Detection Method

Abstract

To provide a detection method with which it is possible to properly detect any of a bright point that exists alone and a bright point cloud that exists densely.SOLUTION: A detection method according to the present invention comprises: a step for preparing sample image data of a sample in which a target substance is labeled with a labeling substrate; a step for obtaining first background image data in which image feature quantity of a single bright point in the sample image is reduced by a first filter; a step for obtaining second background image data in which the image feature quantity of dense bright points in the sample image is reduced by a second filter; a step for synthesizing the first and second background image data on the basis of the brightness information of the sample image data and obtaining synthesized background image data; and a step for finding differential image data that represents a difference between the sample image data and the synthesized background image data.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a detection method. [Background technology]

[0002] In pathological diagnosis, quantifying the expression level of an observation substance using tissue sections or the like can provide very important information for predicting prognosis and determining subsequent treatment plans. Therefore, it is very important to accurately quantify the observation substance in tissue sections or the like. Known methods for quantifying an observation substance include a method in which the observation substance is labeled with a fluorescent substance and detected with high sensitivity (see, for example, Patent Document 1).

[0003] Patent Document 1 describes a method for quantifying the expression level of an observation substance in a tissue section after immunostaining. In the method described in Patent Document 1, first, the fluorescence intensity distribution of a tissue section in which the observation substance is fluorescently labeled is detected. Next, to remove autofluorescence, a region containing only frequency components higher than a predetermined spatial frequency is extracted. Finally, the observation substance is quantified by measuring the number of bright spots derived from the fluorescent substance. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] International Publication No. 2019 / 087853 Summary of the Invention [Problem to be solved by the invention]

[0005] However, the quantification method described in Patent Document 1 targets only bright spots that are not clustered together, and does not target clusters of bright spots. Because a labeling substance containing a fluorescent substance adheres to a specific biological material, it is conceivable that bright spots will be distributed in a clustered manner in a specific narrow area. In this case, the quantification method described in Patent Document 1 can appropriately quantify bright spots that are not clustered together, but cannot appropriately quantify clusters of bright spots.

[0006] An object of the present invention is to provide a detection method that can appropriately detect both a single bright spot and a group of bright spots that are densely packed together. [Means for solving the problem]

[0007] A detection method according to one embodiment of the present invention includes the steps of preparing specimen image data in which a target substance is labeled with a labeling substance, obtaining first background image data by using a first filter to reduce the image features of isolated bright spots in the specimen image, obtaining second background image data by using a second filter to reduce the image features of densely packed bright spots in the specimen image, synthesizing the first background image data and the second background image data based on brightness information of the specimen image data to obtain composite background image data, and calculating differential image data, which is the difference between the specimen image data and the composite background image data. [Effects of the Invention]

[0008] According to the present invention, the target substance can be appropriately detected whether it is a single bright spot or a group of bright spots that are densely packed together. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a flowchart of a detection method according to one embodiment of the present invention. [Figure 2]FIG. 2A is a schematic graph showing the intensity distribution of the specimen image data, and FIG. 2B is a schematic graph showing the intensity distribution of the first background image data and the second background image data. [Figure 3] 3A to 3C are diagrams for explaining the process of obtaining composite background image data. [Figure 4] 4A and 4B are graphs showing the intensity distribution of the differential image data. [Figure 5] FIG. 5 is a flowchart showing how to determine the first threshold value and the second threshold value. [Figure 6] FIG. 6 is a schematic graph showing the relationship between the distance between bright spots and the coincidence rate. DETAILED DESCRIPTION OF THE INVENTION

[0010] A detection method according to an embodiment of the present invention will be described in detail below.

[0011] Fig. 1 is a flowchart of a detection method according to one embodiment of the present invention. Figs. 2A, 2B, 3A-C, and 4A, 4B are diagrams for explaining a detection method according to one embodiment of the present invention. In Figs. 2A, 2B, 3C, and 4A, 4B, the horizontal axis represents distance, and the vertical axis represents fluorescence intensity.

[0012] As shown in Figures 1, 2A and 2B, 3A-C, and 4A and 4B, the detection method of the present invention includes the steps of preparing specimen image data of a specimen in which a target substance is labeled with a labeling substance, obtaining first background image data by reducing the image features of a single bright spot in the specimen image using a first filter, obtaining second background image data by reducing the image features of a densely packed bright spot in the specimen image using a second filter, synthesizing the first background image data and the second background image data based on brightness information of the specimen image data to obtain composite background image data, and calculating differential image data, which is the difference between the specimen image data and the composite background image data.

[0013] FIG. 2A shows the relationship between the distance and fluorescence intensity of a portion of the specimen image data. As shown in FIGS. 1 and 2A, in the step of preparing specimen image data, specimen image data to be used for detection is prepared (S110). The specimen image data is obtained by photographing a biological specimen in which a target substance is labeled with a labeling substance. The number of specimen image data sets is not limited. The number of specimen image data sets may be one or more. By preparing multiple specimen image data sets, quantitative relationships between biological substances among the multiple specimen image data sets, the localization of target substances (substances under observation), and the like can be observed. Here, the type of biological specimen used in the specimen image is not particularly limited as long as the biological substance can be detected. Examples of biological specimens include pathological tissues, tissue sections such as cell line-derived xenografts (CDX) and patient-derived xenografts (PDX), and specimens made from cultured cells.

[0014] Biological substances are substances to be quantified and are contained in biological specimens. The type of biological substance is not particularly limited. Examples of biological substances include nucleic acids (such as single-stranded or double-stranded DNA, RNA, polynucleotides, oligonucleotides, and PNA (peptide nucleic acids), as well as nucleosides, nucleotides, and modified molecules thereof); proteins (such as polypeptides, oligopeptides, and receptors present in the cell membrane of target cells); amino acids (including modified amino acids); carbohydrates (such as oligosaccharides, polysaccharides, and sugar chains); lipids; exosomes; or modified molecules and complexes thereof. Furthermore, more specific examples of biological substances include 5T4, AXL, BCMA, C4.4A, CA6, Cadherin3, Cadherin6, CEACAM5, CD16, CD19, CD22, CD37, CD56, CD71, CD138, CD142, CD352, DLL3, EphA2, EphrinA4, ETBR, FcγRIII, FOLR1, FGFR2, FGFR3, GCC, HER1 (EGFR), HER2, HER3, HER4, IntegrinαV, LAMP1, LIV1, Mesothelin, MUC1, MUC16, NaPi2B, Nectin4, NOTCH3, PD-1, PD-L1, PSMA, PTK7, SLAMF7, SLITRK6, STEAP1, TROP2, Ki67, HER4, ER, and PR.

[0015] The biological material is labeled with a labeling substance. Examples of methods for labeling biological materials include immunostaining methods using antibodies or antibody fragments, and staining methods using molecular recognition groups similar to antibodies. In methods for labeling biological materials using staining methods using molecular recognition groups similar to antibodies, for example, aptamers and SNAP-tags are used as the molecular recognition groups. Examples of labeling substances include phosphor-containing particles and other fluorescent substances.

[0016] In the immunostaining labeling method, a biological specimen containing a biological material is immunostained to obtain an immunostained image in which the biological material is visualized by fluorescent labeling.

[0017] Furthermore, in the secondary reaction of immunostaining, a labeling substance containing fluorescent dye-holding particles or other fluorescent substances can be used. For example, the target substance is labeled with a labeling substance containing fluorescent dye-holding particles.

[0018] Phosphor-aggregated particles are nano-sized particles that have a structure in which multiple phosphors (e.g., fluorescent dyes or semiconductor nanoparticles) are encapsulated within and / or adsorbed onto the surface of an organic or inorganic particle. Examples of fluorescent dyes that make up phosphor-aggregated particles include rhodamine-based dyes, Cy-based dyes, Alexa Fluor®-based dyes, BODIPY-based dyes, squarylium-based dyes, cyanine-based dyes, aromatic ring-based dyes, oxazine-based dyes, carbopyronine-based dyes, and pyrromethene-based dyes. Examples of semiconductor nanoparticle materials that make up phosphor-aggregated particles include II-VI semiconductors, III-V semiconductors, or IV semiconductors. Phosphor-aggregated particles can be prepared according to known methods (see, for example, JP 2013-57937 A).

[0019] Figure 2B shows the relationship between the distance of a portion of the first background image data and the fluorescence intensity. The dotted line in Figure 2B represents the first background image data, and the solid line represents the second background image data. As shown in Figures 1 and 2B, the process of obtaining the first background image data involves using a first filter to reduce the image features of isolated bright spots in the specimen image data (S120). Specifically, the first filter obtains a first background image of isolated bright spots in the specimen image data. Here, "isolated" means that the combined signal value of the base portions of the isolated bright spots is equal to or less than the signal value of any other bright spot due to their proximity to each other. For example, while the base signal pattern of a bright spot depends on the imaging system, if the magnification is 20x, the NA is 0.8, and the pixel pitch on the captured image is 0.325 μm, a bright spot can be considered to exist alone if it is separated by approximately 2 μm or more. The type of first filter is not particularly limited as long as it can reduce the image features of isolated bright spots. Examples of the first filter include a median filter and a frequency filter. In this embodiment, the first filter is a median filter. Also, in this embodiment, the image feature amount is the intensity distribution of light emitted from a labeling substance that exists independently.

[0020] As shown in Figures 1 and 2B, in the step of obtaining second background image data, a second filter is used to reduce the image features of densely packed bright spots in the specimen image data to obtain second background image data (S130). Specifically, the second filter is used to obtain a second background image of densely packed bright spots in the specimen image. Here, "dense" refers to the presence of particles at a distance between bright spots where the sum of the signal values ​​of the bases of the bright spots becomes significantly higher than signal values ​​other than those derived from the labeled substance due to their proximity to each other. For example, while the base signal pattern of bright spots depends on the imaging system, if the magnification is 20x, the NA is 0.8, and the pixel pitch on the captured image is 0.325 μm, bright spots can be considered to be densely packed if they are distributed at a distance of approximately 2 μm between particles. The type of second filter is not particularly limited as long as it can reduce the image features of densely packed bright spots. Examples of the second filter include a median filter and a frequency filter. In this embodiment, the first filter is a median filter. In this embodiment, the image feature amount is the intensity distribution of light emitted from the densely-occurring labeling substances.

[0021] The first light intensity distribution (image feature) in the specimen image data reduced by the first filter and the second light intensity distribution (image feature) in the specimen image data reduced by the second filter differ in the width of the peak in the light intensity distribution, the height of the peak in the intensity distribution, or the width and height of the peak. For example, the first light intensity distribution (image feature) in the specimen image data reduced by the first filter has a narrower peak width and a lower peak height than the second light intensity distribution (image feature) in the specimen image data reduced by the second filter. Conversely, the second light intensity distribution (image feature) in the specimen image reduced by the second filter has a wider peak width and a higher peak height than the first light intensity distribution (image feature) in the specimen image reduced by the first filter. This is thought to be due to overlapping of light intensities depending on the density of bright spots.

[0022] In this embodiment, the first filter, which is a median filter, and the second filter, which is a median filter, process different numbers of pixels at a time. For example, if the labeling substance contains phosphor-accumulating particles and the size of the phosphor-accumulating particles on the screen is 9 pixels x 9 pixels, the number of pixels processed at a time by the first filter is approximately 9 pixels x 9 pixels, and the number of pixels processed at a time by the second filter is approximately 80 pixels x 80 pixels to 90 pixels x 90 pixels. In this way, in this embodiment, by processing a narrow area with the first filter and a wide area with the second filter, it is possible to detect both a single bright spot and a group of bright spots that are densely packed together. Note that when frequency filters are used as the first filter and the second filter, for example, the first filter is a filter that passes only high spatial frequency regions of the spatial frequency information of the original image, and the second filter is a filter that passes only low spatial frequency regions of the spatial frequency information of the original image.

[0023] 3A and 3B show first and second background image data binarized using a first threshold, and FIG. 3C shows composite background image data. As shown in FIG. 1 and 3A-C, in the step of obtaining composite background image data, the first and second background image data are combined based on brightness information of the sample image data (S140). The composite background image data uses the first background image data in areas where a single bright spot exists and the second background image data in areas where a cluster of bright spots exists. The method for obtaining the composite background image data is not particularly limited. The composite background image data may be obtained by setting the areas where a single bright spot exists and the areas where a cluster of bright spots exists in advance, by using the first background image data near the peak position of the single bright spot and the second background image data near the peak position of the cluster of bright spots, or by using the following method.

[0024] Specifically, first, an intensity difference, which is the difference between the second light intensity distribution in the second background image data and the first light intensity distribution in the first background image data, is calculated at each distance. Then, the first background image data is used in areas where the light intensity difference is less than a first threshold, and the second background image data is selected in areas where the intensity difference is equal to or greater than the first threshold. Next, the first background image data in areas where the intensity difference is less than the first threshold and the second background image data in areas where the intensity difference is equal to or greater than the first threshold are combined to obtain combined background image data.

[0025] Here, the first threshold is a value that includes bright spots that exist singly in areas below the first threshold, and bright spots that exist densely in areas above the first threshold. How to determine the first threshold will be described later.

[0026] 4A is a schematic diagram showing differential image data. As shown in FIGS. 1 and 4A, in the process of obtaining differential image data, the intensity difference between the specimen image data and the composite background image data is obtained (S150). Specifically, the difference between the intensity distribution of the specimen image and the intensity distribution of the composite image is obtained. The obtained differential image data is mainly the intensity distribution of the signal light from the labeling substance.

[0027] 4B is a schematic diagram of the difference image data after noise tolerance processing. As shown in FIGS. 1 and 4B, the detection method of this embodiment may further perform noise tolerance processing to remove intensity distributions below a second threshold in the step of obtaining difference image data (S160). The second threshold is a threshold that does not substantially affect the intensity distribution of the signal light from the labeling substance but can remove noise. How to obtain the second threshold will be described later.

[0028] The detection method of this embodiment may further include a step of quantifying the target substance based on the differential image data. For example, a calibration curve showing the relationship between the signal amount and the target substance may be obtained in advance, and the amount of the target substance may be quantified based on the calibration curve. In this case, the target substance may be quantified after performing an expansion process. By performing the expansion process, if the originally extracted bright spot region is narrow and only contains the bright spot peak portion, the signal intensity can be measured more precisely by including more of the bright spot's base portion. For example, the expansion process may be performed using 3 pixels x 3 pixels.

[0029] There are no particular limitations on how the first and second thresholds are determined. The first and second thresholds may be preset values, or may be determined by the following method. Fig. 5 is a flowchart illustrating how the first and second thresholds are determined, and Fig. 6 is a schematic graph showing the relationship between the distance between bright spots and the matching rate. The horizontal axis of Fig. 6 represents the distance between bright spots (µm), and the vertical axis represents the matching rate (%). As shown in Figures 5 and 6, the first threshold and the second threshold were selected by preparing multiple sample image data by combining multiple first images (equivalent to images containing only fluorescence from bright spots with no autofluorescence) containing only labeled substances arranged at different densities and second images (equivalent to images containing only autofluorescence) in which the target substance is not labeled with a labeled substance, reducing the image features of the bright spots for each of the multiple sample image data using a first filter and a second filter, obtaining multiple reduced image data in which the bright spots are extracted using the first threshold and the second threshold, obtaining differential image data which are the differences between the data of the first image data and the multiple reduced images, and comparing the first image data and the multiple differential image data, based on the matching rate between the bright spots in the first image data and the bright spots in the differential image data.

[0030] First, multiple first images containing only labeling substances with different densities are prepared (S210). Specifically, first images in which labeling substances are arranged at a predetermined interval are prepared. For example, first images in which labeling substances containing phosphor-accumulating particles are arranged at intervals of 0.1, 0.3, 0.5, 1.0, 2.0, 3.0, and 10 μm are prepared, respectively. Also, second images of a specimen in which the target substance is not labeled with a labeling substance are prepared (S220). Specifically, data identical to the specimen image data is prepared except that the target substance is not labeled with a labeling substance. Next, multiple first image data and second image data are respectively combined to prepare multiple sample image data (S230).

[0031] Next, differential image data is obtained in the same manner as in the detection method of the present embodiment (S240). Here, the first and second thresholds are set to arbitrary values. Next, the first image and differential image data are compared to determine the match rate between the bright spots in the first image and the bright spots in the differential image data (S250). The match rate is determined for each of the obtained differential image data. The match rate is determined separately for single bright spots and for densely packed bright spots. Based on the results, the first and second thresholds are changed to determine thresholds with the highest match rate or the smallest variation in the match rate (S260). If the match rate or the variation in the match rate is within a predetermined range, the first and second thresholds are used as the thresholds in the detection method (S260; Yes). Alternatively, the first and second thresholds with the highest match rate and the smallest variation in the match rate may be determined. For example, thresholds may be selected such that the average match rate for each distance between bright spots is 50% or higher and the variation in the match rate is smallest. On the other hand, if the match rate or the variation in match rate is outside the predetermined range, the first threshold or the second threshold is changed and the difference image data is calculated again (S260; No). For example, in the example shown in Figure 6, the threshold conditions indicated by the white circle symbols are preferable. The first threshold and the second threshold are used as the thresholds in the detection method.

[0032] (effect) As described above, the detection method according to the present invention can detect both a single bright spot and a group of densely packed bright spots, thereby improving detection accuracy.

[0033] In this embodiment, the second background image data is obtained after the first background image data is obtained, but the first background image data may be obtained after the second background image data is obtained, or the first background image data and the second background image data may be obtained simultaneously.

[0034] In addition, in this embodiment, median filters are used as the first and second filters, but frequency filters may be used as the first or second filter. Even in this case, the detection method of the present invention can properly detect both a single bright spot and a group of bright spots that are densely packed together. [Industrial Applicability]

[0035] The detection method according to the present invention is useful for, for example, pathological diagnosis of cancer and the like.

Claims

1. preparing specimen image data of a specimen in which a target substance is labeled with a labeling substance; obtaining first background image data by reducing image features of single bright spots in the sample image data using a first filter; obtaining second background image data by reducing image feature amounts of densely packed bright spots in the sample image data using a second filter; a step of synthesizing the first background image data and the second background image data based on brightness information of the sample image data to obtain synthesized background image data; determining difference image data that is the difference between the sample image data and the composite background image data; A detection method comprising:

2. The detection method according to claim 1 , further comprising the step of quantifying the target substance based on the difference image data.

3. the image feature is a light intensity distribution, The light intensity distribution in the sample image data reduced by the first filter and the light intensity distribution in the sample image data reduced by the second filter differ in peak width in the intensity distribution, peak height in the intensity distribution, or both the peak width and the peak height. The detection method according to claim 1 .

4. The detection method according to claim 1 , wherein at least one of the first filter and the second filter is a median filter.

5. The detection method of claim 1 , wherein at least one of the first filter and the second filter is a frequency filter.

6. the image feature is a light intensity distribution, In the step of obtaining the composite background image data, an intensity difference is obtained, which is a difference between a second intensity distribution of light in the second background image data and a first intensity distribution of light in the first background image data, and in a region where the intensity distribution difference is less than a first threshold, the first background image data is used, and in a region where the intensity distribution difference is equal to or greater than the first threshold, a composite intensity distribution is obtained using the second background image data. The detection method according to claim 1 .

7. The detection method according to claim 6 , wherein the step of obtaining the difference image data further includes a noise tolerance process for excluding an intensity distribution below a second threshold.

8. The detection method according to claim 2 , wherein the step of quantifying the target substance further comprises performing an expansion treatment.

9. The first threshold value and the second threshold value are preparing a plurality of sample image data by combining a plurality of first image data of a specimen containing only a labeling substance arranged at different densities with a plurality of second image data of a specimen obtained without labeling a target substance with a labeling substance; For each of the plurality of sample image data, the image feature amount of the bright spot is reduced using the first filter and the second filter, and a plurality of reduced image data in which the bright spot is extracted using the first threshold value and the second threshold value is obtained; obtaining the difference image data, which is the difference between the first image data and the plurality of reduced image data; the first image data and the plurality of differential image data are compared, and the selected image data is selected based on a matching rate between the bright spots in the first image data and the bright spots in the differential image data; The detection method according to claim 7.

Images