WO2020054080A1 - Chromosomal abnormality identification device, chromosomal abnormality identification system, measurement device, and analysis device - Google Patents

Abstract

The purpose of this invention is to provide a chromosomal abnormality identification device capable of using chromosome color information for chromosome inspection. To provide such a device, a chromosome data storage device is made to have a database comprising feature information, which includes image data for a spectral feature value image obtained through spectral analysis of the spectrums of light that has passed through a dyed chromosome and feature values obtained through image analysis of the spectral feature value image, and information indicating feature information and whether a chromosome having the feature information is normal. The spectrums of light that has passed through a dyed chromosome to be subjected to identification are acquired for each point on the chromosome as spectral cube data (step S11). Feature information is extracted from the spectral cube data (steps S12, S13). An abnormality in the chromosome to be subjected to identification is identified on the basis of the extracted feature information and the feature information stored in the database (step S14).

Description

Chromosome abnormality determination device, chromosome abnormality determination system, measurement device, and analysis device

<< The present invention relates to a chromosome abnormality judging device, a chromosome abnormality judging system, a measuring device, and an analyzing device.

The work of obtaining images of chromosomes using an optical microscope and judging a number of chromosomal abnormalities from image information is important in the prevention and treatment of diseases caused by chromosomal abnormalities and in tests for determining substances that induce chromosomal abnormalities. Plays a role.
A chromosome is a biological material in a cell nucleus that holds genetic information, and there are a total of 46 chromosomes including 44 autosomes and 2 sex chromosomes per human cell nucleus. There are 22 types of autosomes, each of which is composed of two chromosomes, and the autosomes are distinguished by numbers 1 to 22. Sex chromosomes are distinguished by X and Y, with females having two Xs and males having one X and Y each.

FIG. 15 is a schematic diagram illustrating an example of a chromosome. As shown in FIG. 15, the chromosome is constricted at a portion called centromere 101 and is bisected by centromere 101. The bisected short part is called the short arm 102, and the long part is called the long arm 103. When the chromosome is dyed by the dye, stripes are dyed on the chromosome. This stripe is called a band 104.
When a chromosome abnormality is visually diagnosed by an optical microscope image, the determination is generally performed as follows. That is, the relative length of the chromosome region, the ratio of the length of the short arm and the long arm in the chromosome region, the arrangement of the bands in the chromosome region, i.e., from the image features of the chromosome image such as the band pattern, identification of the type of chromosome Chromosomal abnormalities such as chromosome abnormalities, loss of genetic information, duplication, translocation, etc. are determined.

For example, Patent Literature 1 proposes a chromosome testing device that images a chromosome to be tested, extracts morphological features from the obtained image, and then tests the chromosome based on the result. This chromosome testing device includes a chromosome image database storing chromosome image data, a means for searching the chromosome database for chromosome image data similar to the extracted morphological features, a similar chromosome image data retrieved and a test target. Means for outputting and displaying a captured image of a chromosome.

JP-A-3-123860

By the way, in the above-described apparatus for automatically performing the chromosome test, the test is performed using only the morphological features obtained from the chromosome image data. In addition, it is generally known that chromosome bands have shading, and that not only band patterns but also shading information of each band can be used as a criterion for chromosome abnormality determination, and chromosome color information performs chromosome identification and abnormality determination. It can be one of the important judgment factors at the time.
In an apparatus or the like that automatically performs a chromosome test, the higher the determination accuracy is, the more preferable it is. The improvement in the determination accuracy leads to improvement in usability and versatility of the chromosome test apparatus.

Therefore, the present invention has been made in view of the above-mentioned conventional problems, and a chromosome abnormality determination device, a chromosome abnormality determination system, and a chromosome abnormality determination system capable of performing a chromosome test using chromosome color information obtained by staining. It is intended to provide an apparatus and an analysis apparatus.

According to one aspect of the present invention, an abnormality determination process constructed using feature information representing a feature of a chromosome included in spectral cube data acquired from at least one of an abnormal chromosome and a normal chromosome is executed. Abnormality determination processing unit, and a data acquisition unit that acquires the spectral cube data of the chromosome to be determined, the spectral cube data is a spectrum at each point on the chromosome of light transmitted or reflected through a stained chromosome. A spectrum, the abnormality determination processing unit inputs the characteristic information included in the spectrum cube data of the determination target chromosome acquired by the data acquisition unit, and executes abnormality determination processing for the determination target chromosome using the characteristic information And a chromosome abnormality judging device characterized in that:

Further, according to another aspect of the present invention, a plurality of measurement devices that acquire spectrum cube data of a chromosome to be determined, and an analysis device that performs an abnormality determination of the chromosome to be determined based on the spectrum cube data are connected to a network. A chromosome abnormality determination system connected via the, the measurement device, the data to obtain the spectral spectrum at each point on the chromosome of the determination target of the light transmitted or reflected the stained chromosome of the determination target as spectrum cube data An acquisition unit and a first communication processing unit that transmits the spectrum cube data acquired by the data acquisition unit to the analysis device via the network and receives an abnormality determination result of the analysis device with respect to the transmitted spectrum cube data via the network The analysis device receives the spectrum cube data from the measurement device. Network, and a second communication processing unit that transmits an abnormality determination result in the analysis device for the received spectrum cube data to the measurement device via the network, and at least one of an abnormal chromosome and a normal chromosome An abnormality determination processing unit that executes abnormality determination processing constructed using characteristic information representing chromosome characteristics included in spectrum cube data obtained from the characteristic cube included in spectral cube data of a chromosome to be determined And a chromosome abnormality determination system comprising:

According to another aspect of the present invention, a measurement device that acquires spectrum cube data of a chromosome to be determined and transmits the spectrum cube data to an analyzer that performs an abnormality determination of a chromosome connected via a network. There, a data acquisition unit that acquires a spectral spectrum at each point on the chromosome to be determined of light transmitted or reflected by the chromosome to be stained to be determined as the spectrum cube data, and spectral cube data acquired by the data acquisition unit. And a first communication processing unit for transmitting, via the network, an abnormality determination result of the transmitted spectrum cube data to the analysis device via the network.

Furthermore, according to another aspect of the present invention, there is provided an analysis device for performing abnormality determination of a chromosome to be determined based on spectral cube data of a chromosome to be determined, wherein the spectral cube data of the chromosome to be determined is transmitted via a network. A second communication processing unit that receives and transmits an abnormality determination result in the analysis device for the received spectrum cube data via a network, and the spectrum cube obtained from at least one of an abnormal chromosome and a normal chromosome An abnormality determination processing unit configured to execute abnormality determination processing constructed using characteristic information representing characteristics of chromosomes included in the data using characteristic information included in the spectrum cube data of the chromosome to be determined. An apparatus is provided.

According to one embodiment of the present invention, the chromosome test is performed using the chromosome color information, whereby the accuracy of determining a chromosome abnormality can be improved.

It is a lineblock diagram showing an example of a chromosome abnormality judging device concerning a first embodiment of the present invention. It is an example of the captured image of the chromosome. It is an example of the spectrum of the transmittance | permeability in the measurement point X of FIG. FIG. 4 is an explanatory diagram for explaining a spectral feature image. FIG. 4 is a characteristic diagram illustrating an example of a relationship among wavelength, transmittance, and similarity. FIG. 4 is a characteristic diagram illustrating an example of a relationship between similarity and transmittance. It is an example of a single wavelength image. FIG. 2 is an example of a functional block diagram of a determination processing device according to the first embodiment. It is a flowchart which shows an example of the processing procedure at the time of model learning. It is a flowchart which shows an example of the processing procedure at the time of abnormality determination processing. It is an example of a functional block diagram of a judgment processing device in a second embodiment. It is a flow chart which shows an example of the processing procedure at the time of building a neural network model in a second embodiment. It is a flow chart which shows an example of a processing procedure at the time of abnormal judgment processing in a second embodiment. It is an example of a functional block diagram of a judgment processing device in a third embodiment. It is a schematic diagram which shows an example of a chromosome. FIG. 9 is a configuration diagram illustrating an example of a modification of the chromosome abnormality determination device. It is a lineblock diagram showing an example of the chromosome abnormality judging system concerning a 4th embodiment. It is a lineblock diagram showing an example of a measuring device. FIG. 2 is a configuration diagram illustrating an example of an analysis device. FIG. 2 is an example of a functional block diagram of a determination processing device according to the first embodiment. It is a flowchart which shows an example of the processing procedure at the time of model learning. It is a flowchart which shows an example of the processing procedure of the measuring device at the time of abnormality determination processing. 9 is a flowchart illustrating an example of a processing procedure of the analysis device at the time of abnormality determination processing. It is an example of a functional block diagram of a judgment processing device in a second embodiment. It is a flow chart which shows an example of the processing procedure at the time of building a neural network model in a second embodiment. It is a flow chart which shows an example of a processing procedure of an analysis device at the time of abnormal judgment processing in a second embodiment. It is an example of a functional block diagram of a judgment processing device in a third embodiment. It is a lineblock diagram showing an example of a modification of a chromosome abnormality judging system.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
It is noted that in the following detailed description, numerous specific specific configurations are set forth in order to provide a thorough understanding of embodiments of the present invention. However, it is clear that other embodiments can be implemented without being limited to such specific specific configurations. Further, the following embodiments do not limit the invention according to the claims. In addition, not all combinations of the features described in the embodiments are necessarily essential to the solution of the invention.

First, a first embodiment will be described.
FIG. 1 is a configuration diagram illustrating an example of a chromosome abnormality determination device according to the first embodiment of the present invention.
The chromosome abnormality determination device 1 includes a microscope 2, an electric stage 3 for the microscope 2, an imaging spectroscope 4, an image memory 5, a spectrum analysis device 6, an image analysis device 7, a determination processing device 10, a database, A creator (database creator) 11, a chromosome data storage device 12, and a display device 13 are provided. The image memory 5, the spectrum analysis device 6, the image analysis device 7, the determination processing device 10, the database creation device 11, the chromosome data storage device 12, and the display device 13 include, for example, an input device, a display device, and a storage device. It consists of one personal computer.

The imaging spectroscope 4 receives the reflected light or transmitted light from the chromosome through the microscope 2, condenses the light linearly by a lens and a slit, and obtains position information in one-dimensional direction and spectral information in another one-dimensional direction. The light is separated so that the image is captured by a two-dimensional camera. This captured image is a set of spectra at each measurement point, and each spectrum shows a waveform including color information at each point of the stained chromosome. By moving the electric stage 3 to image the object to be measured, it is possible to obtain spectrum cube data in which the two-dimensional image of the object to be measured is added with the spectrum axis at each measurement point.

The waveform containing the color information is the reflectance and transmittance obtained by removing the influence of the variation in the spectral information of the light source and the measurement sensitivity of the camera element from the image signal captured as the reflection intensity or transmission intensity from the chromosome. Can be obtained in the form of
For example, in one chromosome shown in FIG. 2, the transmittance spectrum at the measurement point X shows a waveform as shown in FIG. In FIG. 3, the horizontal axis represents wavelength (nm) and the vertical axis represents transmittance.
The spectrum cube data captured by the imaging spectroscope 4 is stored in the image memory 5.
The spectrum analysis device 6 performs spectrum analysis on the spectrum cube data stored in the image memory 5 to extract a spectrum feature image, and outputs image data of the spectrum feature image. The image data of the spectral feature image output from the spectrum analyzer 6 is stored in the image memory 5.

Here, the spectral feature image is, for example, a monochrome image composed of chromosome transmission intensity information at each measurement point at an arbitrary wavelength in spectral cube data represented as shown in FIG. That is, it refers to a single wavelength image. Instead of a single-wavelength image at an arbitrary wavelength, a single-wavelength image may be obtained from an integrated value of transmission intensity at a predetermined range of wavelengths before and after including an arbitrary wavelength, or an integrated value of transmission intensity at all wavelengths. .
Further, as shown in FIG. 4A, as the spectral feature image, each of the measurement points at three wavelengths corresponding to, for example, R (red, G (green), and B (blue)) in the spectral cube data A color image obtained from the transmission intensity information from the chromosome in the above may be extracted.

In addition, color calculation is performed based on transmission intensity information at each measurement point of the spectrum cube data, and color information such as L ^* , a ^* , b ^* , X, Y, Z, etc. is obtained, and FIG. As shown in ()), for example, the values of L ^* , a ^* , and b ^* are obtained for each measurement point. Based on this color information, a color image is obtained by assigning L ^* , a ^* , and b ^* values to RGB for each measurement point and forming an image, and the obtained color image is used as a spectral feature image. Alternatively, for each measurement point, a monochrome image may be obtained by forming an image based on any of the color information, and this monochrome image may be used as the spectral feature amount image. The color information is not limited to L ^* , a ^* , b ^* , X, Y, Z, etc., and can be converted into a color image or a monochrome image based on the obtained color information. , May be a value obtained by a spectrum operation, that is, a value capable of generating a two-dimensional image from spectrum information obtained from spectrum cube data.

Here, the transmission spectrum of the part where the chromosome does not exist is used as the reference spectrum, and the transmission spectrum of the part where the chromosome exists is used as the evaluation spectrum. Also, each of the reference spectrum and the evaluation spectrum is considered as a vector, and the similarity between the two is detected. This similarity is calculated using, for example, a cosine similarity that evaluates the similarity by the sum of inner products of two vectors. Since the similarity indicates the similarity between the transmission spectrum of the part where the chromosome does not exist and the transmission spectrum of the part where the chromosome exists, when the similarity is large, the part where the chromosome exists becomes relatively lighter in color and the similarity is small Sometimes the chromosome is relatively dark.
On the other hand, a single-wavelength image obtained as a spectral feature image from spectral cube data has different transmittance depending on the wavelength. The smaller the transmittance, the higher the chromosome is displayed in the single-wavelength image, and the higher the transmittance, the thinner the chromosome is displayed. .

FIG. 5 shows a change in transmittance of a certain chromosome with respect to a change in wavelength for each similarity. In FIG. 5, the horizontal axis is the wavelength, the vertical axis is the spectrum of the transmittance, and the change width of the similarity is constant. In FIG. 5, a characteristic L1 corresponds to a reference spectrum, and characteristics L2 to L10 correspond to transmission spectra (evaluation spectra). FIG. 5 shows that in the range of wavelengths from about 500 nm to about 700 nm, the characteristic L2 has the largest similarity to the characteristic L1, the similarity decreases in the order of the characteristics L2, L3,..., And the characteristic L10 has the smallest similarity. Is represented.

FIG. 6 shows changes in transmittance with respect to changes in similarity for three wavelengths based on the characteristics shown in FIG.
As shown in FIG. 6, the transmittance decreases as the similarity decreases. Therefore, it can be seen that the lower the similarity, the deeper the chromosomes are displayed in the single-wavelength image obtained from the spectrum cube data. In addition, FIG. 5 shows that the transmittance becomes the smallest when the wavelength is about 550 nm in any case of the similarity, and the chromosomes are displayed dark in the single-wavelength image.

In the case of a wavelength of about 550 nm at which the transmittance becomes minimum, the transmittance changes in a relatively low range as shown by the characteristic L11 in FIG. Thus, the transmittance changes relatively largely, and in a range where the similarity is relatively low, the change in transmittance is small with respect to the change in similarity. Therefore, as shown in FIG. 7A, the chromosome is displayed relatively dark in the single-wavelength image of 550 nm, and the contrast between the chromosome and its surroundings is high, but the band stained on the chromosome is The contrast with the surroundings is low, and it is difficult to identify the band.

Conversely, when the wavelength is about 805 nm, as shown by the characteristic L13 in FIG. 6, the change in the transmittance with respect to the change in the similarity is small and the transmittance changes in a relatively high range at any similarity. Therefore, as shown in FIG. 7C, the chromosome is displayed relatively thin in the 805 nm single-wavelength image, the contrast between the chromosome and its surroundings is low, and the contrast between the band and its surroundings is also low. It is difficult to distinguish not only chromosomes but also bands on chromosomes.

On the other hand, when the wavelength is about 650 nm, as shown by the characteristic L12 in FIG. 6, the transmittance changes in a certain range with respect to the change in similarity over the entire similarity range. Therefore, as shown in FIG. 7 (b), the contrast between the chromosome and its surroundings and the contrast between the band and its surroundings are both high in the single wavelength image of 650 nm, and the shapes of the chromosomes and bands are displayed relatively clearly. can do.
In other words, in order to clearly display chromosomes and bands in a single-wavelength image, as shown in FIG. 6, the chromosome density (that is, similarity) and the amount of light transmitted through the chromosome (transmittance) vary over the entire similarity. It is understood that it is preferable that the transmittance is not too large or too small, has a certain size, and that the change width of the transmittance with respect to the change of the similarity has a certain width and is relatively constant. .

Therefore, in the case of spectral cube data having the characteristics shown in FIG. 4, a wavelength satisfying such conditions is obtained. In the case of FIG. 4, a single-wavelength image at 650 nm is acquired and used as a spectral feature image. Thus, the analysis by the image analysis device 7 that performs processing based on the spectral feature amount image and the determination by the determination processing device 10 can be performed with higher accuracy.
The analysis wavelength for obtaining a single-wavelength image is obtained in advance and stored in a predetermined storage area, and when performing spectrum analysis, the stored analysis wavelength is read out and read. Alternatively, a wavelength at which a good single-wavelength image can be obtained may be detected before spectral analysis is performed. Also, here, a case has been described in which a single-wavelength image at a wavelength of 650 nm is obtained and spectrum analysis is performed based on the image. For example, among the spectral cube data, spectral cube data in a wavelength region before and after that including 650 nm is used. Alternatively, the spectrum analysis may be performed.

(4) The image analysis device 7 performs image analysis based on the image data of the spectral feature amount image stored in the image memory 5 and extracted by the spectrum analysis device 6, and extracts a predetermined feature amount. The feature quantity extracted by image analysis includes, for example, the relative length of the chromosome region, the ratio of the length of the short arm to the length of the long arm in the chromosome region, the arrangement of the bands in the chromosome region, that is, the band pattern Etc. are applied.

The determination processing device 10 determines whether the image data of the spectral feature amount image extracted by the spectrum analysis device 6, the feature amount by the image analysis extracted by the image analysis device 7, and the normal or abnormal data stored in the chromosome data storage device 12. By comparing the image data of the spectral feature amount image of the chromosome determined to be a chromosome with the feature amount obtained by image analysis, the abnormality / normality of the chromosome is determined. Further, the determination processing device 10 displays the determination result and the like on the display device 13. After determining whether the chromosome is normal or abnormal, the determination processing device 10 uses the image data of the spectral characteristic amount image used for the determination and the characteristic amount obtained by the image analysis as the characteristic information, and compares the characteristic information with the normal / abnormal determination result. The data is stored in the chromosome data storage device 12 via the database creation device 11 in association with each other.
The chromosome data storage device 12 stores a database in which image data of spectral characteristic amount images and characteristic information including characteristic amounts obtained by image analysis and normal / abnormal judgment results are associated with each other.

FIG. 8 is an example of a functional block diagram of the determination processing device 10 in the first embodiment.
The determination processing device 10 illustrated in FIG. 8 performs chromosome abnormality determination by model learning, and includes an abnormality determination processing unit 21, an abnormality determination parameter determination unit 22, and an abnormality determination result display processing unit 23.
The abnormality determination parameter determination unit 22 determines an abnormality determination parameter used for model learning in the abnormality determination processing unit 21 and notifies the determined abnormality determination parameter to the abnormality determination processing unit 21. The type and number of the feature amounts and the like used as the abnormality determination parameters may be determined manually by the tester, or may be determined automatically based on various information.
The abnormality determination processing unit 21 performs supervised learning using one or a plurality of methods used in machine learning such as a neural network or a support vector machine based on the abnormality determination parameter determined by the abnormality determination parameter determination unit 22. , Construct a classifier for abnormality determination processing.

Then, the abnormality determination processing unit 21 inputs the image data of the spectral characteristic amount image obtained from the chromosome to be determined and the characteristic amount by the image analysis, and performs an abnormality determination process constructed on the basis of the abnormality determination parameter. An abnormality is determined using the model. Further, the abnormality determination processing unit 21 uses the image data of the spectral characteristic amount image of the device under test and the characteristic amount obtained by image analysis as characteristic information, associates this characteristic information with the abnormality determination result at this time, and creates a database creation device. 11 and stored in the chromosome data storage device 12.
The abnormality determination result display processing unit 23 displays the abnormality determination result in the abnormality determination processing unit 21 on the display device 13.

FIG. 9 is a flowchart illustrating an example of a processing procedure at the time of model learning in supervised learning in the chromosome abnormality determination device 1.
At the time of model learning, the abnormality determination parameter determination unit 22 of the determination processing device 10 reads the database stored in the chromosome data storage device 12 (Step S1) and sets the abnormality determination parameters (Step S2). The abnormality determination processing unit 21 builds a model of the abnormality determination process by performing supervised learning of a neural network, a support vector machine, and the like based on the set abnormality determination parameters (step S3. (Corresponding to the determination processing constructing unit), and store it in a predetermined storage area.

FIG. 10 is a flowchart illustrating an example of a processing procedure at the time of abnormality determination processing in the chromosome abnormality determination device 1.
At the time of the abnormality determination process, first, the tester prepares a slide glass on which metaphase cells in which chromosomes can be seen are mounted, and places the slide glass on the electric stage 3. Thereby, the chromosome to be determined on the slide on the motorized stage 3 is enlarged by the microscope 2, spectrally separated by the imaging spectroscope 4 and stored as spectral cube data in the image memory 5 (step S11). (Corresponds to the data acquisition unit).

Next, the spectrum analysis device 6 performs spectrum analysis on the spectrum cube data stored in the image memory 5 to extract a spectrum feature amount image (step S12). Further, the image analysis device 7 performs image analysis on the obtained spectral feature amount image to extract a predetermined feature amount (step S13). In the spectrum analysis performed by the spectrum analyzer 6, the wavelength used as a parameter used when extracting the spectral feature image, such as the wavelength used when a single-wavelength image is extracted as the spectral feature image, and the image analyzed by the image analyzer 7. Various parameters such as a wavelength or a wavelength range used when extracting various characteristic amounts in the analysis are used to perform image extraction using wavelengths set in advance and stored in a predetermined storage area.

Then, the abnormality determination processing unit 21 performs model learning in advance based on the image data of the spectral feature amount image extracted by the spectrum analysis by the spectrum analysis device 6 and the feature amount by the image analysis by the image analysis device 7. The chromosomal abnormality is determined using the abnormality determination model constructed by the above (Step S14).
The abnormality determination result display processing unit 23 displays the abnormality determination result in the abnormality determination processing unit 21 on the display device 13 (step S15), and further, the database creation device 11 causes the image data of the spectral feature amount image used for the abnormality determination to be displayed. Then, the feature amount obtained by the image analysis is associated with the abnormality determination result and stored in the database of the chromosome data storage device 12 (step S16).
Thus, the chromosome abnormality determination processing ends.

Here, the image of the chromosome obtained from the spectrum cube data differs depending on the wavelength as shown in FIG. 7. In the case of the wavelength of 550 nm shown in FIG. 7A and the case of the wavelength of 805 nm shown in FIG. Band cannot be identified, but in the case of the wavelength of 650 nm shown in FIG. 7B, the band can be identified.
In other words, the chromosome abnormality judging device 1 forms a single-wavelength image for a preset wavelength at which a good single-wavelength image with high contrast can be obtained. Therefore, the obtained single-wavelength image, that is, the spectral feature image is an image having a relatively high contrast, and chromosomes and bands can be easily identified.

For example, in image analysis using a binarization process or the like on a captured image of a digital camera, the obtained chromosome image is not entirely black as shown in FIG. As shown in FIG. 7 (c), the obtained chromosome image is so thin that the chromosome itself cannot be identified, and as a result, the chromosome abnormality may not be automatically determined. However, when analysis is performed using a single-wavelength image (spectral feature amount image) acquired by spectrum analysis, chromosome abnormality can be automatically determined. That is, the chromosome abnormality can be automatically determined for more chromosomes, so that usability can be further improved.

In addition, the chromosome abnormality determination apparatus 1 performs abnormality determination on a chromosome spectral feature image and a feature obtained by image analysis using a model of an abnormality determination process obtained by supervised learning. The chromosome abnormality judgment performed by the visual judgment of the house can be automatically performed, and can be sufficiently applied even when the abnormality judgment is performed on a huge amount of chromosome samples.

Next, a second embodiment of the present invention will be described.
The second embodiment differs from the chromosome abnormality judging device 1 in the first embodiment shown in FIG. 1 in that the image analyzing device 7 is omitted, and the configuration of the judgment processing device 10 is different. In the chromosome abnormality judging device 1 according to the second embodiment, the image data of the spectrum feature image extracted by the spectrum analyzing device 6 is directly input to the judgment processing device 10.
The same parts as those in the first embodiment are denoted by the same reference numerals, and detailed description thereof will be omitted.

FIG. 11 is a functional block diagram illustrating an example of the determination processing device 10 according to the second embodiment.
The determination processing device 10 includes an abnormality determination processing unit 21a and an abnormality determination result display processing unit 23.
In the chromosome data storage device 12 in the second embodiment, the image data of the spectral feature image extracted from the spectrum cube data when the chromosome abnormality is determined is stored as characteristic information in the abnormal sample database 12a. Similarly, the image data of the spectrum feature image extracted from the spectrum cube data when the chromosome is determined to be normal is stored as feature information in the normal sample database 12b. The database creation device 11 stores the feature information (that is, the image data of the spectral feature image) in the abnormal sample database 12a or the normal sample database 12b based on the determination result in the abnormality determination processing unit 21a.

The abnormality determination processing unit 21a performs abnormality determination of a chromosome using a neural network model based on supervised learning. That is, the abnormality determination processing unit 21a performs the neural learning by individually performing machine learning on the abnormal sample database 12a storing the characteristic information at the time of abnormality determination and the normal sample database 12b storing the characteristic information at the time of normal determination. Build a network model. Then, by inputting the image data of the spectral feature amount image of the new chromosome to be determined to the abnormality determination processing unit 21a, the image data of the spectral feature amount image of the chromosome to be determined is stored in the chromosome data storage device 12. An abnormality is determined using a neural network model constructed based on the normal or abnormal characteristic information obtained.

FIG. 12 is a flowchart illustrating an example of a processing procedure when constructing a neural network model in the chromosome abnormality determining device 1.
When constructing the neural network model, the abnormality determination processing unit 21a of the determination processing device 10 reads the abnormal sample database 12a and the normal sample database 12b stored in the chromosome data storage device 12 (Step S21), and reads the abnormal sample database 12a and the normal sample. Machine learning is individually performed on the abnormal and normal feature information stored in the database 12b to construct a neural network model (step S22). Then, the constructed neural network model is stored in a predetermined storage area.

FIG. 13 is a flowchart illustrating an example of a processing procedure at the time of abnormality determination processing in the chromosome abnormality determination device 1.
At the time of the abnormality determination process, first, the tester prepares a slide glass on which metaphase cells in which chromosomes can be seen are mounted, and places the slide glass on the electric stage 3. As a result, the chromosome to be determined on the slide on the electric stage 3 is enlarged by the microscope 2 and separated by the imaging spectroscope 4 and stored in the image memory 5 as spectrum cube data (step S31).

(4) Next, the spectrum analyzing device 6 performs a process of extracting a spectral feature image from the spectral cube data stored in the image memory 5 (step S32). Various parameters required for extracting the spectral feature image, such as an arbitrary wavelength when a single-wavelength image is extracted as a spectral feature image in the spectrum analysis by the spectrum analyzer 6, are set in advance and stored in a predetermined storage area. The spectral feature image is extracted using the stored wavelength.

The abnormality determination processing unit 21a performs chromosome abnormality determination using a neural network model constructed in advance based on the image data of the spectral feature image extracted by the spectrum analyzer 6 (step S33).
The abnormality determination result display processing unit 23 displays the abnormality determination result in the abnormality determination processing unit 21a on the display device 13 (step S34), and further, the database creation device 11 uses the image of the spectral feature amount image used for the abnormality determination. The data is stored in the database of the chromosome data storage device 12 in association with the result of the abnormality determination (step S35).
Thus, the chromosome abnormality determination processing ends.

As described above, the chromosome abnormality judging device 1 according to the second embodiment performs an abnormality judgment on a spectral feature amount image acquired from spectrum cube data using a neural network model, and thus cannot be obtained only with a two-dimensional image. An abnormality determination can be performed using the characteristic information. Therefore, the chromosome abnormality judging device 1 according to the second embodiment can also automatically judge a chromosome abnormality for more chromosome images, and can realize the chromosome abnormality judging device 1 with improved usability.

In the second embodiment, a case has been described in which machine learning is individually performed on characteristic information at the time of abnormality and normal time stored in the abnormal sample database 12a and the normal sample database 12b to construct a neural network model. However, it is not limited to this. Machine learning is performed on one of the abnormal feature information stored in the abnormal sample database 12a or the normal feature information stored in the normal sample database 12b, and a neural network model based on the abnormal feature information or It is also possible to construct a neural network model based on characteristic information in a normal state, and determine a chromosome abnormality based on this.

Next, a third embodiment of the present invention will be described.
The third embodiment differs from the chromosome abnormality judging device 1 in the first embodiment shown in FIG. 1 in that the image analysis device 7 is omitted, and the configuration of the judgment processing device 10 is different.
FIG. 14 is a functional block diagram illustrating an example of the determination processing device 10 according to the third embodiment.
The image data of the spectrum feature image extracted by the spectrum analysis in the spectrum analysis device 6 is directly input to the determination processing device 10 in the third embodiment.
The determination processing device 10 according to the third embodiment includes an abnormality determination processing unit 21c and an abnormality determination result display processing unit 23.
In the chromosome data storage device 12 in the third embodiment, the spectral feature image and the determination result of normal or abnormal are stored in the chromosome data storage device 12 via the database creation device 11 in association with each other.

The abnormality determination processing unit 21c performs chromosome abnormality determination using a neural network model based on unsupervised learning. That is, the abnormality determination processing unit 21c constructs a neural network model by performing machine learning using a database including image data of the spectral feature amount image extracted by the spectrum analysis device 6. Then, the abnormality determination processing unit 21c inputs the image data of the spectral feature amount image extracted by the spectrum analysis to the new chromosome to be determined to the abnormality determination processing unit 21c. An abnormality determination is performed on the image data of the image using the constructed neural network model.

In the chromosome abnormality judging device 1 in the third embodiment, the processing procedure at the time of model construction is basically the same as the processing procedure at the time of model construction in the second embodiment shown in FIG. In, when constructing a neural network model (step S22), machine learning is performed using the image data of the spectral feature image at abnormal time and normal time stored in the database of the chromosome data storage device 12, and the neural network Build the model.

Further, in the chromosome abnormality judging device 1 in the third embodiment, the processing procedure at the time of the abnormality judgment processing is basically the same as the processing procedure at the time of the abnormality judgment processing in the second embodiment shown in FIG. In the embodiment, when performing the spectrum analysis (step S32), a predetermined spectral feature image is extracted, and an abnormality is determined for each extracted spectral feature image using the constructed neural network model.
As described above, since the abnormality determination is performed using the neural network model on the image data of the spectral feature image obtained by the spectrum analysis, the abnormality determination is performed using the feature information that cannot be obtained only with the two-dimensional image. Can be. Therefore, also in this case, the chromosome abnormality can be automatically determined for more chromosome images, and the usability of the chromosome abnormality determination device 1 can be improved.

In each of the above embodiments, the case where the reflected light or the transmitted light of the chromosome is separated into a large number of wavelengths using the imaging spectroscope 4 to obtain the spectrum cube data has been described, but the present invention is not limited to this. From the reflected light or transmitted light of the chromosome, a predetermined single wavelength, a predetermined two wavelengths, a predetermined three wavelengths, or the like, may be used to extract only specific wavelengths to obtain image data. Image data may be obtained by extracting only the minimum number of wavelengths necessary for the image processing. In this case, the light may be separated using an optical system such as a prism, or only a predetermined wavelength may be extracted using a filter. Further, for example, only the wavelengths corresponding to R, G, and B may be extracted. In this case, by providing an image sensor having a filter for extracting only the wavelengths corresponding to R, G, and B, only the wavelengths corresponding to RGB may be extracted.

FIG. 16 shows an example of a schematic configuration of a chromosome abnormality judging device 1a configured to extract only wavelengths necessary for obtaining a spectral feature image using a filter and obtain image data.
That is, the chromosome abnormality judging device 1a is different from the chromosome abnormality judging device 1 shown in FIG. 1 in that, instead of the imaging spectroscope 4, a monochrome camera 14 and a filter provided in the monochrome camera 14 and having a characteristic of transmitting only a specific wavelength are provided. 15 is provided. With such a configuration, it is possible to acquire image data in which only a specific wavelength is extracted from the reflected light or transmitted light of the chromosome.
In this manner, when image data of an arbitrary wavelength is acquired and an abnormality determination of a chromosome is performed using a neural network model, an abnormality determination is performed by constructing a neural network model corresponding to the extracted wavelength. What should I do?

Next, a fourth embodiment of the present invention will be described.
In the chromosome abnormality determination system according to the fourth embodiment, in the chromosome abnormality determination device 1 according to the first embodiment, the processing for acquiring the spectrum cube data and the processing for analyzing the spectrum cube data are performed by different apparatuses. It is something to do.
FIG. 17 is a configuration diagram illustrating an example of a chromosome abnormality determination system according to the fourth embodiment of the present invention.
The chromosome abnormality determining system 1001 includes a plurality of measurement devices 1002 and an analysis device 1003, and the plurality of measurement devices 1002 and the analysis device 1003 are connected via a communication line 1004 such as a LAN (Local Area Network). .
The plurality of measurement devices 1002 have the same functional configuration, and as shown in FIG. 18, a microscope 1201, an electric stage 1202 for the microscope 1201, an imaging spectrograph 1203, an image memory 1204, a communication processing device ( (Corresponding to the first communication processing unit) 1205 described in the claims.

The microscope 1201, the motorized stage 1202, the imaging spectroscope 1203, and the image memory 1204 have the same functional configuration as the microscope 2, the motorized stage 3, the imaging spectroscope 4, and the image memory 5 in the first embodiment, respectively.
That is, the imaging spectroscope 1203 enters the reflected light or transmitted light from the chromosome through the microscope 1201, condenses it linearly by a lens and a slit, and performs positional information in one-dimensional direction and spectral information in another one-dimensional direction. The light is separated so that it becomes information, and an image is taken with a two-dimensional camera. This captured image is a set of spectra at each measurement point, and each spectrum shows a waveform including color information at each point of the stained chromosome. By moving the motorized stage 1202 to image the object to be measured, it is possible to obtain spectral cube data in which a two-dimensional image of the object to be measured is added with a spectral axis at each measurement point.
The waveform containing the color information is the reflectance and transmittance obtained by removing the influence of the variation in the spectral information of the light source and the measurement sensitivity of the camera element from the image signal captured as the reflection intensity or transmission intensity from the chromosome. Can be obtained in the form of

The spectral cube data captured by the imaging spectroscope 1203 is stored in the image memory 1204.
The communication processing device 1205 is configured by, for example, a personal computer, transmits the spectrum cube data stored in the image memory 1204 to the analysis device 1003 via the communication line 1004, and performs an abnormality on the transmitted spectrum cube data in the analysis device 1003. Upon receiving the determination result, the abnormality determination result is stored in the storage device 1205a and displayed on the display device 1205b.

As shown in FIG. 19, the analysis device 1003 includes a communication processing device (corresponding to a second communication processing unit described in the claims) 1301, an image memory 1302, a spectrum analysis device 1303, and an image analysis device 1304. , A determination processing device 1305, a database creation device (corresponding to a database creation unit described in the claims) 1306, a chromosome data storage device 1307, a display device 1308, and a determination result storage device 1309. The image memory 1302, the spectrum analysis device 1303, the image analysis device 1304, the determination processing device 1305, the database creation device 1306, the chromosome data storage device 1307, the display device 1308, and the determination result storage device 1309 are, for example, an input device, a display device, and a storage device. It is composed of one personal computer equipped with the device.

The image memory 1302, the spectrum analysis device 1303, the image analysis device 1304, the determination processing device 1305, the database creation device 1306, the chromosome data storage device 1307, and the display device 1308 are respectively the image memory 5, the spectrum analysis device in the first embodiment. 6, has the same functional configuration as the image analysis device 7, the determination processing device 10, the database creation device 11, the chromosome data storage device 12, and the display device 13.
The communication processing device 1301 exchanges data with each of the measuring devices 1002 via the communication line 1004, and associates the spectrum cube data received from the measuring device 1002 with the identification information for specifying the measuring device 1002 as the transmission source. To be stored. Further, the communication processing device 1301 reads out the determination result of the determination processing device 1305 with respect to the spectrum cube data received from each measurement device 1002 from the determination result storage device 1309 and transmits it to the corresponding measurement device 1002. The transmission of the determination result by the communication processing device 1301 may be performed each time the determination result is obtained by the determination processing device 1305, or may be performed when the determination result for a series of determinations is obtained. Good. Further, when receiving a transmission request from the measurement device 1002, the communication processing device 1301 may read out the determination result stored in the determination result storage device 1309 and transmit it.

A spectrum analysis device (corresponding to a feature information extraction unit described in the claims) 1303 performs spectrum analysis on the spectrum cube data stored in the image memory 1302 to extract a spectrum feature amount image, and The image data of the quantity image is output. The image data of the spectral feature image output from the spectrum analyzer 1303 is stored in the image memory 1302.
The image analysis device 1304 performs image analysis based on the image data of the spectral feature amount image stored in the image memory 1302 and extracted by the spectrum analysis device 1303, and extracts a predetermined feature amount. The feature quantity extracted by image analysis includes, for example, the relative length of the chromosome region, the ratio of the length of the short arm to the length of the long arm in the chromosome region, the arrangement of the bands in the chromosome region, that is, the band pattern Etc. are applied.

The determination processing device 1305 includes the image data of the spectral feature amount image extracted by the spectrum analysis device 1303, the feature amount by the image analysis extracted by the image analysis device 1304, and the normal or abnormal data stored in the chromosome data storage device 1307. By comparing the image data of the spectral feature amount image of the chromosome determined to be a chromosome with the feature amount obtained by image analysis, the abnormality / normality of the chromosome is determined. Further, the determination processing device 1305 displays the abnormality determination result and the like on the display device 1308, and stores the determination result in the determination result storage device 1309. After determining whether the chromosome is normal or abnormal, the determination processing device 1305 uses the image data of the spectral characteristic amount image used for the determination and the characteristic amount obtained by the image analysis as characteristic information, and compares the characteristic information with the determination result of normality or abnormality. The data is stored in the chromosome data storage device 1307 via the database creation device 1306 in association with each other.

The chromosome data storage device 1307 stores a database in which feature information including image data of spectral feature amount images and feature amounts obtained by image analysis and normal / abnormal determination results are associated with a plurality of chromosomes. A database is created for each measurement device 1002, and stored in the corresponding database for each measurement device 1002 that has transmitted the spectrum cube data corresponding to the spectral feature image.

FIG. 20 is an example of a functional block diagram of the determination processing device 1305 according to the fourth embodiment.
The determination processing device 1305 shown in FIG. 20 performs chromosome abnormality determination by model learning. The functional block diagram of the determination processing device 1305 is the same as the functional block diagram of the determination processing device 10 in the first embodiment shown in FIG. It is the same, and includes an abnormality determination processing unit 21, an abnormality determination parameter determination unit 22, and an abnormality determination result display processing unit 23.

FIG. 21 is a flowchart illustrating an example of a processing procedure at the time of model learning in supervised learning in the analysis device 1003.
At the time of model learning, in the determination processing device 1305 shown in FIG. 19, the abnormality determination parameter determination unit 22 reads the database stored in the chromosome data storage device 1307 (step S101). At this time, the abnormality determination parameter determination unit 22 reads the database corresponding to the measurement device 1002 that is the transmission source of the spectrum cube data to be subjected to the abnormality determination. Then, an abnormality determination parameter is set (step S102). The abnormality determination processing unit 21 performs supervised learning of a neural network, a support vector machine, or the like based on the set abnormality determination parameters to construct a model of the abnormality determination process (Step S103. (Corresponding to the determination processing constructing unit), and store it in a predetermined storage area.

Next, a processing procedure for performing chromosome abnormality determination in the chromosome abnormality determination system 1001 will be described with reference to FIGS. FIG. 22 is a flowchart illustrating an example of a processing procedure at the time of measurement in the measurement apparatus 1002, and FIG. 23 is a flowchart illustrating an example of a processing procedure at the time of abnormality determination in the analysis apparatus 1003.
When performing chromosome abnormality determination, first, the tester prepares a slide glass on which metaphase cells in which chromosomes can be seen are mounted, and places the slide glass on the motorized stage 1202. As a result, the chromosome to be determined on the slide on the motorized stage 1202 is enlarged by the microscope 1201 and separated by the imaging spectroscope 1203 and stored as spectral cube data in the image memory 1204 (step S201). (Corresponds to the data acquisition unit).

Subsequently, the tester operates the communication processing device 1205 to transmit the spectrum cube data stored in the image memory 1204 to the analysis device 1003 via the communication line 1004. Thus, the spectrum cube data stored in the image memory 1204 is transmitted to the analysis device 1003 via the communication line 1004 together with the identification information for specifying the measurement device 1002 (Step S202).
Subsequently, when the communication processing device 1205 receives the abnormality determination result from the analysis device 1003 via the communication line 1004 (step S203), the received abnormality determination result is displayed on the display device 1205b and stored in the storage device 1205a. .
The tester can recognize the presence or absence of a chromosome abnormality by referring to the abnormality determination result displayed on the display device 1205b.
On the other hand, when the analysis device 1003 receives the spectrum cube data from the measurement device 1002 via the communication line 1004, the communication processing device 1301 stores the received spectrum cube data in the image memory 1302 (step S301).

Subsequently, spectrum analysis is performed on the spectrum cube data stored in the image memory 1302 by the spectrum analyzer 1303 to extract a spectrum feature image (step S302). The image analysis device 1304 performs image analysis on the obtained spectral feature amount image to extract a predetermined feature amount (step S303). In the spectrum analysis by the spectrum analysis device 1303, the wavelength as a parameter used when extracting the spectral feature image, such as the wavelength when extracting a single wavelength image as the spectrum feature image, and the image by the image analysis device 1304. Various parameters such as a wavelength or a wavelength range used when extracting various characteristic amounts in the analysis are used to perform image extraction using wavelengths set in advance and stored in a predetermined storage area.
Then, the abnormality determination processing unit 21 of the determination processing device 1305 determines in advance based on the image data of the spectral feature amount image extracted by the spectrum analysis by the spectrum analysis device 1303 and the feature amount by the image analysis by the image analysis device 1304. A chromosome abnormality is determined using the model of the abnormality determination process constructed by performing the model learning (step S304).

The abnormality determination result display processing unit 23 of the determination processing device 1305 displays the determination result of the abnormality determination processing unit 21 on the display device 1308, and stores it in the determination result storage device 1309 (step S305). Further, the communication processing device 1301 transmits the determination result stored in the determination result storage device 1309 to the measuring device 1002 that is the transmission source of the spectrum cube data from which the spectral feature amount image to be determined is extracted (step S306). Further, the database creation device 1306 stores the image data of the spectral feature amount image used for the abnormality determination and the feature amount obtained by the image analysis in association with the determination result in the database of the chromosome data storage device 1307 (step S307). At this time, the spectrum cube data corresponding to the spectrum feature amount image used for the abnormality determination is stored in the database corresponding to the transmission source measuring device 1002.
Thus, the chromosome abnormality determination processing ends.

As described above, the chromosome aberration determination system 1001 according to the fourth embodiment includes a process for acquiring spectrum cube data and a process for analyzing spectrum cube data in the chromosome abnormality determination device 1 according to the first embodiment. In a different device. Therefore, the same operation and effect as those of the first embodiment can be obtained. Further, as shown in FIG. 17, a measurement device 1002 for acquiring spectrum cube data and an analyzer 1003 for analyzing spectrum cube data are connected by a communication line 1004. Therefore, for example, by installing the measuring device 1002 in a plurality of hospitals and arranging the analyzing device 1003 in a different center or the like from the measuring device 1002, a plurality of hospitals and test institutions can transmit the spectrum cube data up to transmission. Then, the determination result of the chromosome abnormality can be obtained.

Here, when performing chromosome analysis, generally, for example, after collecting a sample such as blood, it is necessary to perform pretreatment, such as culturing the sample, fixing, preparing a sample, and staining, Chromosome imaging and analysis will be performed. In other words, from the time of collecting the specimen to the time of analyzing the chromosome abnormality, it is necessary to perform a preprocessing operation and an analysis operation, and an operator performing each operation is required.
Therefore, for example, by disposing the measuring device 1002 in an institution that performs pre-processing such as a hospital or an inspection institution, and performing pre-processing and imaging, the institution that performs pre-processing performs an analysis operation. The determination result of the chromosome abnormality can be obtained without arranging an operator who performs the chromosome abnormality.

Next, a fifth embodiment of the present invention will be described.
This fifth embodiment corresponds to the second embodiment, in which the image analyzer 1304 is omitted from the analyzer 1003 of the chromosome abnormality determination system 1001 in the fourth embodiment shown in FIG. The configuration of the determination processing device 1305 is different. In the analysis device 1003 according to the fifth embodiment, the image data of the spectral feature image extracted by the spectrum analysis device 1303 is directly input to the determination processing device 1305.
The same parts as those in the second and fourth embodiments are denoted by the same reference numerals, and detailed description thereof will be omitted.

FIG. 24 is a functional block diagram illustrating an example of the determination processing device 1305 according to the fifth embodiment.
The determination processing device 1305 includes an abnormality determination processing unit 21a and an abnormality determination result display processing unit 23.
In the chromosome data storage device 1307 in the fifth embodiment, the image data of the spectral feature image extracted from the spectrum cube data when the chromosome abnormality is determined is stored as feature information in the abnormal sample database 1307a. Similarly, the image data of the spectral feature image extracted from the spectrum cube data when the chromosome is determined to be normal is stored as feature information in the normal sample database 1307b. The database creation device 1306 stores the feature information (that is, the image data of the spectral feature image) in the abnormal sample database 1307a or the normal sample database 1307b based on the determination result in the abnormality determination processing unit 21a. The database creation device 1306 creates an abnormal sample database 1307a and a normal sample database 1307b for each measurement device 1002 that is the source of the spectrum cube data from which the spectral feature image has been extracted.

The abnormality determination processing unit 21a performs abnormality determination of a chromosome using a neural network model based on supervised learning. That is, the abnormality determination processing unit 21a includes an abnormality sample database 1307a in which characteristic information at the time of abnormality determination corresponding to the measuring apparatus 1002 as the transmission source of the spectrum cube data from which the spectral characteristic amount image to be determined is extracted is stored. The neural network model is constructed by individually performing machine learning on the normal sample database 1307b in which the characteristic information of is stored. Then, by inputting the image data of the spectral feature amount image of the new chromosome to be determined to the abnormality determination processing unit 21a, the image data of the spectral feature amount image of the chromosome to be determined is stored in the chromosome data storage device 1307. An abnormality is determined using a neural network model constructed based on the normal or abnormal characteristic information obtained.

FIG. 25 is a flowchart illustrating an example of a processing procedure when constructing a neural network model in the analysis device 1003.
When constructing the neural network model, the abnormality determination processing unit 21a of the determination processing device 1305 stores the abnormal sample database 1307a and the normal sample stored in the chromosome data storage device 1307 corresponding to the measurement device 1002 of the chromosome to be determined. The database 1307b is read out (Step S401), and machine learning is individually performed on the abnormal and normal feature information stored in the abnormal sample database 1307a and the normal sample database 1307b to construct a neural network model (Step S402). ). Then, the constructed neural network model is stored in a predetermined storage area.

Next, a processing procedure for performing chromosome abnormality determination in the chromosome abnormality determination system 1001 will be described with reference to FIG. FIG. 26 is a flowchart illustrating an example of a processing procedure at the time of abnormality determination in the analysis device 1003. The processing procedure at the time of measurement in the measuring apparatus 1002 is the same as the processing procedure at the time of measurement in the fourth embodiment shown in FIG.
When performing chromosome abnormality determination, first, the tester prepares a slide glass on which metaphase cells in which chromosomes can be seen are mounted, and places the slide glass on the motorized stage 1202. As a result, the chromosome to be determined on the slide on the electric stage 1202 is enlarged by the microscope 1201 and dispersed by the imaging spectroscope 1203 and stored as spectral cube data in the image memory 1204 (step S201 in FIG. 22).

Subsequently, the tester operates the communication processing device 1205 to transmit the spectrum cube data stored in the image memory 1204 to the analysis device 1003 via the communication line 1004. Thus, the spectrum cube data stored in the image memory 1204 is transmitted to the analysis device 1003 via the communication line 1004 together with the identification information for specifying the measurement device 1002 (Step S202).
On the other hand, in the analysis device 1003, when receiving the spectrum cube data from the measurement device 1002 via the communication line 1004, the communication processing device 1301 stores the received spectrum cube data in the image memory 1302 (step S311).

Subsequently, the spectrum analyzing device 1303 performs a spectrum feature amount image extraction process (step S312) on the spectrum cube data stored in the image memory 1302. Various parameters required for extracting the spectral feature image, such as an arbitrary wavelength when a single-wavelength image is extracted as a spectral feature image in the spectrum analysis by the spectrum analyzer 1303, are set in advance and stored in a predetermined storage area. The spectral feature image is extracted using the stored wavelength.
The abnormality determination processing unit 21a of the determination processing device 1305 determines a chromosome abnormality using a neural network model constructed in advance based on the image data of the spectral feature image extracted by the spectrum analysis device 1303 (step S313).

The abnormality determination result display processing unit 23 of the determination processing device 1305 displays the determination result of the abnormality determination processing unit 21a on the display device 1308. Also, it is stored in the determination result storage device 1309 (step S314). Further, the communication processing device 1301 transmits the determination result stored in the determination result storage device 1309 to the measuring device 1002 that is the transmission source of the spectrum cube data from which the spectral feature image to be determined is extracted (step S315). Further, the database creation device 1306 associates the image data of the spectral feature amount image used for the abnormality determination with the determination result of the abnormality determination, and stores the spectrum cube data in the chromosome data storage device 1307 from which the spectral feature amount image has been extracted. The data is stored in the database corresponding to the measuring device 1002 of the transmission source (step S316).
Thus, the chromosome abnormality determination processing ends.
Therefore, the chromosome abnormality determination system 1001 according to the fifth embodiment can obtain the same operation and effects as those of the second embodiment.

Next, a sixth embodiment of the present invention will be described.
The sixth embodiment corresponds to the third embodiment, in which the image analyzer 1304 is omitted from the analyzer 1003 of the chromosome abnormality determination system 1001 in the fourth embodiment shown in FIG. The configuration of the determination processing device 1305 is different. In the analysis device 1003 according to the sixth embodiment, the image data of the spectral feature image extracted by the spectrum analysis of the spectrum analysis device 1303 is directly input to the determination processing device 1305.
The same parts as those in the third embodiment and the fourth embodiment are denoted by the same reference numerals, and detailed description thereof will be omitted.
FIG. 27 is a functional block diagram illustrating an example of the determination processing device 1305 according to the sixth embodiment.
The determination processing device 1305 includes an abnormality determination processing unit 21c and an abnormality determination result display processing unit 23.

In the chromosome data storage device 1307 in the sixth embodiment, the spectrum feature image and the determination result of normal or abnormal are associated with each other, and the spectrum cube data from which the spectral cube data is extracted. Then, it is stored in the chromosome data storage device 1307 via the database creation device 1306.
The abnormality determination processing unit 21c performs chromosome abnormality determination using a neural network model based on unsupervised learning. That is, the abnormality determination processing unit 21c constructs a neural network model by performing machine learning using a database including image data of the spectrum feature amount image extracted by the spectrum analysis device 1303. Then, the abnormality determination processing unit 21c inputs the image data of the spectral feature amount image extracted by the spectrum analysis to the new chromosome to be determined to the abnormality determination processing unit 21c. An abnormality determination is performed on the image data of the image using the constructed neural network model.

Note that, in the analysis device 1003 in the sixth embodiment, the processing procedure at the time of model construction is basically the same as the processing procedure at the time of model construction in the fifth embodiment shown in FIG. 25, but in the sixth embodiment, When constructing a neural network model (step S402 in FIG. 25), the chromosome data storage device 1307 stores the spectral feature amount image to be determined in the database corresponding to the measuring device 1002 from which the spectral cube data is extracted. Machine learning is performed using the image data of the spectral feature image at the time of abnormality and at the time of normal, and a neural network model is constructed.

In the chromosome abnormality determination system 1001 according to the sixth embodiment, the processing procedure of the measurement device 1002 and the analysis device 1003 when performing the chromosome abnormality determination is basically the same as that of the fifth embodiment shown in FIGS. , The analysis apparatus 1003 extracts a predetermined spectral feature amount image when performing spectrum analysis (step S312), and constructs a neural network constructed for each extracted spectral feature amount image. An abnormality is determined using the model.
As described above, since the abnormality determination is performed using the neural network model on the image data of the spectral feature image obtained by the spectrum analysis, the abnormality determination is performed using the feature information that cannot be obtained only with the two-dimensional image. Can be. Therefore, also in this case, the chromosome abnormality can be automatically determined for more chromosome images, and the usability of the chromosome abnormality determination system 1001 can be improved.

In the fourth to sixth embodiments as well, image data is obtained by extracting only specific wavelengths such as a predetermined single wavelength, a predetermined two wavelengths, a predetermined three wavelengths, etc. from the reflected light or transmitted light of the chromosome. Alternatively, only the minimum number of wavelengths necessary for acquiring the spectral feature image may be extracted to acquire the image data.

FIG. 28 shows an example of a schematic configuration of a chromosome abnormality determination system 1001a that extracts only wavelengths necessary for obtaining a spectral feature image using a filter and obtains image data. The chromosome abnormality determination system 1001a is obtained by changing the measurement device 1002 to a measurement device 1002a in the chromosome abnormality determination system 1001 shown in FIG. The measuring apparatus 1002a in the measuring apparatus 1002 shown in FIG. 18 includes, instead of the imaging spectroscope 1203, a monochrome camera 1211 and a filter 1212 provided in the monochrome camera 1211 and having a characteristic of transmitting only a specific wavelength. With such a configuration, it is possible to acquire image data in which only a specific wavelength is extracted from the reflected light or transmitted light of the chromosome.

In this case as well, when acquiring image data of an arbitrary wavelength and performing chromosome abnormality determination using a neural network model, construct a neural network model corresponding to the extracted wavelength and perform abnormality determination. What should I do?

特性 In the first to sixth embodiments, the characteristics of the obtained spectral cube data may vary depending on the type of staining solution, staining method, and the like. Therefore, when extracting a single-wavelength image of an arbitrary wavelength as a spectral feature image, for example, for each type of staining solution or staining method, the optimal wavelength as a single-wavelength image, the type of staining solution, staining method, etc. And stored as wavelength information in association with the wavelength information. Then, the tester inputs and sets the type and staining method of the staining solution when the new chromosome to be determined is stained, so that the staining solution input and set from the wavelength information stored in advance is set. A wavelength corresponding to a type, a staining method, or the like may be searched, and a single wavelength image or the like may be created using the wavelength. In this way, even when the characteristics and the like of the spectrum cube data change due to the type of the staining solution, the staining method, and the like, the image analysis by the image analysis device 7 or 1304 that performs processing based on the spectral feature amount image can be performed. It is possible to obtain a single-wavelength image suitable for abnormality determination in the determination processing device 10 or 1305, and to perform image analysis and abnormality determination based on the single-wavelength image, thereby improving analysis accuracy and determination accuracy. .

Further, in each of the above embodiments, the chromosome data storage device 12 or 1307 is provided, and the content of the determination processing in the determination processing device 10 or 1305 is learned using the database stored in the chromosome data storage device 12 or 1307. But it is not limited to this. For example, in a device different from the chromosome abnormality determination device 1 or the chromosome abnormality determination system 1001, a neural network model is constructed, and the constructed neural network model is mounted on the chromosome abnormality determination device 1 or the analysis device 1003, and is mounted. The abnormality determination may be performed using a neural network model.
Further, in each of the above embodiments, a case has been described in which an abnormality determination process is performed by constructing a model of an abnormality determination process by model learning or by constructing a neural network model. For example, by directly comparing the characteristic information at the time of normal or abnormal included in the database stored in the chromosome data storage device 12 or 1307 with the characteristic information extracted from the chromosome to be determined. Alternatively, an abnormality determination may be performed.

Note that the scope of the present invention is not limited to the illustrated and described exemplary embodiments, but includes all the embodiments having the same effects as those aimed at by the present invention. Furthermore, the scope of the present invention may be defined by any desired combination of the particular features of each disclosed feature.

DESCRIPTION OF SYMBOLS 1, 1a Chromosome abnormality determination apparatus 2 Microscope 4 Imaging spectroscope 5 Image memory 6 Spectrum analysis apparatus 7 Image analysis apparatus 10 Judgment processing apparatus 11 Database creation apparatus 12 Chromosome data storage apparatus 13 Display apparatus 14 Monochrome camera 15 Filter 1001, 1001a Chromosome abnormality Judgment system 1002 Measurement device 1003 Analysis device 1004 Communication line 1201 Microscope 1203 Imaging spectroscope 1204 Image memory 1205 Communication processing device 1211 Monochrome camera 1212 Filter 1301 Communication processing device 1302 Image memory 1303 Spectrum analysis device 1304 Image analysis device 1305 Judgment processing device 1306 Database Creation device 1307 Chromosome data storage device 1308 Display device

Claims (18)

An abnormality determination processing unit that performs an abnormality determination process constructed using feature information representing the characteristics of the chromosome included in the spectrum cube data obtained from at least one of the abnormal chromosome and the normal chromosome,
A data acquisition unit for acquiring the spectrum cube data of the chromosome to be determined,
Has,
The spectral cube data is a spectral spectrum at each point on the chromosome of light transmitted or reflected through the stained chromosome,
The abnormality determination processing unit,
Inputting the characteristic information included in the spectrum cube data of the chromosome to be determined acquired by the data acquisition unit, and performing the abnormality determination process on the chromosome to be determined using the characteristic information. Chromosome abnormality determination device. The abnormality determination processing unit,
In the abnormality determination process, the characteristic information included in the spectral cube data of the chromosome to be determined is compared with the characteristic information included in the spectral cube data of at least one of an abnormal chromosome and a normal chromosome. The chromosome abnormality judging device according to claim 1, wherein the chromosome abnormality judging device executes a process for performing the chromosome abnormality. The abnormality determination processing unit,
As the abnormality determination processing, based on the characteristic information included in the spectrum cube data of at least one of the abnormal chromosome and the normal chromosome and information indicating whether the chromosome having the characteristic information is normal or not. The chromosome abnormality judging device according to claim 1, wherein a process constructed by performing model learning using the determined abnormality judging parameter is executed. The abnormality determination processing unit,
As the abnormality determination process, a feature information group including a plurality of the feature information included in the spectrum cube data of the abnormal chromosome, and a feature information group including a plurality of the feature information included in the spectrum cube data of the normal chromosome The chromosome abnormality judging device according to claim 1, wherein a process using a neural network model constructed by individually performing machine learning on the chromosome is executed. The abnormality determination processing unit,
By performing machine learning on the characteristic information included in the spectral cube data of at least one of an abnormal chromosome and a normal chromosome and information indicating whether the chromosome having the characteristic information is normal or not. The chromosome abnormality judging device according to claim 1, wherein a process using the constructed neural network model is executed. Of the spectral cube data, the characteristic information using data in a wavelength range including a wavelength selected in advance according to the characteristics of the density of the stained chromosome and the amount of light transmitted or reflected through the stained chromosome The chromosome abnormality judging device according to any one of claims 1 to 5, further comprising a feature information extracting unit that extracts a chromosome. A database in which the characteristic information included in the spectral cube data of at least one of an abnormal chromosome and a normal chromosome and information indicating whether the chromosome having the characteristic information is abnormal are stored,
7. The system according to claim 1, further comprising a database creation unit that stores the feature information input to the abnormality determination processing unit and a determination result of the feature information by the abnormality determination processing unit in the database. The chromosome abnormality judging device according to any one of the above. 8. The chromosome abnormality judging device according to claim 7, further comprising: an abnormality judging process constructing unit that constructs the abnormality judging process based on the feature information stored in the database and the judgment result. A plurality of measurement devices that obtain spectrum cube data of the chromosome to be determined, and an analysis device that performs an abnormality determination of the chromosome to be determined based on the spectral cube data, in a chromosome abnormality determination system connected via a network. So,
The measurement device, a data acquisition unit that acquires a spectral spectrum at each point on the chromosome of the determination target of light transmitted or reflected by the stained chromosome of the determination target as the spectral cube data,
A first transmitting the spectrum cube data acquired by the data acquisition unit to the analyzer via the network, and receiving an abnormality determination result in the analyzer for the transmitted spectrum cube data via the network; A communication processing unit;
The analysis device receives the spectrum cube data from the measurement device via the network, and transmits an abnormality determination result in the analysis device to the received spectrum cube data to the measurement device via the network. A second communication processing unit,
Abnormality chromosome and abnormality determination processing constructed using characteristic information representing the characteristics of the chromosome included in the spectral cube data obtained from at least one of the normal chromosome, the spectrum of the chromosome of the determination target An abnormality determination processing unit that executes using the feature information included in the cube data,
A chromosome abnormality determination system, comprising: A measurement device that acquires spectrum cube data of a chromosome to be determined and transmits the spectrum cube data to an analyzer that performs an abnormality determination of a chromosome connected via a network,
A data acquisition unit that acquires a spectral spectrum at each point on the chromosome of the determination target of light transmitted or reflected through the stained chromosome of the determination target as the spectrum cube data,
A first transmitting the spectrum cube data acquired by the data acquisition unit to the analyzer via the network, and receiving an abnormality determination result in the analyzer for the transmitted spectrum cube data via the network; And a communication processing unit. An analysis device that performs an abnormality determination of the chromosome to be determined based on the spectral cube data of the chromosome to be determined,
A second communication processing unit that receives, via a network, the spectrum cube data of the chromosome to be determined, and transmits an abnormality determination result in the analyzer for the received spectrum cube data through the network.
Abnormality chromosome and abnormality determination processing constructed using characteristic information representing the characteristics of the chromosome included in the spectral cube data obtained from at least one of the normal chromosome, the spectrum of the chromosome of the determination target An abnormality determination processing unit that executes using the feature information included in the cube data,
An analysis device comprising: The abnormality determination processing unit,
In the abnormality determination process, the characteristic information included in the spectral cube data of the chromosome to be determined is compared with the characteristic information included in the spectral cube data of at least one of an abnormal chromosome and a normal chromosome. The analysis device according to claim 11, wherein the analysis device performs a process of performing the analysis. The abnormality determination processing unit,
As the abnormality determination processing, based on the characteristic information included in the spectrum cube data of at least one of the abnormal chromosome and the normal chromosome and information indicating whether the chromosome having the characteristic information is normal or not. 12. The analysis apparatus according to claim 11, wherein a process constructed by performing model learning using the determined abnormality determination parameter is executed. The abnormality determination processing unit,
As the abnormality determination process, a feature information group including a plurality of the feature information included in the spectrum cube data of the abnormal chromosome, and a feature information group including a plurality of the feature information included in the spectrum cube data of the normal chromosome The analysis apparatus according to claim 11, wherein the apparatus performs processing using a neural network model constructed by individually performing machine learning on the analysis. The abnormality determination processing unit,
By performing machine learning on the characteristic information included in the spectral cube data of at least one of an abnormal chromosome and a normal chromosome and information indicating whether the chromosome having the characteristic information is normal or not. 12. The analysis device according to claim 11, wherein a process using the constructed neural network model is executed. The spectral cube data is a spectral spectrum at each point on the chromosome of the determination target of the light transmitted or reflected through the stained chromosome of the determination target,
Of the spectral cube data, the characteristic information using data in a wavelength range including a wavelength selected in advance according to the characteristics of the density of the stained chromosome and the amount of light transmitted or reflected through the stained chromosome The analysis device according to any one of claims 11 to 15, further comprising: a feature information extraction unit configured to extract the information. A database in which the characteristic information included in the spectral cube data of at least one of an abnormal chromosome and a normal chromosome and information indicating whether the chromosome having the characteristic information is abnormal are stored,
A database creation unit that stores the feature information input to the abnormality determination processing unit and the determination result of the abnormality determination processing unit for the feature information in the database,
The analyzer according to any one of claims 11 to 16, further comprising: 18. The analysis apparatus according to claim 17, further comprising: an abnormality determination processing constructing unit configured to construct the abnormality determination processing based on the characteristic information stored in the database and the determination result.

Images