JP4922109B2 - Cell analysis method, apparatus and program - Google Patents

Description

  The present invention relates to a cell analysis method, a cell analysis program, and a cell analysis device used for pathological diagnosis of cancer and the like using living cells and drug screening.

Evaluating the effects of various drugs and searching for particularly effective drugs are generally referred to as drug screening. In particular, in drug screening using living cells, after the drug is administered, an appropriate stimulus is applied to the cultured cells, and the difference in the cells from the cells that have not been stimulated is quantified to measure the drug effect.
For example, in drug screening using fluorescence, a fluorescent protein gene is administered to a cell, and the amount of fluorescence emitted from the cell is detected after a lapse of time, thereby quantifying the chemical reaction of the cell and measuring the drug effect. To do. In this case, the greater the expression level of a specific gene in the cell, the greater the amount of fluorescence.

In general, the drug reaction as described above is observed by an examiner by visually observing cell changes before and after drug administration with a microscope or the like. However, the work by the inspector not only consumes a great deal of labor, but also has a problem of lack of quantitativeness. Therefore, in recent years, a cell analysis apparatus that automatically performs such cell analysis has been proposed (for example, see Patent Document 1).
In Patent Document 1, a container containing cells is placed on a stage, a sample is imaged by a camera to obtain a cell image to be analyzed, and the chemical reaction of the cell is quantified by analyzing the cell image. A cell analysis device is disclosed.
JP 2001-307066 A

  However, when the drug reaction in the cell is automatically quantified by image analysis, there is a problem that the analysis accuracy decreases due to noise or the like included in the image. For example, cell brightness (fluorescence amount) in cell images has various effects such as the amount of illumination light applied to the sample during image acquisition, autofluorescence of the container containing the sample, and changes in the thickness of the container. Since these appear in the cell image as noise, the analysis accuracy decreases.

  This invention is made | formed in view of the situation mentioned above, Comprising: It aims at providing the cell analysis method and apparatus which can improve the evaluation precision of the drug reaction of a cell, and a program.

In order to solve the above problems, the present invention employs the following means.
The present invention provides an imaging step of imaging each cell group cultured under different conditions, a luminance measurement step of measuring the luminance of each cell from the cell image acquired in the imaging step, and luminance data for each of the conditions A distribution characteristic creating step for creating a distribution characteristic, a normalization process for normalizing the distribution characteristic created for each of the conditions, and averaging the distribution characteristics after standardization in the normalization process The average distribution characteristic is obtained by the above, a deviation from the average distribution characteristic is obtained for each distribution characteristic, and luminance data showing a tendency that the deviation is different from other distribution characteristics is deleted or corrected so as to be close to the average distribution characteristic. A cell analysis method including a correction step is provided.

According to the present invention, a plurality of cell images are acquired by imaging each cell group cultured under different conditions. Subsequently, the luminance of each cell is measured in each acquired cell image, a distribution characteristic of luminance data is created for each culture condition, and the distribution characteristic is normalized. Thus, by normalizing distribution characteristics, errors due to differences in culture conditions can be reduced, and highly reliable distribution characteristics can be obtained. Therefore, by comparing these distribution characteristics between the culture conditions, it becomes possible to more accurately evaluate the drug reaction of the cells.
According to the present invention, it is possible to delete luminance data with low reliability. As a result, it is possible to acquire distribution characteristics with higher reliability.

  In the cell analysis method, the normalization step includes a reference setting step of setting a reference distribution characteristic serving as a reference based on a plurality of the distribution characteristics, and the distribution so that each of the distribution characteristics matches the reference distribution characteristic. A distribution characteristic conversion step of normalizing the distribution characteristic by linearly converting each luminance data in the characteristic.

  Since the reference distribution characteristics are set and each luminance data is linearly converted so that each distribution characteristic matches the reference distribution characteristics, normalization can be easily performed.

  In the cell analysis method, the distribution characteristic conversion step may linearly convert each luminance data in the distribution characteristic so that each distribution characteristic peak and the reference distribution characteristic peak coincide with each other.

For example, when the reaction evaluation based on the drug concentration is performed, not all cells respond to the drug, and most of them are unresponsive in many cases. In addition, since the non-reactive cells exhibit substantially the same luminance, it can be determined that the peak in the distribution characteristic is the data indicated by the non-reactive cells. Therefore, by combining the peaks in each distribution characteristic, it is possible to easily reduce the error between the distribution characteristics.
Furthermore, when performing the reaction evaluation based on the drug concentration, it is preferable to compare luminance data indicating a high value between the drug concentrations. In such a case, according to the present invention, the luminance data in the distribution characteristic is linearly converted to normalize the distribution characteristic so that the peak of the distribution characteristic coincides with the peak of the reference distribution characteristic. Therefore, it is possible to accurately grasp the change in the luminance data in the luminance range that is important when performing the measurement, and it is possible to obtain a highly reliable evaluation result of the drug reaction.

Further, the present invention is a cell analysis apparatus that performs cell analysis using a cell image acquired by imaging a plurality of cells cultured under different conditions, and measures the luminance of each cell from the cell image Brightness measurement means for generating, distribution characteristic creation means for creating distribution characteristics of brightness data for each condition, normalization means for normalizing the distribution characteristics created for each condition, and normalization by the normalization means The average distribution characteristic is obtained by averaging the distribution characteristics after being obtained, and a deviation from the average distribution characteristic is obtained for each distribution characteristic, and the luminance data indicates a tendency that the deviation is different from other distribution characteristics. A cell analyzing apparatus is provided that includes correction means that deletes or corrects to approximate the average distribution characteristic .

A cell analysis program for performing cell analysis using a cell image acquired by imaging a plurality of cells cultured under different conditions, the luminance measuring the luminance of each cell from the cell image A measurement step, a distribution characteristic creation step for creating a distribution characteristic of luminance data for each condition, a normalization step for normalizing the distribution characteristic created for each condition, and standardized in the normalization step The average distribution characteristic is obtained by averaging the distribution characteristics later, and the deviation from this average distribution characteristic is obtained for each distribution characteristic, and the luminance data showing the tendency that the deviation is different from the other distribution characteristics is deleted. Alternatively, the present invention provides a cell analysis program that causes a computer to execute a correction step for correcting the average distribution characteristic so as to approximate it.

  According to this invention, there exists an effect that the evaluation precision of the drug reaction of a cell can be improved.

  Hereinafter, an embodiment of a cell analysis method, a cell analysis program, and a cell analysis apparatus of the present invention will be described with reference to the drawings.

FIG. 1 is a diagram showing a configuration of a cell analysis apparatus according to an embodiment of the present invention.
As shown in FIG. 1, the cell analyzer according to the present embodiment includes a stage 101 on which a sample plate 1 is mounted, an excitation light source 102, a power supply unit 103 connected to the excitation light source 102, and an excitation light source 102. A shutter unit 104 that turns on and off the light, a lens 105 that converts the light into parallel light, a filter unit 106 that selects the wavelength of the excitation light to be incident on the sample, and reflects the excitation light to be incident on the sample to emit fluorescence. A mirror set 107 that transmits light, an objective lens 108 that condenses excitation light on a specimen, an imaging lens 109 that condenses the fluorescence that is collected by the objective lens 108 and passes through the mirror set 107, and the imaged fluorescence. Controls the CCD camera 110 for imaging, the shutter unit 104, the filter unit 106, and the mirror set 107. And Nibasaru control box 111, and stage controller 115 controls the stage 101, along with processing the fluorescence captured by the CCD camera 110, and a computer 112 for controlling the universal control box 111 and the stage controller 115.

  As the sample plate 1, as shown in FIG. 2, for example, a plate in which 15 wells 1a having a diameter of 20 mm and a depth of 15 mm are arranged in 3 rows and 5 columns is used. Cells stained with a fluorescent dye to be examined are seeded in each well 1a. The fluorescent dye used in the present embodiment is a dye that can emit fluorescence while keeping cells alive, such as GFP, and is stained in the cytoplasm. In addition, a barcode 1 b is attached to the sample plate 1. The bar code 1b is read by a bar code reader (not shown), so that the bar code 1b is collated with the database, and the consistency between the sample plate 1 and the data in the database is maintained.

The stage 101 is mounted with the sample plate 1 and can be electrically driven and driven two-dimensionally in the horizontal direction to position the sample plate 1 at an arbitrary position.
As the excitation light source 102, a mercury light source or the like is used. The shutter unit 104 is disposed on the optical path of the excitation light emitted from the excitation light source 102 and incorporates a shutter plate 104a that can be opened and closed.

  The filter unit 106 includes two excitation filters 106a, and can irradiate the sample with two types of excitation light having different wavelengths. The mirror set 107 includes a dichroic mirror 107A that reflects excitation light and transmits fluorescence, and an excitation filter 107D and an absorption filter 107C attached thereto.

The mirror set 107 is stored in a mirror holder (not shown) for a necessary excitation wavelength, and an optical path can be arbitrarily switched by an electric unit (not shown). On the reflected light path of the mirror set 107, the objective lens 108 and the sample plate 1 are arranged. An imaging lens 109 and a CCD camera 110 are disposed on the transmission light path of the mirror set 107. The CCD camera 110 is, for example, a 1 million pixel camera with 12 bits and 1000 × 1000 pixels.
A keyboard 113, a monitor 114, and a mouse 116 are connected to the computer 112.

A cell analysis method using the cell analysis apparatus according to this embodiment configured as described above will be described with reference to the drawings.
First, in the first row of the well 1a of the sample plate 1 (see FIG. 2), for example, a compound A that is a drug candidate for implanting cells is administered at a concentration a. Similarly, in the second row, cells are implanted and Compound A as a drug candidate is administered at a concentration b. Further, on the third line, cells are implanted and compound A as a drug candidate is administered at a concentration c. Thus, in this embodiment, a plurality of cells are cultured under three different conditions. Each well 1a is given a well number. Thereby, the culture condition of the cell can be discriminated by the well number.

  Subsequently, the inspector sets the sample plate 1 on the stage 101, activates a control program (not shown) on the computer 112, reads the barcode (ID: 10001) of the sample plate 1, and starts observation and photographing. . Thereby, a cell image is acquired for every well 1a, and a well number and a cell image are matched and preserve | saved at the hard disk in the computer 112 (imaging process).

  When image acquisition is completed for all wells, the examiner activates a cell analysis program on the computer 112 and instructs the start of analysis. Thereby, the cell analysis program is executed.

Hereinafter, a cell analysis method realized by executing a cell analysis program by a computer will be specifically described with reference to FIG. FIG. 3 is a flowchart showing the procedure of the cell analysis method according to the present embodiment.
First, in step SA1, the cell image stored in the hard disk of the computer 112 is loaded into a memory (not shown).
In step SA2, a plurality of cell images are classified for each culture condition based on the well number. In step SA3, the luminance of each cell is measured from the cell image for each condition, and a luminance data set is created for each condition (luminance measurement step). Here, in general, the luminance varies depending on the location even within one cell. For this reason, in this embodiment, the average luminance of the whole cell or the maximum luminance is adopted as the luminance data.

In subsequent step SA4, a distribution characteristic is created from the luminance data set under each condition (distribution characteristic creation step). In the present embodiment, since the cells are cultured under the three conditions of concentrations a, b, and c, three distribution characteristics are created.
FIG. 4 shows an example of luminance distribution characteristics. As shown in FIG. 4, the luminance distribution characteristic is obtained as a graph in which the horizontal axis represents luminance and the vertical axis represents frequency. The frequency may be, for example, the number of cells or a ratio obtained by dividing the number by a predetermined value. As shown in FIG. 4, for example, in the case of a drug response, the distribution characteristic has a peak, and the frequency gradually decreases as the luminance increases.

Next, in step SA5, the above distribution characteristics are normalized (standardization process). FIG. 5 is a flowchart showing the procedure of the normalization process performed in step SA5.
First, in step SB1, a reference distribution characteristic is set (reference setting step). The reference distribution characteristic is used to determine a distribution characteristic that serves as a reference for normalization. For example, one of the three distribution characteristics created in step SA4 is set as the reference distribution characteristic.

  In subsequent step SB2, one of the distribution characteristics other than the reference distribution characteristic is selected as a normalization target, and in subsequent step SB3, one or more of the reference distribution characteristic and the selected distribution characteristic (hereinafter referred to as “selection distribution characteristic”) are selected. Set a representative point. As representative points, for example, as shown in FIG. 6, the peak, the minimum luminance value, the upper 10% boundary luminance, and the like can be cited as examples.

Next, in step SB4, as shown in FIG. 6, the luminance data set related to the selection distribution characteristic is converted into the following conversion formula (1) so that each representative point of the selection distribution characteristic matches those of the reference distribution characteristic. Is used for linear conversion (distribution characteristic conversion step).
For example, when the luminance data is x, the luminance data x is linearly converted using the following equation (1).
X ′ = ax + b (1)
In the equation (1), X ′ is luminance data after linear conversion, a and b are variables, and x is measured luminance data.

By changing the variable b, the selection distribution characteristic can be translated along the X axis. This data operation corresponds to, for example, an operation for reducing noise caused by autofluorescence or the like of the sample plate 1. That is, since the plastic sample plate 1 emits fluorescence, the measured luminance of the cell is a value affected by the autofluorescence. For this reason, by adjusting the variable b, it is possible to correct data variation due to autofluorescence.
Further, by changing the variable a, it is possible to change the maximum frequency and the luminance data taking the maximum frequency. This data operation corresponds to an operation for reducing a measurement error of luminance data due to a change in intensity of illumination light.

  In step SB4, the selection distribution characteristics are linearly converted by arbitrarily changing the values of a and b within a predetermined range, and an error at each representative point between the converted selection distribution characteristics and the reference distribution characteristics is calculated. The values of a and b when the accumulated error is minimized are specified. Thus, by performing data normalization, it is possible to reduce noise components generated by various factors as described above.

Next, in step SB5, it is determined whether or not normalization has been performed for all distribution characteristics. As a result, when normalization has not been performed for all distribution characteristics, the process returns to step SB2, and a distribution characteristic that has not been normalized is selected, and the processing after step SB3 is performed for the selected distribution characteristic. Similarly, the distribution characteristics are normalized.
On the other hand, in step SB5, when normalization has been performed for all distribution characteristics, the normalization process ends and the flow returns to the flow of FIG.

  In step SA6 of FIG. 4, each normalized distribution characteristic is compared, and luminance data that shows a different tendency compared to other distribution characteristics is deleted or corrected (correction step).

  In the correction step, an average distribution characteristic is obtained by averaging the normalized distribution characteristics, and a deviation from the average distribution characteristic is obtained for each distribution characteristic. If the deviation is equal to or greater than a predetermined threshold value, it is possible to further reduce noise by deleting the luminance data of the portion or correcting the luminance data to approximate the average characteristic.

  FIG. 7 shows average distribution characteristics and normalized distribution characteristics. As shown in FIG. 7, noise removal is performed by adopting, adjusting, or deleting luminance data of other characteristics for the region P in which the deviation is greater than a predetermined threshold with respect to the average distribution characteristics. We are going to plan.

In this way, by obtaining the average distribution characteristic from the normalized distribution characteristic and deleting or correcting the luminance data showing a unique tendency with respect to the average distribution characteristic, a highly reliable distribution characteristic can be obtained. Is possible.
In addition, when cells are cultured using different types of drugs, it is difficult to determine whether the drug has a specific tendency or a noise has a specific tendency. In that case, it is possible to leave the luminance data showing a unique tendency with respect to the average distribution characteristic as it is without deleting or correcting it, and to analyze this unique tendency in a later evaluation.

  When a highly reliable distribution characteristic is obtained in this way, the chemical reaction caused by the drug is quantified and evaluated by comparing the distribution characteristic for each condition finally obtained in step SA7 in FIG. Perform the evaluation process.

  As described above, according to the cell analysis method, apparatus, and program according to the present embodiment, the luminance distribution characteristic created for each culture condition is normalized, so that the error in the luminance data due to the difference in the culture condition is reduced. Therefore, it is possible to obtain luminance data and distribution characteristics with high reliability. Therefore, by comparing these distribution characteristics between the culture conditions, it becomes possible to more accurately evaluate the drug reaction of the cells.

  For example, when evaluating the difference in response due to the drug concentration, it is preferable to compare luminance data indicating high values between the drug concentrations. At this time, in order to compare the number of cells reacting with the drug, there is an evaluation method in which a threshold value of brightness is set, and the number of cells that have a brightness higher than the threshold value is counted as a reaction. Even when such an evaluation method is adopted, the cell analysis method according to the present embodiment linearly converts the luminance data in the distribution characteristic so that the peak of each distribution characteristic matches the peak of the reference distribution characteristic. Therefore, it is possible to accurately compare the tendency of the distribution characteristics in the luminance range, which is important when evaluating the drug reaction. Thereby, an appropriate luminance threshold value can be set in the drug evaluation according to these comparison results, and the drug evaluation can be performed with high accuracy.

  In the cell analysis method according to the above-described embodiment, in the normalization process performed in step SA5 of FIG. 4, the linear conversion of the distribution characteristics is performed so that the representative points of the reference distribution characteristics coincide with the representative points of the selected distribution characteristics. However, this method is an example, and for example, the following method 1 or method 2 may be adopted.

[Method 1]
First, as shown in FIG. 8, when the minimum luminance of the reference distribution characteristic is min_i, the maximum luminance is max_i, the minimum luminance of the distribution characteristic is min_j, and the maximum luminance is max_j, the following equations (2) and (3) Are used to calculate the differences Li and Lj between the maximum luminance and the minimum luminance in the respective distribution characteristics.
Li = max_i−min_i (2)
Lj = max_j−min_j (3)
Subsequently, the above-described a is obtained by dividing the difference Li relating to the reference distribution characteristic by the difference Lj relating to the distribution characteristic using the following equation (4).
a = Li / Lj = (max_i-min_i) / (max_j-min_j) (4)
The above b is calculated using the following equation (5).
b = min_j−min_i (5)

  As described above, a and b may be acquired using the minimum luminance and the maximum luminance of each distribution characteristic, and the selection distribution characteristic may be normalized using the acquired values of a and b. The reason why a and b can be obtained by such a method is related to the fact that each distribution characteristic shows a similar characteristic. In other words, each distribution characteristic has a peak in a low-luminance region, and the frequency (number) gradually decreases as the luminance increases. Since the shapes of the distribution characteristics are substantially the same as described above, the values of a and b can be easily obtained by using the above-described method, and the selection distribution characteristics can be easily standardized. .

[Method 2]
In Method 2, as in Method 1 described above, when the minimum luminance of the reference distribution characteristic is min_i, the maximum luminance is max_i, the minimum luminance of the distribution characteristic is min_j, and the maximum luminance is max_j, the change numerical value ranges of a and b are set. The candidate values of a and b are respectively set by determining in advance as follows and dividing the setting range by a preset number.
| A | <(max_i-min_i) / (max_j-min_i)
| B | <min_j-min_i

  Then, the distribution characteristics are converted while changing the set combination of the candidate value a and the candidate value b as needed, and the combination of the values of a and b when the cumulative error d between the reference distribution characteristic and the distribution characteristic in each combination is minimized. Ask for. The accumulated error d is expressed by the following equation (6).

d = sam_k | Pi (tk) −Pj (tk) | (6)
However, 0 ≦ k ≦ n

Here, as shown in FIG. 9, Pi (tk) represents the frequency of the reference distribution characteristic at the luminance tk, and Pj (tk) represents the frequency of the selected distribution characteristic at the luminance tk. As tk is set in finer increments, the error between the selection distribution characteristic and the reference distribution characteristic can be determined more finely.
By performing the above process for each distribution characteristic, it is possible to obtain the values of a and b for each distribution characteristic when each distribution characteristic is most approximated to the reference distribution characteristic.

As mentioned above, although embodiment of this invention was explained in full detail with reference to drawings, the specific structure is not restricted to this embodiment, The design change etc. of the range which does not deviate from the summary of this invention are included.
For example, in the above-described embodiment, the distribution characteristic is created for each condition. However, the present invention is not limited to this. It is good also as performing. In this way, by performing normalization for each well, it is possible to further reduce noise due to well differences and noise in illumination during photographing.
In the above embodiment, one of the three distribution characteristics is set as the reference distribution characteristic. However, the present invention is not limited to this example. For example, a new reference distribution characteristic may be provided from the three distribution characteristics. Alternatively, any reference distribution characteristic that is set in advance may be used.

It is a whole lineblock diagram showing the cell analysis device concerning one embodiment of the present invention. It is a perspective view which shows an example of the sample plate used for the cell analyzer of FIG. It is a flowchart which shows the procedure of the cell analysis method which concerns on one Embodiment of this invention. An example of luminance distribution characteristics is shown. It is the flowchart shown about the procedure of the normalization process. It is a figure for demonstrating an example of a representative point and the normalization method of distribution characteristics. It is a figure which shows an average distribution characteristic and the distribution characteristic after normalization. It is explanatory drawing used for description about the method 1 of a normalization process. It is explanatory drawing used for description about the method 2 of a normalization process.

Explanation of symbols

1 Sample plate 102 Excitation light source 110 CCD camera 111 Universal control box 112 Computer

Claims (5)

An imaging step of imaging each cell group cultured under different conditions;
A luminance measurement step of measuring the luminance of each cell from the cell image acquired in the imaging step;
A distribution characteristic creation step of creating a distribution characteristic of luminance data for each of the conditions;
A normalization step of normalizing the distribution characteristics created for each condition;
An average distribution characteristic is obtained by averaging the distribution characteristics after normalization in the normalization step, and a deviation from the average distribution characteristic is obtained for each distribution characteristic, and the deviation is compared with other distribution characteristics. And a correction step of correcting the luminance data indicating different tendencies so as to be deleted or approximated to the average distribution characteristic . The standardization process includes
A reference setting step for setting a reference distribution characteristic as a reference based on the plurality of distribution characteristics;
The cell analysis according to claim 1, further comprising a distribution characteristic conversion step of normalizing the distribution characteristic by linearly converting each luminance data in the distribution characteristic so that each distribution characteristic matches the reference distribution characteristic. Method.   The cell analysis method according to claim 2, wherein in the distribution characteristic conversion step, luminance data in the distribution characteristic is linearly converted so that a peak of each distribution characteristic coincides with a peak of the reference distribution characteristic. A cell analysis device that performs cell analysis using a cell image acquired by imaging a plurality of cells cultured under different conditions,
A luminance measuring means for measuring the luminance of each cell from the cell image;
A distribution characteristic creating means for creating a distribution characteristic of luminance data for each condition;
Normalization means for normalizing the distribution characteristics created for each condition;
An average distribution characteristic is obtained by averaging the distribution characteristics after being normalized by the normalization means, and a deviation from the average distribution characteristic is obtained for each distribution characteristic, and the deviation is compared with other distribution characteristics. And a correction unit that corrects the luminance data indicating different tendencies to be deleted or approximated to the average distribution characteristic . A cell analysis program for performing cell analysis using a cell image obtained by imaging a plurality of cells cultured under different conditions,
A luminance measurement step of measuring the luminance of each cell from the cell image;
A distribution characteristic creation step of creating a distribution characteristic of luminance data for each of the conditions;
A normalization step for normalizing the distribution characteristics created for each of the conditions;
An average distribution characteristic is obtained by averaging the distribution characteristics after normalization in the normalization step, and a deviation from the average distribution characteristic is obtained for each distribution characteristic, and the deviation is compared with other distribution characteristics. A cell analysis program for causing a computer to execute correction steps for deleting brightness data showing different tendencies or correcting the brightness data to approach the average distribution characteristic .

Images