JP2006333710A - Automatic system for judging quality of cell - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an automatic system for judging the quality of a cell designed to achieve the automation of judgment about the quality of the cell. <P>SOLUTION: An automatic apparatus 1 for culturing having the automatic system for judging the quality is provided with an analytical program 12. Furthermore, the analytical program 12 is designed to drive a program 13 for extracting a characteristic quantity that is an image processing program for extracting the characteristic (characteristic quantity) from a picked up image of the cell and extract the characteristic of the cell. The analytical program 12 is designed to drive a discrimination program 15 for judging the quality of the cell from the extracted characteristic or a combination of the plurality of characteristics and judge the quality of the cell. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a system for automatically determining the quality of a cell.

  The conventional cell culture apparatus does not have an automatic image analysis function for judging the quality of the culture state from the cell image. Therefore, it was necessary for the experimenter to take out the culture vessel from the culture apparatus by himself and observe the growth state of the cells using a separately installed microscope.

  In addition, when cultivating cells, it is known that the characteristics of the cultured cells may change as the generation progresses (when subspecies are generated), causing the subspecies to proliferate. It has been. In order to prevent this, cell culture experimenters have examined past cells to determine whether the cells they are using have lost their basic characteristics or have maintained their shape. I had to contact the supplier (cell bank, etc.) and pay attention to cell changes. In addition, in order to prevent this, it was necessary to immediately proliferate cells obtained from cell banks and the like, store them in a frozen state at an early stage, lyse and culture new cells every few months, and compare and confirm them.

  In recent years, biotechnology research fields have been diversified, so a lot of time has been spent in education in specialized fields, and there has been a lot of research on cell culture, and the number of researchers who cannot accurately judge the quality of cultured cells has increased. Yes. In addition, when a large number of cells or many types of cells are cultured, it takes a lot of time for the experimenter to determine the quality of each cell. In addition, since the criteria for determination are unclear, the quality of cultured cells varies for each experimenter, and the reproducibility of experiments using cultured cells is a problem.

  This invention is made | formed in view of such a situation, and makes it a subject to provide the automatic quality determination system of the cell which can implement | achieve automation of the quality determination of a cell.

The first means for solving the above-mentioned problem is a plurality of feature extraction means for extracting the features of the image of the captured cell,
A plurality of quality determination means for determining quality of the cell from the extracted features or a combination of a plurality of features;
An analysis recipe storage means for storing a combination of the feature extraction means and the pass / fail judgment means as an analysis recipe to be used according to the type of the cell;
The analysis recipe corresponding to the cell type is selected from the analysis recipe storage means, the feature extraction means stored in the analysis recipe is driven to extract the characteristics of the cell, and then the analysis recipe An automatic quality determination system for cells, comprising analysis means for driving the stored quality determination means to determine the quality of the cells.

The second means for solving the above-mentioned problem is a feature extraction means for extracting the characteristics of the image of the imaged cell,
A plurality of quality determination means for determining quality of the cell from the extracted features or a combination of a plurality of features;
According to the type of the cell, an analysis recipe storage unit that stores a combination of the feature to be extracted by the feature extraction unit and the quality determination unit to be used as an analysis recipe;
The analysis recipe corresponding to the cell type is selected from the analysis recipe storage means, and the feature extraction means stored in the analysis recipe is passed to the feature extraction means to extract the features of the cells. Subsequently, an automatic quality determination system for cells, comprising: analysis means for driving the quality determination means stored in the analysis recipe to determine the quality of the cells.

  A third means for solving the above-mentioned problem is the first means or the second means, wherein the characteristics are a growth state, the number of cells, a cell size, a feature of a cell outline, a cell-to-cell Adhesion, Adjacent Adhesion, Nucleus Number, Nucleus Shape, Nucleus Size, Nucleolus Number, Nucleolus Size, Nucleolus Concentration, Cytoplasmic Granule Amount, Cytoplasmic Granule Luminance The size of the vacuole, the number of vacuoles, the shape of the cytoplasm, the occupied area, and the crowded state.

  ADVANTAGE OF THE INVENTION According to this invention, the automatic quality determination system of the cell which can implement | achieve automation of the quality determination of a cell can be provided.

  Hereinafter, an automatic quality determination system for cells according to an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a diagram showing an overview of an automatic quality determination system for cells, which is an example of an embodiment of the present invention. In the following description, the term “feature amount” is used. In many cases, the feature is represented by a quantity, and thus, this is done. However, these “feature quantities” are concepts including those not represented by a quantity (for example, presence or absence).

  An automatic culture apparatus 1 having an automatic quality determination system has an analysis program 12. The analysis program 12 drives a feature amount extraction program 13 which is an image processing program for extracting the feature (feature amount) from the captured cell image to extract the feature of the cell. Subsequently, the identification program 15 that determines the quality of the cell from the extracted feature or a combination of a plurality of features is driven to determine the quality of the cell.

The amount of features to be extracted from the captured cell image depends on the cell type, growth state (single, colony formation, sheet-like), number of cells, cell size, and features of cell outline (spherical, foaming). Shape, jagged shape), adhesion between cells, connectivity between adjacent areas,
Number of nuclei, shape of nuclei, size of nuclei, number of nucleolus, size of nucleolus, concentration of nucleolus, amount of cytoplasmic granules, brightness of cytoplasmic granules, size of vacuoles, vacuoles Number, length of cytoplasm projections (number, length, thickness), occupied area, crowded state, etc.

  These features are extracted by appropriately using existing image processing programs. For example, the size of the nucleus can be extracted by image edge extraction by Laplacian or texture analysis by histogram feature amount. For example, if the Fourier transform is performed on the number of nuclei, a difference occurs in the spectrum amplitude of the high-frequency component inside and outside the nuclei. Thus, the number of nuclei can be determined by extracting the region of the nuclei.

  In addition, when the cytoplasmic projection is observed with a phase difference optical system, the cytoplasmic projection portion is observed to be blackened. Therefore, if binarization is performed in the vicinity of the background gradation and then the sesame salt noise is removed, only the cytoplasmic protrusions can be extracted. When the average value of the lengths of the straight line portions is obtained, the “average length of cytoplasmic projections” can be measured.

  When observed with a bright contrast phase difference optical system, it is possible to separate the gradation of cells and background. Therefore, the cell outline can be extracted by removing the sesame salt noise after binarization with the background threshold value measured in advance. For the connected images, “complexity” is calculated, and those larger than a certain threshold are removed as dead cells. In addition, connected images that are clearly larger than the standard size (for example, 3 to 4 times) are regarded as a mass of multiple cells, and an image skeleton is extracted. Cell image. Finally, if the area of the connected image is obtained, the cell size can be obtained.

  In this embodiment, a feature quantity extraction program 13 is prepared for each of these features, and which feature quantity extraction program 13 is used is stored in the analysis recipe 14 stored in the recipe database 11. ing. The analysis recipe 14 is created for each cell. That is, it is not necessary to extract all these features for all the cells, and only the features necessary for the pass / fail judgment need be extracted according to the cell type. The analysis recipe 14 stores only a feature quantity extraction program 13 for extracting feature quantities necessary for quality determination.

  In this way, when a new cell is added as a determination target or when a criterion for pass / fail determination becomes new, the extracted feature amount is determined or updated only by rewriting the analysis recipe 14. Can do. Further, when it becomes necessary to extract a new feature, it can be dealt with by adding a feature amount extraction program 13.

  When the features necessary for the pass / fail judgment are extracted in this way (there may be a single feature, but in many cases, a combination of a plurality of features), the analysis program 12 refers to the analysis recipe 14. Then, the identification program 15 stored therein is started. The identification program 15 is for determining the quality of a cell from the extracted feature or a combination of a plurality of features, but usually one is determined corresponding to the cell to be determined. However, there may be a case where a common identification program 15 can be used for a plurality of types of cells. In this case, it is not necessary to create the identification program 15 for each cell, and only the use of the common identification program 15 is stored in the analysis recipe 14 for each cell.

  The identification program 15 determines the quality of the cell using a statistical analysis means such as multivariate analysis or a pattern recognition algorithm such as a neural network.

  FIG. 2 is a flowchart summarizing the operation of the above-described cell automatic quality determination system. First, in step S1, the cell type is manually input. Then, the automatic quality determination system selects the analysis recipe 14 corresponding to the cell type in step S2. In step S3, one feature quantity stored in the analysis recipe is selected. In step S4, the feature quantity extraction program 13 corresponding to the selected program is selected and activated in step S5. Then, the feature quantity extraction program 13 extracts and records the feature quantity (step S6). In step S7, it is determined whether or not all the feature values stored in the analysis recipe 14 have been detected. If there are undetected ones, the process returns to step S3.

  When all the feature values stored in the analysis recipe 14 are extracted, the process proceeds to step S8, the identification program 15 stored in the analysis recipe 14 is selected, and activated in step S9. Then, the quality of the cell is determined by the action of the identification program 15 (step S10).

  In the above embodiment, the feature quantity extraction program 13 is created in correspondence with the feature quantity to be extracted. However, all feature quantities can be extracted by using the feature quantity extraction program 13 as one. In this way, the same effect can be obtained by bypassing the process of extracting feature amounts that are not stored in the analysis recipe 14. Further, one feature amount extraction program 13 may simultaneously extract a plurality of feature amounts (for example, the number and size of cells and the number and size of vacuoles). In this case, what is extracted in the analysis recipe 14 and actually used is stored, and the information is passed to the feature quantity extraction program 13 so that unused feature quantities are not used for quality determination. Alternatively, the corresponding feature amount extraction program 13 may not receive or use such information. Also, even in the same feature quantity extraction program 13, there are a plurality of internal processing methods, and when these are selected and used, similarly, which processing is used in the analysis recipe 14 is determined. The information may be stored and passed to the identification program 15 when the feature amount extraction program 13 is activated.

  Table 1 shows the recipe for extracting feature values stored in the analysis recipe 14 for the MK cells (monkey kidney cells) and PC12 (cells derived from rat brown cell types), and Table 2 shows the identification program 15 to be activated. . In the case of MK, the feature quantity is the number of nuclei, the size of the cell, and the number of vacuoles. However, the number of nuclei and the size of the cell are extracted by the program A which is the same feature quantity extraction program 13. ing. The number of vacuoles is extracted by the program B which is the feature quantity extraction program 13. The program X is used as the identification program 15.

In the case of PC12, the feature quantity is the number of nuclei, the size of the cell, the average length of the cytoplasm, and the number of vacuoles. It is extracted by a certain program D. The number of nuclei is extracted by the program E which is the feature quantity extraction program 13. Program C extracts the size of the test cell. The program Y is used as the identification program 15.
(Table 1)

(Table 2)

The identification program determines the quality of the cell based on the extracted feature values according to the criteria shown in Tables 3 and 4.
(Table 3)

(Table 4)

It is a figure which shows the outline | summary of the automatic quality determination system of the cell which is an example of embodiment of this invention. It is the flowchart which put together the operation | movement of the automatic quality determination system of a cell.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Automatic culture apparatus, 11 ... Recipe database, 12 ... Analysis program, 13 ... Feature quantity extraction program, 14 ... Analysis recipe, 15 ... Identification program

Claims (3)

A plurality of feature extraction means for extracting features of the imaged cell image;
A plurality of quality determination means for determining quality of the cell from the extracted features or a combination of a plurality of features;
An analysis recipe storage means for storing a combination of the feature extraction means and the pass / fail judgment means as an analysis recipe to be used according to the type of the cell;
The analysis recipe corresponding to the cell type is selected from the analysis recipe storage means, the feature extraction means stored in the analysis recipe is driven to extract the characteristics of the cell, and then the analysis recipe An automatic quality determination system for cells, comprising: an analysis means for driving the stored quality determination means to determine the quality of the cells. Feature extraction means for extracting features of the image of the imaged cell;
A plurality of quality determination means for determining quality of the cell from the extracted features or a combination of a plurality of features;
According to the type of the cell, an analysis recipe storage unit that stores a combination of the feature to be extracted by the feature extraction unit and the quality determination unit to be used as an analysis recipe;
The analysis recipe corresponding to the cell type is selected from the analysis recipe storage means, and the feature extraction means stored in the analysis recipe is passed to the feature extraction means to extract the features of the cells. Then, an automatic quality determination system for cells, comprising: analysis means for driving quality determination means stored in the analysis recipe to determine the quality of cells. The characteristics are growth state, number of cells, cell size, characteristics of cell outline, adhesion between cells, connectivity between adjacent cells, number of nuclei, shape of nucleus, size of nucleus, number of nucleolus Nucleolus size, nucleolus concentration, cytoplasmic granule amount, cytoplasmic granule brightness, vacuole size, number of vacuoles, shape of cytoplasm, occupied area, crowded state 2. The automatic quality determination system for cells according to claim 1, wherein

Images