WO1989003431A1

WO1989003431A1 - Cell culture monitor method

Info

Publication number: WO1989003431A1
Application number: PCT/JP1987/000750
Authority: WO
Inventors: Fumihiko Yonemori; Masayoshi Yamaguchi
Original assignee: Sumitomo Electric Industries, Ltd.
Priority date: 1987-10-06
Filing date: 1987-10-06
Publication date: 1989-04-20

Abstract

This invention relates to a cell culture monitor method which digitalizes a cultured cell image inputted to an image processor through a TV camera connected to an observation means, measures the number of cultured cells and the state of their propagation by conducting processing by use of a spatial filter and can thus count accurately and rapidly the number of cells.

Description

Specification

Cell culture monitoring method

Technical field

Akira Homoto sought a method of monitoring cell culture, and in more detail, digitized culture images captured by an image processing device, and processed live cells by processing using an air filter. Cell culture monitoring, which consists of measuring the number of cells and the cell culture status of the cells.

A place where cells are cultivated in containers such as microphone plates, dishes, and culture bottles. If the concentration of the production becomes too high and it becomes a state called fulgus, it will die. . It is necessary to change the culture medium or to carry out hybrid culture before it becomes fructose. So far, humans have been observing the state of fine breast breeding in containers using Ying Kang, and have decided to use this to determine when to exchange or cultivate cell culture. But force According to the conventional method, it was very ambiguous and the quantification of the cultured cells was unclear. In addition, it was very difficult to grasp the growth status of the culture of the daisies.

In addition, there are cases where it is necessary to measure both the number of production and the number of small vagina at the same time, such as when performing the sigma-binding by the limiting dilution method or when analyzing the shaku. However, even if blood cell measurement is used, the error will be large and the measurement will be irritated by the sighting system of Yingye.

-Instead of observing by scholarship, a method of recognizing teeth using image processing has been proposed. In this case, it is difficult to approve the background when the background such as the colony on the agar culture is simple.钧 is included, so the pattern and the background are both complex.

Departure J¾ © Purpose Another object of the present invention is to provide a cell culture method capable of accurately counting the number of cells even when having a complicated pattern and background.

The purpose of ^ of Takaaki Honaki is to provide a method for monitoring a cell culture which can perform a large amount of processing in a short time.

Structure of the invention

According to the present invention, the number of living cells in culture can be obtained by digitizing the cultured cell image which is embroidered on the observation means, captured by the image processing crab via the TV camera, and processed using the air filter. Also, a cell culture method specializing in measuring the proliferation state is provided.

The cell culture monitoring method of the present invention will be described with reference to the accompanying drawings. The image of the cells in the culture vessel is input to the image processing device using a TV camera attached to the microscope, which is an observation means. Perform image processing according to the procedure shown in Fig. 1 and count the number of living cells in culture. After the binarization, it is preferable to carry out a 辗 treatment. Well, by the procedure shown in Fig. 2, If measurement of the number of cultures and the number of saturates can be performed.

Cultivation 鎺跑 image acquisition

-Displacement is performed at the focal point allocation where the brightness of the center of the culture is bright and the brightness of the periphery is dark. Normally, by shifting the focus of the incisor by 5 ~ ± 250 # ιη, the brightness of the 細 ho and the circumference S of the live artillery is reduced and the image becomes darker, and the center latitude becomes a brighter image. If the defocus is less than 5 # β, the entire raw tendon remains a dark image.If the defocus is larger than 250, the image becomes unclear, so within the range S of ± 5 to 250 It is appropriate to shift the focus.

At this point, when one of the cultures has a contact ratio of ρ in contact with the square of the specified size of the empty matrix, that is, the size of the empty filter matrix is also nn. Pixels (n≥5>) and the size of the culture production is generally 10-21a, so the size of a line from 10 / nXi to 21 / nX2 2 / η [Ίπθ and-are taken in at a rate of- If the size of the filter matrix is too large or too small, the recognition rate will deteriorate. A spatial filter with a size of 5 × 5 pixels or more is used. This is because even if a pixel smaller than 5 x S pixels is used, the recognition efficiency is improved, and it is not practical.

The ^ -shape of the spatial filter is a circular shape, in which the coefficients of the periphery are negative, the coefficient of the center is positive, and the other coefficients are 0 (Zσ). However, only the production area can be enhanced. As an example, Fig. 3 shows the shape and coefficients of a 7x7 pixel spatial filter. ,

A filter to emphasize the production component, the shape of the target cell, and the brightness distribution pattern of the target living cell image are collected for multiple cells, and the maximum pattern is obtained by obtaining the flat pattern. be able to. This is because the spatial filtering also emphasizes the high degree of integration between the spatial filter and the shape (reference pattern). This is because the shape of the cell is the cell pattern itself. Fig. 4 shows the shape and coefficients of a 7 × 7 pixel filter for a simple filter.

In this case, the value of the center pixel y ij is given by:

— 1 + 3 + 3

A S1- -3 η = -3 x: Wandering of the gallery in front of Puerto Tarinda.

y 素 of the national element after filtering

A ·: Coefficient

ίι _Ηη : Teacher of each pixel of filter 1

Decisions are made. and two Ueka in ϋ those threshold, the pixel of the pixel ho no茚分the portion of the living cells, "0, to _δ胬饞the two攛化the eyes at one豳瘃hazel measured raw钿胞Calculate the maximum binarization threshold value such that the number or actual number of living teeth and the number of broken pixels of ⁴ pixels that are considered to have living cells after binarization are the same, and obtain multiple images The average value of the minute can also be regarded as the fixed planting。 because the brightness of Pacglan is not related to the brightness and the correlation of the floor is observed. — The number of _d 1 living cells that are generated based on the number of 索 mins of field 1 "cocoon prime number) becomes almost constant, and the pixel occupancy is almost constant.

^ tei — Calculates the number of living cells according to the number of pixels of a living vesicle cell.

D: Before calculating the number of pixels from the image ^:

By performing one or more of the 8 recent inflation treatments or a combination thereof, it is possible to determine the difference that occurs when the number of pixels is converted into the number of cells. This is because by performing the expansion process, it is possible to repair the degenerated wisteria and the missing cocoon image, and because the number of pixels occupied by one cell increases, the child of one pixel becomes a child. It is because it becomes less. When the image after the tooth alignment by the empty ring filter ring is found, the birth of kito is the beginning of many years, many of which are re-delicated at a low value, invert to the bira and die. The field element of the area where 存在 exists is set as “、”, and the pixel of the area where 钿遛 does not exist is set as “-0”. ^ Number of pixels per minute assuming there are cells

Calculate the number of dead cells based on (· prime number). 1 餹 dead cells Since the number of pixels occupied by is approximately constant, the equation:

^^ ^-Pixels when there is ¾

^ ^mm '-Death 画素 1 橱

Follow the steps to find the garden.

At this time, depending on the plant, the production benefit part (翰 :) may be inserted: the image after the space filter is requested. Binary images binarized by fighting for the number of viable cells or binary images improperly expanded are removed from the binary image binarized and decomposed by subtracting them on the image. can do. Ama here! The process is repeated several times. § §1 If too much processing is performed, necessary images will be erased. Noise around the living cells ^ By performing one or more of four near-swelling or s near-swelling treatments or a combination of these on one or more images, the same as in the live case The error that occurs when converting the number of cells from birch to the number of cells. _ One g one

The combination of image capture at the focal point where the brightness of the surrounding area is dark and the brightness of the central area is bright, and the live cell image enhancement using an empty filter ^, only the cells having the shape and characteristics of living cells As a result, live cells and dead cells can be easily separated, and the number of living cells in a complex pattern and in the background can be extracted to accurately count the number of cells. In addition, a large amount of cultured cells can be processed in a short time.

Therefore, it is possible to quickly and appropriately determine the timing of exchanging the cultivated seeds and determining the culture time.

Example

Examples will be described below.

Example 1 "

A complex image, such as the presence of a living cell in a dead cell, foreign body, etc., is shifted by 30 z ffi from a fully focused state to a complex image. Image processing equipment for cell images f PIAS-1 (produced by Nippon PC Systems, Inc.), the rate at which the cultured live cell sections are inscribed in a 7 × 7 pixel empty filter, that is, the i-pixel size is 1.8 # ffl X 1. The image was digitized using the image processing equipment FIAS-1 with the agility rate of S m, and the image was filtered using a spatial filter with a shape extension coefficient as shown in No. 513. If the false recognition rate is within the range of 10%, the error rate is 67%, if the false recognition rate is within the range of 20%.

A 70% recognition rate was obtained. Here, the accusation rate and the recognition rate can be recognized by image processing.

Number of living cells present

Deception rate-The number of raw pongs that actually exist in the original image

Number of live cells assumed to exist in image processing __

False recognition rate

The number of viable cells determined to be present by image processing was determined. Example 2 A complex g-image コ such that there is a co-σ 21 of living cells in a dead cell, foreign body, etc., and the focus is shifted by 30 m from the completely focused focus state. The cell image was taken into the image processor PIP-40000 (ADS (铮)) through the image reading device of the CCD camera.

As the magnification, the rate at which the outline of the cultured live cell is inscribed in the empty filter of the 7 x 7 grid, that is, the rate at which the size of one pixel is 2.4 2.4 iim, is used.

The digital image is digitized by the image processing device PI 00-40000, and a filter with a shape and coefficient as shown in Fig. 6 is filtered and binarized based on the number of productions by visual measurement. When the threshold was used, a false recognition rate of 10% was obtained in the region of 73%, and a rubber recognition rate of 205% was 8 I%. If only the binarization process is performed on the same image, the recognition rate is 5 l% in the range of 10% and the recognition rate is 60% in the range of 20%. .

An image in which live cells and dead cells existed sparsely was captured by the above method, and the numbers of live cells and dead cells were counted. Regarding the number of viable cells, the number of image processing measurement δ 0 The cell number was 4%, and the error was 0% in the total value of the number of image processing cells, which was 45.3 in comparison with the visual observation. Further, ShimegiYasushi儒差measurement error in 3 9 Samburu total钿跑number (viable Hoso跑number +%钿齒number) is soil 1 in 1 Sigma 0, 2% and was the _β where error , Image processing t measurement-visual measurement

It was determined by the target number of measurements. Example 2 Except for using a binarization method based on the number of seedling cells instead of a binning bench value based on the number of living cells based on the target measurement, Example 2 I returned. Implementation If the spatial filter with the same shape coefficient as that of If 2 is also filtered, the recognition rate, 7 3 おいて for 1 鼷籙, and 8 2% for 20 おいてMet. Also, the error of converting the return prime number into the number of viable cells by performing the re-expansion once for the image after binarization is 8% from 17% when the expansion process is not performed. It became%. Brief description of drawings

Fig. 1 is a flowchart showing the procedure for counting the number of living cells. Fig. 2 ^, the flowchart showing the procedure for counting the number of living cells and the number of cells.

O "" chart,

FIG. 3 is a diagram showing the shape and coefficient of the spatial filter of the 7 × 7 grid, and FIG. 4 is a diagram showing the shape and coefficient of an optimized empty filter of 7 × 7 pixels.

FIG. 5 and FIG. 6 are diagrams showing the size and relationship of the air filters used in Example 1 and Example 2, respectively.

Claims

The scope of the claims

(1) Observation means Images of cultured cells captured in the image processing bag via a connected TV camera are digitized, and processed using a blank filter. Cell culture monitoring method characterized by measuring proliferation 扰篛

2) The method of claim 1, wherein the number of living cells in culture is measured.

(3) The method of monitoring the culture of claim 1, wherein the method of measuring the production of cultured cells and cells is described.

(4) The focus is shifted from the point of perfect focus to obtain an image in which the center of the cultured culture is bright and the circumference of the periphery is dark. The cell culture monitor according to any one of claims 1 to 3, characterized in that a culture image is captured at the position.

(5) From the point where the focus is perfectly focused. One 1 δ

Incorporating culture images in a new market: and the method of monitoring cultures described in item 4 ia of the specialty of the contract.

(6) The size of the air filter "" is n x n

(ti 5), wherein the cell culture 5 & nita method _{e according} to claim 5

(7) The cell culture according to claim 6, wherein the carrier of the cultured cells is capable of taking in culture images at an inscribing rate of matrix square of a set-size spatial filter. Monitor method.

(8) The cell culture according to claim 6 or 7, wherein the image of the cultured cell is taken in at a rate such that the size of one branch is 10 / n-21 / nκι. Monitor "how".

(9) It is important to emphasize the cultured live cell image using an empty rush filter with a set size of a circular rush filter having a circular shape, a central coefficient of positive and a peripheral coefficient of negative or 0. 9. The method for monitoring a culture according to claim 1, claim 6, or claim 8 in the claims. (10) The shape of the blank filter with the set size is the same as the shape of the blank filter obtained from the average pattern of the cell luminance distribution pattern.

In addition, a method for monitoring a cell culture according to any one of the first, sixth, eighth, and ninth aspects of the present invention, which is characterized by enhancing a cultured cell image.

(1 1) The image after the spatial filtering processing is binarized, and from the binary image,

_Am ^ _-Total number of pixels with live cells ■

Fine ^ ⁼⁼ "Number of pixels occupied by TS¾ raw fine teeth

Calculate the number of cultures required by ::

The method for monitoring culture according to any one of Items 2 to 10.

C1 2) Multiple digitization thresholds such that the number of target measurement cells and the number of seedlings in one image match the number of roses that indicate that there is a post-binary raw fine The method of claim 1, wherein the classification, the average value, and the fixed binary value are reduced.

C13> After Kuoryu filtering © The cocoon image is binarized, and the number of viable cells is calculated based on the number of pixels after 4 or 8 dilations. The method for monitoring a culture according to any one of claims 2 to 11, wherein the method is characterized in that:

(14) The image after emptying filtering was binarized at a 闞 value lower than the threshold value for determining the number of viable cells, and further recultivated.

It is assumed that there are dead cells.

Female

Number of pixels occupied by 1-value dead cells

Claim 3 characterized in that the number of dead cells required in the step is calculated.

- 1 2 wherein the cell culture Nita ¹ "" The method according to any one of.

(15) Binary binarization of the image after blanking filtering using a threshold for determining the number of dead cells, and binarization from the binary image that has been re-tiled, with binarizing the number of viable cells, or 15. The cell culture device according to claim 14, wherein the binary image subjected to the appropriate expansion process is subtracted on a screen to extract a component corresponding to dead cells. Method.

(16) With respect to the binary image from which a portion corresponding to a dead cell is extracted, one or more times, 4 near expansion, 8 near expansion processing, or a combination thereof is performed, and then a fine number is obtained. Claim 5 Item 5. A method for monitoring a cell culture according to any one of the items _δ.