn + m. ζη =A kl←mn / A mn←mn Let's assume that.

[0107] Figure 11 shows an example of the structure of a calibration light source 1101 that performs the operations described above. The calibration light source 1101 has a white light source 1102 inside. The white light source 1102 is, for example, a white LED or a halogen lamp. Light from the white light source 1102 is separated by wavelength by a lens 1103 and a grating 1104. A slit 1105 is placed at the imaging point of the separated light. The slit 1105 cuts out only the components of a specific wavelength from the light of the white light source 1102. The grating 1104 is designed to change its angle, and by adjusting this angle, the wavelength of the light passing through the slit 1105 is controlled. The light that has passed through the slit 1105 is introduced into a specific fiber in the output fiber array 1108 by a lens 1106 and a mirror 1107. One of the mirrors 1107 is designed to change its angle, and by adjusting this angle, the light is introduced into which fiber in the output fiber array 1108 is controlled.

[0108] The matrix F can be measured by simultaneously injecting the dye-labeled sample into all capillaries. In the matrix acquisition method of the first embodiment, the dye had to be injected into only one capillary at a time and the signal measured for each of the N capillaries, whereas in the matrix acquisition method of the second embodiment, it is possible to measure simultaneously in all capillaries, thus reducing the time required for measurement to 1 / N.

[0109] Since matrix F is a matrix obtained by repeatedly arranging matrix F1, which contains the spectra of each dye, for the number of capillaries, matrix F1 may be measured by measuring the dye spectrum in a specific capillary, and matrix F1 may be obtained by repeating this for the number of capillaries. In conventional crosstalk reduction methods and the crosstalk of the first embodiment, information on spectral crosstalk caused by the dye and information on crosstalk caused by the photodetector and information on the detection efficiency of each wavelength are mixed in a single processing matrix. Therefore, it was necessary to run each dye through each capillary to acquire the information. However, in the crosstalk reduction method of the second embodiment, information on the effect caused by the spectrometer is aggregated in matrix G, so it is not necessary to measure the same sample for each capillary when measuring matrix F.

[0110] As described above, the crosstalk reduction method of the second embodiment makes it easier to obtain the processing matrix, particularly matrix F which includes information on the dye spectrum, compared to the conventional crosstalk reduction method described in Patent Document 1 and the crosstalk reduction method of the first embodiment. This is particularly useful when adding dyes to be used for analysis. If it is desired to use a new dye for which no information exists in the computer performing the crosstalk reduction processing, the conventional crosstalk reduction method described in Patent Document 1 or the crosstalk reduction method of the first embodiment requires introducing the new dye into all capillaries sequentially and obtaining the fluorescence spectrum information of the new dye. On the other hand, with the crosstalk reduction method of the second embodiment, it is sufficient to introduce the new dye into all capillaries simultaneously, or into a specific single capillary, and measure the fluorescence spectrum.

[0111] Alternatively, matrix F may be provided in a form that is pre-stored within the computer. Crosstalk information that depends on each individual spectrometer is included in matrix G. Therefore, matrix F does not need to be measured for each individual photodetector; spectral information of each dye measured in a specific individual may be stored in the computer of another individual, and matrix F may be generated from that information. That is, spectral information of dyes measured in a certain photodetector may be published on the Web, users may download the spectral information of the dyes from the Web and save it in their computers, and matrix F may be generated from the saved spectral information.

[0112] Figure 12 shows an example of the instrument's operation when acquiring matrix F. First, the user sets the dye or set of dyes from which to acquire spectra (S1201). Next, the user selects the method for acquiring fluorescence spectra (S1202). If the method of measuring dye spectra in all capillaries is selected, the instrument first prepares for electrophoresis (S1203). After preparation is complete, the user sets the sample labeled with the dye to be measured (S1204). After the sample is set, the instrument injects the sample into the capillary and performs electrophoresis (S1205). The spectrum is acquired when the sample has migrated to the fluorescence measurement point (S1206). This cycle is repeated for all dyes from which spectra are to be acquired (S1207). If the sample is prepared so that all dye spectra can be acquired at once, this cycle only needs to be performed once. This is, for example, a sample in which dyes A, B, C, etc., to be measured are labeled on DNA with chain lengths of 100 bases, 150 bases, and 200 bases, respectively, and then mixed. When electrophoresis is performed, dyes A, B, C, etc., emit light sequentially at different times. Such calibration samples are commercially available and relatively easy to obtain. Because the DNA reaches the fluorescence measurement points in order of its chain length during electrophoresis, dyes A, B, and C emit light sequentially over time. After measuring all the fluorescence spectra, the fluorescence spectra are arranged to generate matrix F (S1208).

[0113] When the method of measuring fluorescence spectra with a single capillary is selected, the operation of the instrument is the same as the method of measuring fluorescence spectra with all capillaries (S1209~S1214). However, the sample is injected into only a single capillary. When matrix F is generated, the fluorescence spectra obtained with a single capillary are repeatedly arranged.

[0114] If the method of generating matrix F from internal information is selected, the fluorescence spectrum (emission spectrum information) of the set dye is read from memory (S1215). The fluorescence spectrum may be stored in the computer from the beginning, or the spectrum information may be provided via the internet or other means and downloaded. Matrix F is generated from the read fluorescence spectrum (S1216). The generated matrix F is saved to the computer (S1217).

[0115] <Variation> The present invention is not limited to the embodiments described above, but includes various modifications. For example, the embodiments described above are described in detail to make the present invention easier to understand, and are not necessarily limited to those having all the configurations described. Furthermore, it is possible to replace parts of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add configurations from other embodiments to the configuration of one embodiment. In addition, it is possible to add, delete, or replace parts of the configuration of each embodiment with other configurations.

[0116] For example, in the first embodiment, an example was described in which an optical fiber array 510 is provided as a spatial filter, but the spatial filter may also be a filter having through holes that allow light to pass through. The shape of the through holes in this filter may be circular or rectangular. Figure 13 is a configuration diagram of a modified photodetector 1300. Unlike the photodetector 500, the photodetector 1300 has a pinhole array 1310 in which a plurality of pinholes (circular through holes) are arranged in an array, corresponding to each of the light-emitting points of a plurality of light-emitting body holders (capillaries). This pinhole array 1310 may also be a slit array in which slits (rectangular through holes) are arranged in an array.

[0117] In the above embodiment, a photodetector is used that comprises multiple light-emitting holders, each emitting light from multiple types of light-emitting elements. However, if there are multiple types of light-emitting elements, there may be only one light-emitting holder, or if there are multiple light-emitting holders, there may be only one type of light-emitting element.

[0118] Furthermore, although the calculations in the above embodiment involved various operations using matrices, equivalent operations using information in a non-matrix format may also be used, as long as equivalent results can be obtained. [Explanation of Symbols]

[0119] 100 Light detection device 101 Capillary Array 102 Lens 103 Grating 104 Optical image sensor 105 Spectral image 106 channels 201 Capillary Array 202 Lens 203 Optical Image Sensor 500 Light detection device 501 Capillary Array 501a Capillary 502 Lens 503 Grating 504 Optical Image Sensor 505 Spectrometer 510 Fiber Optic Array 510a optical fiber 550 calculator 551 processors 552 Main memory 553 Auxiliary storage 554 Input / Output Interfaces 560 Display section 701 Capillary Array 702 Lens 703 Optical Image Sensor 710 Fiber Optic Array 720 Light-blocking wall 801 Calibration light source 802 Excitation light source 803 Flow channel 804 Dye injection mechanism 805 lens 806 wavelength filter 807 Miller 808 Lens 809 Output Fiber Array 1101 Calibration light source 1102 White light source 1103 Lens 1104 Grating 1105 Slit 1106 Lens 1107 Miller 1108 Output Fiber Array 1300 Light detection device 1310 Pinhole Array

Claims

1. Multiple interchangeable light-emitting element holders, each containing multiple types of light-emitting elements, A light sensor having multiple detection channels for detecting light emitted from multiple types of light emitters in multiple wavelength bands, A spatial filter that fixes the incident position of the light emitted from the plurality of types of light emitters to the light sensor, The system includes a computer that processes the signal output from the optical sensor, The aforementioned computer is It internally contains crosstalk information to reduce crosstalk present between the multiple detection channels, By performing calculations on the output from the photosensor corresponding to the multiple wavelength bands of each of the multiple light-emitting points from which the multiple types of light-emitting elements emit light, using the crosstalk information, the crosstalk present between the detection channels of the photosensor is reduced, and the respective concentration ratio or signal quantity ratio of the multiple types of light-emitting elements is derived for each of the multiple light-emitting element holders. Even if the light-emitting element holder is replaced with another light-emitting element holder, the calculation is performed using the crosstalk information. A light detection device characterized by the following features.

2. The crosstalk information is the general inverse matrix C of matrix C, which contains crosstalk information present between the plurality of detection channels. - Alternatively, it is information equivalent to the generalized inverse matrix, The calculation for deriving the concentration ratio or the signal quantity ratio is performed using the matrix C with respect to the output vector s from the light sensor. - Apply C from the left - The operation that derives s, or an equivalent operation. The light detection device according to claim 1.

3. The matrix C is a plurality of matrices C that contain only crosstalk information between some of the light-emitting holders or dyes. n It is a combination of the components. The light detection device according to claim 2.

4. The spatial filter is an optical fiber array having multiple optical fibers provided corresponding to each of the light-emitting points of the multiple light-emitting element holders. The light detection device according to feature 1.

5. The spatial filter has a plurality of pinholes or slits provided corresponding to each of the light-emitting points of the plurality of light-emitting element holders. The light detection device according to feature 1.

6. The plurality of light-emitting elements are sequentially passed through each of the plurality of light-emitting element holders, the output of the light sensor is recorded when one of the plurality of light-emitting element holders is emitting light from one of the plurality of light-emitting element holders, the outputs of the light sensor are arranged to generate a matrix C, and the general inverse matrix C of matrix C is generated. － to [C t C] -1 C t Alternatively, calculate and store it using an equivalent operation. The light detection device according to claim 2.

7. When said crosstalk information is defined such that G is a matrix representing crosstalk generated by said photodetection device, and F is a matrix representing spectral crosstalk generated due to overlapping emission spectra of dyes, an inverse matrix G of said matrix G -1 and, [F t F] -1 F t or a generalized inverse matrix F of said matrix F obtained by an operation equivalent thereto - and an operation using F - G -1 s or an operation equivalent thereto, a dye concentration ratio or a signal amount ratio is obtained The light detection device according to claim 1.

8. The matrix G is obtained by sequentially introducing monochromatic light corresponding to the wavelength detection category of the photosensor to each emission acquisition position of the spatial filter. The light detection device according to feature 7.

9. The matrix F is obtained by sequentially passing the multiple types of light-emitting elements through all or part of the multiple light-emitting element holders. The light detection device according to feature 7.

10. The matrix F is generated from the emission spectrum information of the multiple types of light emitters stored inside the computer. The light detection device according to feature 7.

11. The aforementioned crosstalk information is acquired by connecting the calibration light source to the spatial filter. The calibration light source sequentially introduces calibration light to each light source acquisition position of the spatial filter. The light detection device according to claim 2 or 7, characterized by the features described above.

12. The calibration light source introduces light having the same spectrum as the emission spectrum of the multiple types of light emitters. The light detection device according to claim 11.

13. The calibration light source introduces monochromatic light corresponding to the wavelength detection category of the light sensor. The light detection device according to claim 11.

14. If the output from the aforementioned light sensor becomes saturated, a warning will be displayed. The light detection device according to feature 1.

15. To prepare a photodetector comprising: multiple interchangeable light-emitting element holders in which multiple types of light-emitting elements emit light internally; a photosensor having multiple detection channels for detecting the light emitted from the multiple types of light-emitting elements in multiple wavelength bands; a spatial filter for fixing the incident position of the light emitted from the multiple types of light-emitting elements to the photosensor; and a computer for processing the signal output from the photosensor. The light sensor measures the emission of light from the multiple types of light-emitting elements. Using crosstalk information stored in the computer for reducing crosstalk between the multiple detection channels, the output from the photosensor corresponding to the multiple wavelength bands of each of the multiple light-emitting points from which the multiple types of light-emitting elements emit light is calculated, the crosstalk between the detection channels of the photosensor is reduced, and the respective concentration ratio or signal amount ratio of the multiple types of light-emitting elements is derived for each of the multiple light-emitting element holders, and The calculation is performed using the crosstalk information even if the light-emitting element holder is replaced. A signal processing method characterized by the following:

16. The aforementioned crosstalk information is the generalized inverse matrix C of matrix C containing the crosstalk information. - Alternatively, it is equivalent information, and the calculation for deriving the concentration ratio or the signal quantity ratio is performed with respect to the matrix C of the output vector s from the light sensor. - Apply C from the left - The operation that derives s, or an equivalent operation. The signal processing method according to feature 15.

17. The crosstalk information used in the above calculation is such that G is a matrix representing the crosstalk generated in the photodetector, and F is a matrix representing the spectral crosstalk generated by the overlap of the emission spectra of the dyes, and the inverse matrix G is the inverse matrix of G. -1 And, [F t F] -1 F t Alternatively, the generalized inverse matrix F of the matrix F obtained by an equivalent operation. - Calculation F using and - G -1 The pigment concentration ratio or signal level ratio is obtained by s or an equivalent operation. The signal processing method according to feature 16.

Images