JP7235027B2 - Cell evaluation instruments, incubators and programs - Google Patents

Description

TECHNICAL FIELD The present invention relates to a cell evaluation device, an incubator, and a program.

Stem cells such as ES (Embryonic Stem) cells and iPS (induced Pluripotent Stem) cells are capable of differentiating into all tissues and proliferating indefinitely, so they are applied in, for example, regenerative medicine. . Human ES cells and iPS cells are also differentiated and cultured into nerve cells.

It is known that when stem cells are differentiated and cultured into nerve cells, a unique and characteristic structure emerges in the cells during the differentiation process of nerve cells. Therefore, in evaluating the state of differentiation into nerve cells, it is effective to evaluate structures inherent in the process of differentiation into such nerve cells.

The evaluation of the state of the structure as described above is carried out visually. However, the evaluation by visual observation requires human labor, and is highly dependent on the subjectivity of the observer and cannot be quantified. In addition, for example, since the cells are taken out of the incubator and observed, the quality of the cells deteriorates during observation, and the quality becomes unstable.

Therefore, for example, a method of confirming the presence of proteins corresponding to structures in nerve cells by staining them with a fluorescent antibody is known (see, for example, Non-Patent Document 1). With this technique, the structure can be specified by the state of the stained cells, so that the evaluation results can be reproducible.

Leif Dehmelt, Gunnar Poplawski, Eric Hwang and Shelley Halpain "NeuriteQuant: An open source toolkit for high content screens of neuronal morphogenesis"

However, in the method of Non-Patent Document 1, the work for staining the cells requires a relatively long time and effort, which imposes a heavy burden on human resources. In addition, in the technique of Non-Patent Document 1, cells are stained and invasive, so the cells to be observed cannot be used continuously for differentiation culture.

The present invention has been made in view of such circumstances, and an object of the present invention is to enable noninvasive and stable evaluation of the state of differentiation into nerve cells while reducing human labor.

In order to solve the above-described problems, a cell evaluation apparatus as one aspect of the present invention includes an image acquisition unit that acquires an image of nerve cells that is transilluminated and noninvasively captured, and an image of the acquired image. An image region whose brightness is below a certain level or above a certain level is extracted, and a portion of the single-focal image in which a shape created by a change in the brightness of the image area is a thin line whose brightness is lower than that of the surrounding area is defined as the nerve cell. Extract as neurites, extract granular parts with a certain or more brightness gradient at the edge of the image region in the single focus image as dead cells, and in the single focus image, the brightness of the edge of the image region with a certain area or more a first extraction unit for extracting a portion having a constant value or more as living cells ; and determining the growth state of the nerve cells based on the number of the living cells or the dead cells extracted by the first extraction unit. and a first determination unit.

Further, an incubator as one aspect of the present invention includes a temperature-controlled room for culturing cells, and an observation unit for capturing the images of cells.

Further, the program as one aspect of the present invention comprises an image acquisition step of acquiring an image of nerve cells non-invasively captured by transillumination in a computer; Extracting the above image area, extracting a portion in which the shape created by the change in brightness of the image area in the single focus image is a thin line whose brightness is lower than the surrounding area as a neurite of the nerve cell, In the single focus image, a granular portion with a brightness gradient of the edge of the image region above a certain level is extracted as a dead cell, and in the single focus image, a portion with an area above a certain level and a brightness at the edge of the image area above a certain level is extracted. a first extraction step of extracting live cells ; and a first determination step of determining the growth state of the nerve cells based on the number of the live cells or the dead cells extracted by the first extraction step; is executed.

INDUSTRIAL APPLICABILITY As described above, according to the present invention, it is possible to stably and noninvasively evaluate the state of differentiation into nerve cells while reducing human labor.

1 is a front view showing a structural example of an incubator according to an embodiment of the present invention; FIG. It is a figure which shows the structure example of the incubator in this embodiment from a plane. It is a figure which shows the structural example of the incubator of this embodiment. FIG. 2 is a diagram showing the relationship between the process of differentiation from stem cells to nerve cells and phenomena occurring in the cells. FIG. 2 is a diagram showing rosettes and neurites appearing in the process of differentiation into nerve cells. 4 is a flow chart showing an example procedure for the incubator of the present embodiment to evaluate the state of differentiation of cells that are induced to differentiate into transcytic cells. FIG. 4 is a diagram showing a more detailed procedure example for the incubator of this embodiment to determine the state of the rosette. FIG. 4 is a diagram showing a more detailed procedure example for the incubator of the present embodiment to determine the state of neurites; It is a figure which shows typically the generation|occurrence|production method of the omnifocal image by the omnifocal image generation part of this embodiment. FIG. 5 is a diagram showing an example of the procedure of image processing executed by the rosette extracting unit of the present embodiment to extract rosettes; It is a figure which shows the example of a result about the soft matching which the Rosetta extraction part of this embodiment performs. FIG. 10 is a diagram showing an example of a result when the rosette extracting unit of the present embodiment performs processing for extracting a region with a low density difference from an omnifocal image; FIG. 10 is a diagram showing an example of noise removal results for regions with low density differences by the rosette extraction unit of the present embodiment; FIG. 10 is a diagram showing an example of a result of rosette extraction processing executed by a rosette extraction unit using an image after soft matching according to the present embodiment and an extracted image of a region with a low density difference; FIG. 4 is a diagram showing an example of neurites, dead cells, and live cells extracted from a phase-contrast image by the neurite extraction unit of the present embodiment; FIG. 4 is a diagram showing a processing procedure example for extracting neurites, dead cells, and live cells by the neurite extraction unit of the present embodiment. FIG. 10 is a diagram showing an example of a processing result of soft matching executed by the neurite extracting unit of the present embodiment in response to extraction of dead cells and nerve cells; Examples of results of closing and binarization performed by the neurite extraction unit of the present embodiment to extract living cells, and differential image generation and binarization performed by the neurite extraction unit to extract neurites are shown. FIG. 4 is a diagram showing; FIG. 10 is a diagram showing an example of the results of binarization and common object extraction processing executed by the neurite extracting unit of the present embodiment for extracting living cells; FIG. 10 is a diagram showing an example of an image used when generating a difference image executed by the neurite extracting unit of the present embodiment for extracting nerve cells; FIG. 10 is a diagram showing an example of a result of processing from thinning to neurite extraction performed by the neurite extraction unit of the present embodiment for extracting nerve cells; FIG. 10 is a diagram showing an example of processing results of dilation and differential image generation performed by the neurite extracting unit of the present embodiment for extracting dead cells; FIG. 10 is a diagram showing an example of a result of common object extraction processing executed by the neurite extracting unit of the present embodiment for extracting dead cells; FIG. 10 is a diagram showing an example of segmentation performed by the neurite extraction unit when extracting dead cells and live cells.

[Incubator configuration example]
An incubator 11 to which a cell evaluation device according to an embodiment of the present invention is applied will be described with reference to FIGS. 1 to 3. FIG.
1 and 2 show a structural example of the incubator 11. FIG. 1 is a front view of the incubator 11, and FIG. 2 is a plan view of the incubator 11. FIG. 3 shows a configuration example of the incubator 11. As shown in FIG.

The incubator 11 shown in these figures has an upper casing 12 and a lower casing 13 . The upper casing 12 rests on the lower casing 13 in the assembled state of the incubator 11 . The internal space between the upper casing 12 and the lower casing 13 is vertically partitioned by a base plate 14 .

First, inside the upper casing 12, a temperature-controlled room 15 for culturing cells is provided. The temperature-controlled room 15 has a temperature control device 15a and a humidity control device 15b, and the temperature-controlled room 15 is maintained in an environment suitable for culturing cells (for example, an atmosphere with a temperature of 37° C. and a humidity of 90%) ( 1 and 2, illustration of the temperature control device 15a, the humidity control device 15b, the collection device 51, the seeding device 52, and the medium exchange device 53 is omitted).

A large door 16, a middle door 17, and a small door 18 are arranged in front of the temperature-controlled room 15. - 特許庁The large door 16 covers the front surfaces of the upper casing 12 and the lower casing 13 . The middle door 17 covers the front surface of the upper casing 12 and isolates the environment between the temperature-controlled room 15 and the outside when the large door 16 is opened. The small door 18 is a door for loading and unloading the culture vessel 19 for culturing cells, and is attached to the middle door 17 . Carrying in and out of the culture container 19 through the small door 18 makes it possible to suppress environmental changes in the temperature-controlled room 15 . The large door 16, the middle door 17, and the small door 18 are kept airtight by packings P1, P2, and P3, respectively.

A stocker 21 , an observation unit 22 , a container transfer device 23 , and a transfer table 24 are arranged in the temperature-controlled room 15 . Here, the carrier 24 is arranged in front of the small door 18 and the culture container 19 is carried in and out through the small door 18 .

The stocker 21 is arranged on the left side of the temperature-controlled room 15 when viewed from the front surface of the upper casing 12 (lower side in FIG. 3). The stocker 21 has a plurality of shelves, and each shelf of the stocker 21 can accommodate a plurality of culture vessels 19 . Each culture container 19 contains cells to be cultured together with a culture medium.

The observation unit 22 is arranged on the right side of the temperature-controlled room 15 when viewed from the front of the upper casing 12 .
Here, the observation unit 22 is arranged by being fitted into the opening of the base plate 14 of the upper casing 12 . The observation unit 22 has a sample stage 31, a stand arm 32 projecting above the sample stage 31, and a main body portion 33 containing a microscopic optical system for phase contrast observation and an imaging device 33a. The sample stage 31 and the stand arm 32 are arranged in the temperature-controlled room 15 , while the main body portion 33 is accommodated in the lower casing 13 .

The sample table 31 is made of a translucent material, and the culture vessel 19 can be placed thereon. The sample table 31 is configured to be horizontally movable, and the position of the culture vessel 19 placed on the upper surface can be adjusted. Also, the stand arm 32 incorporates an LED light source 33b. The imaging device 33a can acquire a microscopic image of the cells by imaging the cells in the culture container 19, which are transilluminated from above the sample stage 31 by the stand arm 32, via the microscopic optical system.

The container conveying device 23 is arranged in the center of the temperature-controlled room 15 when viewed from the front of the upper casing 12 . The container transfer device 23 transfers the culture container 19 between the stocker 21 , the sample table 31 of the observation unit 22 and the transfer table 24 .

The container transport device 23 also has a vertical robot 34 having an articulated arm, a rotary stage 35 , a mini stage 36 and an arm section 37 . The rotating stage 35 is attached to the tip of the vertical robot 34 via a rotating shaft 35a so as to be horizontally rotatable by 180°. Therefore, the rotating stage 35 can have the arm portions 37 opposed to the stocker 21 , the sample table 31 and the carrier 24 .

Also, the mini-stage 36 is attached to the rotary stage 35 so as to be horizontally slidable. An arm portion 37 for gripping the culture container 19 is attached to the mini stage 36 .

Next, the lower casing 13 will be explained. Inside the lower casing 13, the body portion 33 of the observation unit 22 and the control device 40 of the incubator 11 are accommodated.

The control device 40 is connected to the temperature adjustment device 15a, the humidity adjustment device 15b, the observation unit 22, the container transport device 23, the recovery device 51, the seeding device 52 and the medium exchange device 53, respectively. This control device 40 comprehensively controls each part of the incubator 11 according to a predetermined program.

As an example, the control device 40 controls the temperature adjustment device 15a and the humidity adjustment device 15b to maintain the temperature-controlled room 15 at predetermined environmental conditions. Further, the control device 40 controls the observation unit 22, the container transport device 23, the collection device 51, the seeding device 52, the culture medium exchange device 53, etc., and automatically executes the observation sequence of the culture container 19. FIG. Further, the control device 40 executes culture state evaluation processing for evaluating the culture state of cells based on the images acquired in the observation sequence.

The incubator 11 of the present embodiment is used, for example, to differentiate and culture human-derived skin cells into nerve cells. FIG. 3 shows an example of the functional configuration of the control device 40 corresponding to differentiation culture into nerve cells. The operation of each functional unit in the control device 40 is realized, for example, by causing a CPU (Central Processing Unit) forming the control device 40 to execute a program.

The control device 40 includes a storage unit 41, an image input unit 42, an omnifocal image generation unit 43, a rosette extraction unit 44, a neurite extraction unit 45, a Rosetta correspondence determination unit 46, and a nerve cell differentiation culture. A protrusion correspondence determination unit 47 , a comprehensive determination unit 48 and a device control unit 49 are provided.

The storage unit 41 stores various data required by the control device 40 .
The image input unit 42 inputs a captured image captured by the imaging device 33a. In this embodiment, the imaging device 33a captures a microscopic image observed by a microscopic optical system for phase-contrast observation, that is, a phase-contrast image obtained by converting the phase difference of light into contrast. In addition, the imaging device 33a changes the focal position along the Z direction (the depth direction of the sample) from the predetermined imaging position on the xy plane, performs imaging at each focal position, and outputs the imaging image. . Each captured image obtained in this manner is a single focus image. The image input unit 42 inputs these single focus images as phase contrast images.

The omnifocal image generation unit 43 generates an omnifocal image using a plurality of single-focus images input by the image input unit 42 . Specifically, the omnifocal image generator 43 extracts, for example, in-focus regions at positions in the cell thickness direction, and generates an omnifocal image by synthesizing these regions. Note that such an omnifocal image is also called an EDF (Extend Depth of Focus) image, for example.
For example, in the process of differentiation from stem cells to nerve cells, changes occur in the undulations of the cell surface according to the appearance of structures called rosettes. For this reason, in a single-focus image, there are both in-focus and high-contrast regions and out-of-focus and low-contrast regions. Since the position of the structure in the image is indefinite, it is difficult to obtain an appropriate judgment result even if a monofocal image is used to evaluate the differentiated state of nerve cells. Therefore, in this embodiment, an omnifocal image is used for evaluation when extracting the rosette 100 . Thereby, an appropriate evaluation result can be obtained.

The rosette extraction unit 44 extracts from the all-in-focus image (first original image) rosettes, which are one of the structures that appear uniquely in cells in the process of differentiation into nerve cells.

The neurite extraction unit 45 (structure extraction unit) extracts neurites, which are one of the structures in the process of differentiation into nerve cells, from the monofocal image (second original image). In the differentiation process in which neurites appear, dead cells (hereinafter also referred to as "dead cells") and cell bodies of living cells (hereinafter also referred to as "live cells") coexist in addition to neurites. be in a state to Therefore, the neurite extraction unit 45 distinguishes and extracts neurites, dead cells, and living cells.

The rosette correspondence determination unit 46 determines the state of the rosettes extracted by the rosette extraction unit 44 . In addition, the rosette correspondence determination unit 46 determines a predetermined item related to cell differentiation induction based on the determined state of the rosette.
A neurite correspondence determination unit (structure correspondence determination unit) 47 determines the state of the rosette extracted by the neurite extraction unit 45 . In addition, the neurite correspondence determination unit 47 determines a predetermined item related to cell differentiation induction based on the determined neurite state.
Based on the determination result of the Rosetta correspondence determination unit 46 and the determination result of the neurite correspondence determination unit 47, the comprehensive determination unit 48 determines a predetermined item related to cell differentiation induction.

The device control unit 49 controls operations of the imaging device 33a, the temperature adjustment device 15a, the humidity adjustment device 15b, the container transport device 23, the collection device 51, the seeding device 52, and the medium exchange device 53.

[Rosetta and neurites]
The incubator 11 of the present embodiment evaluates the differentiation state of nerve cells based on the state of rosettes and neurites that appear in the process of differentiation into nerve cells. Therefore, the rosette and the neurite will be described with reference to FIGS. 4 and 5. FIG.

FIG. 4 shows the process of differentiation of iPS (induced pluripotent stem) cells into nerve cells over time. When iPS cells are cultured to induce differentiation into nerve cells, the iPS cells differentiate into neuroepithelial cells, neural progenitor cells, and mature nerve cells in that order. Rosetta appears when neuroepithelial cells are differentiated into neural progenitor cells.

FIG. 5(a) shows an example of a phase-contrast image obtained by observing a cell with rosettes 100 appearing under a phase-contrast microscope.
The rosette 100 has, for example, a shape in which cylindrical cells are arranged radially. The rosette 100 has a characteristic shape and appears early as a unique structure in the process of differentiation into nerve cells. Therefore, the fact that the appearance of Rosetta 100 was confirmed indicates that differentiation of stem cells such as iPS cells into nerve cells was successfully induced.
Considering that the Rosetta 100 has the properties described above, it can be said that it is effective to evaluate the state of the Rosetta 100 in evaluating the state of differentiation into nerve cells. Therefore, in this embodiment, the state of the Rosetta 100 is first determined in order to evaluate the state of differentiation into nerve cells.

In addition, neurites appear in the process of differentiation from neural progenitor cells to mature nerve cells, and a phenomenon occurs in which the neurites elongate.
FIG. 4 shows an example using iPS cells, which are stem cells. When the differentiation culture is performed, the differentiation process is similar to that shown in FIG.

FIG. 5(b) shows the structure of a nerve cell 200. FIG. A nerve cell 200 is a structure having a cell body 210 and a neurite 300 (axon 220 and dendrite 230 ) extending from the cell body 210 .
The cell body 210 has a nucleus and has a round, protruding shape.
Axons 220 have the function of outputting signals. In addition, the axon 220 is thin, uniform in thickness, long, and has almost no endoplasmic reticulum or ribosomes. Also, the elongation speed of the axon 220 is high.
Dendrites 230 have the function of receiving signals. Also, the dendrite 230 is relatively thick, but becomes thinner toward the tip. Also, its extension speed is slow.
This nerve cell 200 is also one of the structures that appear inherently in the differentiation process of nerve cells. In addition, in the process of culturing neurons, a large number of neurons 200 overlap each other, making it difficult to distinguish between axons 220 and dendrites 230 . Therefore, in the present embodiment, the axon 220 and the dendrite 230 are extracted as the neurite 300 as an elongated portion of the nerve cell 200 without distinguishing between them. At the same time, living cell bodies 210 in nerve cells 200 are also extracted as living cells.

[Evaluation procedure]
FIG. 6 shows a procedure example for the incubator 11 to evaluate the differentiation state of cells that are induced to differentiate into nerve cells.
First, in the incubator 11, human-derived iPS cells or ES cells are housed in the culture container 19 as stem cells to be cultured to induce differentiation into nerve cells (step S101). Alternatively, different types of differentiated cells may be accommodated in the culture vessel 19 for the purpose of inducing differentiation by direct reprogramming (step S102).

The stem cells accommodated in the culture vessel 19 in step S101 or S102 are cultured to induce differentiation into neuroepithelial cells (step S103). After differentiation into neuroepithelial cells, Rosetta 100 appears in the process of differentiation into progenitor cells (step S104). It should be noted that the cells in which the Rosetta 100 has appeared in this manner can be proliferated by differentiation culture by subculturing.

The control device 40 executes image processing for extracting the rosette 100 using the image (first captured image) captured by the imaging device 33a (step S105). For this purpose, the control device 40 causes the imaging device 33a to image the rosettes 100 that have appeared in the cells in the process of differentiation. As described above, the images captured by the imaging device 33a are phase-contrast images observed with a phase-contrast microscope, and are a plurality of single-focus images at different focal positions. After that, the control device 40 controls the imaging device 33a to capture a plurality of single-focus images, for example, at each imaging time at a predetermined time interval.

Then, the image input unit 42 in the control device 40 inputs a plurality of single-focus images captured by the imaging device 33a. In addition, the omnifocal image generator 43 sequentially generates one omnifocal image using a plurality of single-focus images captured at the same imaging time. Thereby, the omnifocal image generator 43 generates a plurality of omnifocal images corresponding to each imaging time. A plurality of omnifocal images generated in this manner are so-called time-lapse images in time series. Then, the rosette extraction unit 44 extracts the rosette 100 from the omnifocal image by image processing in step S105.

Next, the rosette correspondence determination unit 46 uses the rosettes 100 extracted in step S105 to determine the state of the rosettes 100 that have appeared in the cells (step S106). There are various judgment items here, and they are not particularly limited, but for example, the quality and quantity of the rosette 100 and the period required for its appearance are used as criteria to judge various judgment items.

As an example, the rosette correspondence determination unit 46 determines the position, differentiation state, differentiation efficiency, etc. of the appearing rosette 100 (step S107). As the differentiated state, specifically, the shape, quality, etc. of each rosette 100 or the entire rosette 100 is determined. In addition, the differentiation efficiency here can be obtained, for example, as the number of appearances of Rosetta 100 with respect to the number of colonies that have appeared.

Based on the determined state of the rosette 100, the rosette correspondence determining unit 46 determines the next step for the cell in which the rosette 100 appears (step S108) and predicts the future (step S109). In other words, the rosette correspondence determining unit 46 can further determine predetermined items related to cell differentiation and culture based on the determined state of the rosette 100 .

The next steps determined by the Rosetta correspondence determination unit 46 in step S108 include the following.
In other words, the rosette correspondence determination unit 46 can determine, as the next step, that the determination result of the state of the rosette 100 should be used as information for lot management.
In addition, the rosette correspondence determining unit 46 can determine, as the next step, that the cells in which the rosettes 100 have appeared should be collected and used as a sample for transplantation. When it is determined that the cells should be collected in this manner, the device control section 49 of the control device 40 controls the collection device 51 to collect the cells.
In addition, the rosette correspondence determination unit 46 can determine, as the next step, that the cells in which the rosettes 100 have appeared should be cryopreserved. Cryopreserved cells are used, for example, for transplantation, storage for mass production purposes, or lot control.
Further, the rosette correspondence determining unit 46 can determine, as the next step, that the cells in which the rosettes 100 appear should be sorted and examined. Incidentally, when it is determined that inspection should be performed in this manner, various inspections and analyzes are performed (step S110). This inspection and analysis may be performed after the cells are collected, or may be performed using the captured images without collecting the cells. In addition, the rosette correspondence determination unit 46 can determine, as the next step, to continue culturing the cells in which the rosettes 100 have appeared.

Further, the Rosetta correspondence determination unit 46 can perform the future prediction in step S109 as follows. In other words, the Rosetta correspondence determination unit 46 can predict the yield of differentiation into neurons. In addition, the Rosetta correspondence determination unit 46 can predict what type of neuron differentiation tendency, such as dopaminergic or glutamatergic, for example.

In addition, when the differentiation-inducing culture into neuroepithelial cells in step S103 is continued, the cells further progress to differentiation into neural progenitor cells (step S111). Neurospheroids appear at this stage of differentiation into neural progenitor cells (step S112). This neurospheroid can also be propagated by passaging.
In addition, after the appearance of neurospheroids, differentiation-inducing culture into nerve cells is performed (step S113). Neurites 200 appear in the process of differentiation-inducing culture in step S113 (step S114). Then, when the neurite 200 appears in this way, a phenomenon occurs in which the projection (axon 220 and dendrite 230) of the neurite 200 is elongated.

The control device 40 causes the imaging device 33a to image nerve cells in which the neurites 300 are elongated as described above, and extracts the neurites 300 using the captured image (second captured image) ( step S115). In the image processing for this neurite extraction, the image input unit 42 inputs a single focus image at each imaging time. Then, the neurite extraction unit 45 extracts the neurite 300 from the generated phase contrast image as the single focus image.

Next, the neurite correspondence determination unit 47 uses the neurites 300 extracted in step S115 to determine the state of the neurites 300 appearing in the cell (step S116). Items that serve as criteria for this determination include, for example, the rate of process extension, the number of living cells, the number of dead cells, the amount of increase in dead cells, and the proliferation rate of neurites 300 and living cells. The neurite correspondence determining unit 47 determines predetermined determination items based on these items, for example. It should be noted that the determination items regarding the neurite 300 are also diverse and are not particularly limited.

As an example, the neurite correspondence determination unit 47 determines the differentiation state and quality of the neurite 300 (step S117).

Further, the neurite correspondence determining unit 47 determines the next step for the neuron in which the neurite 300 appears (step S118).
The next steps determined by the neurite correspondence determining unit 47 in step S118 include the following.
For example, the neurite correspondence determination unit 47 can determine, as the next step, that the determination result of the state of the neurite 300 this time should be used as information for lot management.
In addition, the neurite correspondence determining unit 47 can determine, as the next step, that the nerve cells in which the neurites 300 appear should be collected and used as samples for transplantation.
Further, the neurite correspondence determination unit 47 can determine, as the next step, that the nerve cells in which the neurite 300 appears should be sorted and examined.
In addition, the neurite correspondence determining unit 47 can determine, as the next step, that the neuron in which the neurite 300 appears should be used for screening, assay, or the like.

Further, the comprehensive determination unit 48 makes a comprehensive determination using the determination results of the Rosetta correspondence determination unit 46 in steps S107 and S109 and the determination results of the neurite correspondence determination unit 47 in step S117 (step S119).

The comprehensive determination unit 48 can output, for example, the differentiation state and differentiation efficiency in the differentiation process including the appearance of the Rosetta 100 and the neurite 300 as the result of the comprehensive determination (step S120).
Further, the comprehensive determination unit 48 can determine the next step for the neuron at the stage where the neurite 300 appears as the comprehensive determination result (step S118).

Further, the comprehensive determination unit 48 can create a quality database reflecting the comprehensive determination result (step S121). The quality database stores, for example, the state of differentiation of the Rosetta 100, the state of differentiation of the neurite 300, the culture conditions and the period until the Rosetta 100 is differentiated to the stage where the neurite 300 emerges. It also stores gene and protein marker profiles. In addition, inspection results after final differentiation can be stored. This quality database is stored in the storage unit 41 .
Then, the content of the quality database can be fed back to, for example, the next culture step (step S122). Specifically, the comprehensive determination unit 48 sets various items in the next culturing step based on the contents of the quality database stored in the storage unit 41 . Then, the device control unit 49 notifies the user of the setting items, or automatically executes temperature and humidity adjustment, seeding, culture medium replacement, etc. according to the set items. This realizes the feedback of the overall determination result to the next culture step.

Further, after the neurites 300 appear, differentiation-inducing culture into mature neurons is performed (step S123). Then, the mature nerve cells are sorted for examination, for example (step S124). Also, the cells differentiated into mature nerve cells can be used for medical treatment or the like (step S125).

The flowchart of FIG. 7 shows a more detailed procedure example for the incubator 11 to determine the state of the rosette 100 . The procedure shown in this figure corresponds to, for example, steps S104 to S109 in correspondence with FIG.

In determining the state of the Rosetta 100, cells at the stage of differentiation to neuroepithelial cells are cultured (step S201). Then, the control device 40 switches the lighting for irradiating the culture container 19 to, for example, lighting suitable for the appearance of rosettes (step S202). Further, the device control unit 49 in the control device 40 performs collection and seeding (step S203).
Then, the image input unit 42 in the control device 40 inputs the captured image of the cells at the stage immediately after seeding (step S204). The Rosetta extracting unit 44 performs image processing for extracting colonies from the omnifocal image (original image) generated by the omnifocal image generating unit 43 using the captured images (a plurality of single-focus images) (step S205). ). Colonies correspond to the earliest state of Rosetta 100 .
Note that in step S205, the rosette extracting unit 44 processes each of a plurality of omnifocal images as time-lapse images. This point also applies to the steps of image processing that will be described later with reference to the same figure.

Next, the Rosetta correspondence determination unit 46 predicts differentiation efficiency based on the quality, size, texture, and other conditions of the extracted colonies (step S206). Specifically, the Rosetta correspondence determination unit 46 predicts, for example, the number of colonies successfully differentiated into Rosetta among the total number of extracted colonies.
After that, the Rosetta correspondence determination unit 46 predicts, for example, the density of the Rosetta 100, the number of days until the differentiation into the Rosetta 100, etc., in addition to the prediction result of the differentiation efficiency described above, and continues the culture based on these prediction results. It is determined whether the data should be deleted or discarded (step S207).
If it is determined that the nerve cells should be discarded, the device control unit 49 controls, for example, the container conveying device 23 so that the neurons cultured so far are discarded (step S208). On the other hand, when it is determined that the culture should be continued, the device control section 49 controls predetermined devices in the upper casing 12 so that the culture is continued (step S209).

When culturing is continued in step S209, the image input unit 42 inputs a captured image of cells appearing just before medium replacement (step S210). The rosette extraction unit 44 executes image processing for extracting rosettes from the omnifocal image generated using the input captured image (step S211).

Next, the device control unit 49 exchanges the culture medium at appropriate timing (step S212). Next, the image input unit 42 inputs the captured image of the cells immediately after the medium exchange (step S213), and the Rosetta extractor 44 extracts the colony from the omnifocal image generated using the input captured image. (step S214).

The cells at the stage of step S213 are further cultured (step S215). Then, the cells at the timing just before the subsequent land exchange are imaged, and the image input unit 42 inputs the imaged image (step S216). The rosette extraction unit 44 executes image processing for extracting the rosette 100 from the omnifocal image generated using the input captured image (step S217).

Then, the rosette correspondence determination unit 46 determines the state of the rosette based on the colony extracted in step S214 and the rosette 100 extracted in step S217 (step S218). Specifically, the rosette correspondence determination unit 46 determines, for example, whether or not the rosette 100 is formed. Also, when the rosette 100 is formed, its quality is determined.
After that, the rosette correspondence determination unit 46 again determines whether to continue the subsequent culture or discard it based on the state of the rosette 100 determined in step S218 (step S219).
If the Rosetta correspondence determination unit 46 determines that the cells should be discarded, it discards the cells that have been determined so far (step S220). If it determines that the culture should be continued, the medium replacement is performed. After that, culture is continued (step S221).

When the medium is exchanged in step S221, the cells immediately after the medium exchange (step S222) are in the process of appearing the rosette 100 . Then, in the process of further culturing the cells (step S223), the image input unit 42 inputs the captured image of the cells just before collection (step S224), and the Rosetta extraction unit 44 receives the input captured image. Image processing for extracting the rosette 100 generated using the image is executed (step S225).

Then, the rosette correspondence determination unit 46 determines the state of the rosette 100 extracted in step S225 (step S226). At this stage, the rosette correspondence determination unit 46 determines the quality of the rosette 100, for example.

Next, the Rosetta correspondence determination unit 46 determines whether the subsequent steps should be continued or discarded based on the determination result of step S226 (step S227).
When determining that the cells should be discarded, the Rosetta correspondence determination unit 46 discards the cells that have been determined so far (step S228).
Further, when it is determined in step S228 that the culture should be continued, the content of the determination is further divided into two.
One determination is, for example, that the Rosetta 100 is sufficiently grown to proceed to a process for differentiation into neural progenitor cells. In this case, the device control unit 49 causes the recovery device 51 to recover cells for the next step of differentiation into neural progenitor cells (step S229).
Another determination content is, for example, that the rosette 100 has not grown sufficiently, so that the process is shifted to a process for further growing the rosette 100 . In this case, the device control unit 49 controls the collection device 51 and the seeding device 52 so that the collection and seeding steps of step S203 are performed.

The flow chart of FIG. 8 shows a more detailed example procedure for determining the state of neurites 300 . The procedure shown in this figure corresponds to, for example, steps S114 to S118 in correspondence with FIG.

In determining the state of the neurites 300, cells that have differentiated to neural progenitor cells are cultured (step S301). Then, the device control unit 49 in the control device 40 causes the recovery device 51 and the seeding device 52 to perform recovery and sowing (step S302), and continues culturing (step S303).

Then, in the process of culturing in step S303, the image input unit 42 inputs a captured image of neurons in which neurites 300 appear, that is, a single focus image as a phase contrast image (step S304). Further, the neurite extracting unit 45 executes image processing for extracting the neurite 300 from the input single focus image (original image) (step S305).

Here, in the cells in which the neurites 300 appear, as described above, dead cells (dead cells) and living cells other than the neurites 300 (live cells) are mixed. there is These dead and live cells are also effective materials for determining the state of differentiation of neurons. Therefore, depending on the image processing in step S305, not only the neurites 300 but also dead cells and live cells are extracted.

Then, based on the neurite 300, the dead cell, and the living cell extracted in step S305, the neurite correspondence determination unit 47 determines, for example, the length of the elongated projection of the neurite 300, the length of the neurite 300, the living cell, and the dead cell. Recognize the respective number of cells, as well as the growth rate of neurites 300 and viable cells. Also, using this recognition result, the growth state, differentiation state, etc. of nerve cells are determined (step S306).

Then, based on these determination results, the neurite correspondence determining unit 47 also determines subsequent steps in the same step S306.
Specifically, the neurite correspondence determining unit 47 can determine that neurons cultured so far should be discarded. When determined in this way, the device control unit 49 controls, for example, the container transport device 23 so as to discard the neurons cultured so far (step S307). Further, when it is determined that the medium should be replaced and then the next step should be performed, the device control unit 49 causes the medium replacement device 53 to perform the medium replacement (step S308), and then proceeds to the next step. (Step S309). Alternatively, when it is determined to proceed to the next step without changing the culture medium, the device control section 49 controls a predetermined device in the upper casing 12, for example, so as to proceed to the next appropriate step.

[Image processing for rosetta extraction]
Next, a specific example of image processing for extracting the rosette 100 will be described with reference to FIGS. 9 to 14. FIG.
As described above, the original image used to extract the rosette 100 is a phase-contrast image obtained by imaging cells through a phase-contrast microscope, and is an omnifocal image that is in focus over the entire screen area. be. Also, an omnifocal image is formed corresponding to, for example, each imaging time at a predetermined time interval. That is, in the present embodiment, a time-lapse image is formed from a plurality of omnifocal images in time series.

FIG. 9 schematically shows a procedure for generating an omnifocal image by time-lapse. For example, at imaging time t0, the device control unit 49 of the control device 40 causes the imaging device 33a to capture N single-focus images Psf-1 to Psf-N captured at different focal positions. As described above, the single-focus images Psf-1 to Psf-N are phase-contrast images of cells observed through a phase-contrast microscope. An image input unit 42 in the control device 40 inputs these single focus images Psf-1 to Psf-N. Then, the omnifocal image generation unit 43 extracts and synthesizes, for example, high-contrast image regions from the single-focus images Psf-1 to Psf-N, thereby generating an omni-focal image Paf-0 corresponding to the imaging time t0. do.

Similarly, at imaging time t1 after a predetermined time has passed since imaging time t0, N single-focus images Psf-1 to Psf-N are similarly captured by the imaging device 33a. Then, the omnifocal image generation unit 43 uses the single-focus images Psf-1 to Psf-N input by the image input unit 42 to generate the omnifocal image Paf-1 corresponding to the imaging time t1 in the same manner as described above. Generate.
Thereafter, similarly, the control device 40 uses the single-focus images Psf-1 to Psf-N captured by the imaging device 33a at predetermined time intervals to generate the omnifocal image Paf-N.

As described above, the incubator 11 of the present embodiment generates an omnifocal image Paf by a phase difference image at predetermined time intervals. A time-lapse image is formed from the omnifocal image Paf generated at regular time intervals in this manner. Then, the rosette extraction unit 44 extracts the rosette 100 from each of the omnifocal images Paf thus generated.

The original image to be processed when the neurite extraction unit 45 extracts the neurite 300 may be a single focus image, but may be an omnifocal image Paf formed as a time-lapse image as shown in FIG. good too. Since the neurite 300 is not as thick as the rosette 100, extraction with sufficient precision is possible even with a single focus image, but the use of the omnifocal image Paf further improves the extraction precision.
However, in the case of the neurite 300, its elongation speed is considerably faster than the growth speed of the rosette 100. FIG. For this reason, the time interval of the time-lapse images is set to about 6 hours when extracting the rosette 100, while about 30 minutes to 1 hour is set to extract the neurite 300. FIG.

Note that the omnifocal image generator 43 may generate the omnifocal image Paf at the timing when the single-focus images Psf-1 to Psf-N corresponding to each imaging time are input. Further, the single-focus images Psf-1 to Psf-N input corresponding to each imaging time are stored in the storage unit 41, and then the single-focus images read out from the storage unit 41 at a predetermined timing. The omnifocal image Paf may be generated using Psf-1 to Psf-N.

Next, the flowchart of FIG. 10 shows an example of the procedure of image processing executed by the rosette extraction unit 44 to extract the rosette 100 . The processing shown in this figure corresponds to, for example, step S105 in FIG. 6 and steps S205, S211, S214, S217 and S225 in FIG. Further, the processing shown in this figure targets the omnifocal image generated at one imaging time in FIG.

First, the rosette extraction unit 44 in the control device 40 inputs an omnifocal image generated for rosette extraction as an original image (step S401). Note that the rosette extracting unit 44 may input the omnifocal image generated by the omnifocal image generating unit 43 and stored in the storage unit 41, for example.

Next, the rosette extraction unit 44 performs soft matching for extracting an image region having features corresponding to the rosette 100 from the original image (step S402). Soft matching is one type of supervised classification.
FIG. 11 shows an example of soft matching results in step S403. FIG. 11 shows the original image. In the original image, which is a phase-contrast image, the rosettes 100 appearing in the cells are observed as dark clustered areas. Focusing on this point, the soft matching in step S403 predetermines the texture for extracting the dark area, and extracts the area corresponding to this texture from the original image. In FIG. 11, the black area indicated by the white arrow is the area extracted by the soft matching in step S403. The region extracted in this way is a region that is highly likely to be the rosette 100 .

In addition, in the original image as the phase contrast image, the rosette 100 is uniformly dark, and the circular structure characteristic of the rosette 100 has a small density difference in the surrounding portion.
Therefore, in parallel with the soft matching in step S403, the rosette extracting unit 44 extracts regions where the density difference is equal to or less than a certain value from the original image as the phase difference image (steps S403 to S407).

For this purpose, the rosette extraction unit 44 first performs erosion processing on the original image (step S403).
FIG. 12(a) shows an example of the original image generated in step S402. FIG. 12B shows an image obtained by subjecting the original image to erosion in step S403. An image subjected to erosion in this way is in a state in which the dark portion of the original image is enlarged.

Next, the rosette extraction unit 44 generates a difference image by subtracting the original image of FIG. 12(a) from the eroded image of FIG. 12(b) (step S404). FIG. 12(c) is the difference image generated by this step S404. Next, the rosette extraction unit 44 dilates the difference image generated in step S404 (step S405). FIG. 12D shows an image that has undergone dilation in step S405. In the image of FIG. 12D generated in this manner, the absolute value of the density difference is enlarged.

Next, the rosette extraction unit 44 binarizes the image after dilation in FIG. 12D using a predetermined threshold value (step S406). The image in FIG. 13(a) shows the result of binarizing the image in FIG. 12(d) in step S406. In this figure, the black part with the lowest luminance is the area (object) extracted as having a density difference below a certain level.

After that, the rosette extraction unit 44 deletes objects whose size is smaller than a certain size among the objects in the binarized image of FIG. 13A in order to further remove noise (step S407). As a specific example, in the binarized image of FIG. 13(a), small-sized objects are present at positions enclosed by dashed lines. Depending on step S407, for example, such objects are deleted. As a result, the image shown in FIG. 13(b) is obtained. In the area enclosed by the dashed line in the image of FIG. 13B, objects with a density below a certain level and a size below a certain level have been deleted and do not exist.

Next, the rosette extraction unit 44 extracts a portion (object product) in which objects are common between the image subjected to the soft matching in step S403 and the image subjected to the processing in steps S403 to S407 (step S408). ).

FIG. 14(a) shows the same image as in FIG. 11 as the image subjected to the soft matching in step S402, and FIG. 14(b) shows the image in FIG. shows the same image as
As the process of step S408, the rosette extraction unit 44 extracts a matching portion between the object shown in the image of FIG. 11 and the object shown in FIG. 13(b). In other words, the rosette extracting unit 44 extracts a region where the coordinates of the portion shown in black in the image of FIG. 11 and the portion shown in black in the image of FIG. 13(b) match. FIG. 14(c) shows the regions extracted in step S409 in black as indicated by white arrows. The area extracted in this way is the rosette 100 .
Then, the rosette extracting unit 44 outputs the image showing the rosette extracted in step S408 (rosetta extracted image) (step S409).

Note that the processing shown in the flowchart of FIG. 10 is performed on an original image as one omnifocal image generated corresponding to one imaging time in FIG. 9 . As can be understood from the description so far, the omnifocal images form a time-lapse image by aggregation for each imaging time.
Along with this, the rosette extraction unit 44 performs the processing described with reference to FIG. 10 for each omnifocal image corresponding to each imaging time. Then, the rosette correspondence determination unit 46 can make a determination based on the state change over time by comparing the state of the rosette 100 extracted from the focused image at each imaging time, for example.

[Image processing for neurite extraction]
Next, a procedure example of image processing executed by the neurite extraction unit 45 to extract the neurite 300 will be described.
As described above, the neurite extraction unit 45 extracts not only the neurites 300 but also dead cells and live cells. Therefore, the characteristics of the neurites 300, dead cells, and living cells observed in the original image, which is a phase-contrast image, will be described with reference to FIG.
Note that the phase-contrast image used for extracting the neurite 300 is not an all-focus image but a single-focus image. Since the rosette 100 has a three-dimensional structure in which cells are gathered, the neurite 300 is not so tall. Therefore, the neurite 300 can be extracted with sufficiently high accuracy without using an all-in-focus image.

FIG. 15(a) shows an example of a monofocal image obtained by imaging neural progenitor cells in which neurites 300 have emerged. FIG. 15(b) shows the neurites 300, the dead cells 240, and the live cells 250 present in the monofocal image of FIG. 15(a). Note that the living cell 250 is a living cell other than the neurite 300 , and specifically, is in a state of being alive as a cell body 210 . Dead cells 240 also include cell bodies 210 and neurites 300 that have died.

A single focus image is a phase-contrast image captured through a phase-contrast microscope. A phase-contrast image is characterized in that, for example, a portion with large undulations is bright and a portion with small undulations is dark.
As understood from FIGS. 15(a) and 15(b), the neurite 300 is observed as a linear dark portion in the phase-contrast image. Therefore, the neurite extracting unit 45 extracts a thin line-shaped portion having a lower brightness than the surroundings as a neurite 300 in the single focus image (first original image).
Also, the dead cells 240 are observed in the phase-contrast image as granular portions such as bright small circles bordered by dark lines. Therefore, the neurite extracting unit 45 extracts, as dead cells 240, granular portions having a brightness of a certain level or more and a brightness gradient of the edge of the granular part of the single focus image.
In the phase-contrast image, the living cell 250 is observed as a relatively large dark mass with a bright border around it. Therefore, the neurite extracting unit 45 extracts, as the living cell 250, a portion of the monofocal image in which the brightness is below a certain level, the area is above a certain level, and the edge brightness is above a certain level.

The flowchart of FIG. 16 shows an example of a processing procedure for the neurite extraction unit 45 to extract neurites 300, dead cells 240, and live cells 250. In FIG.
First, the neurite extracting unit 45 inputs a single focus image captured by the imaging device 33a for extracting the neurite 300 as an original image (step S501). At this time, the neurite extracting unit 45 may input a single-focal image stored in the storage unit 41 captured by the imaging device 33a to extract the neurite 300, for example.

The neurite extraction unit 45 executes soft matching for extracting an image region having features corresponding to dead cells 240 from the input phase-contrast image as the original image (step S502). As the soft matching in step S502, the neurite extracting unit 45 performs a process of extracting, as an object, a portion surrounded by a bright outline, for example.

FIG. 17 shows an example of the processing result of soft matching as step S502. FIG. 17(a) shows an example of the original image input in step S501. By the soft matching in step S502, the neurite extracting unit 45 extracts, as an object, the area surrounded by black in FIG. 17(b) from the original image of FIG. 17(a). Objects extracted in this manner are likely to be dead cells 240 .

Further, the neurite extracting unit 45 performs closing image processing on the original image input in step S501 (step S503), and further binarizes the closing image (step S503) to extract the living cell 250. S504).

FIGS. 18(a) to 18(c) show the results of image processing for closing and binarization in steps S503 and S504.
FIG. 18(a) is an example of the original image input in step S501. The image shown in FIG. 18(b) is obtained by the neurite extraction unit 45 closing the original image shown in FIG. 18(a). This closing process fills in the thin black grooves. As a result, the black clumps corresponding to the live cells 250 are preserved, but for example, the fine black parts corresponding to the neurites 300 and the bright parts around the small dark circles, such as the dead cells 240, are reduced to the surrounding density. filled with
Next, the neurite extraction unit 45 extracts the area of the black portion as an object by binarizing the image of FIG. 18(b) in step S504. The object extracted in this way is the area shown in black in FIG. 18(c). Objects extracted in this manner are likely to be living cells 250 .

In order to extract the neurite 300, the neurite extraction unit 45 also generates a difference image by subtracting the original image input in step S501 from the image closed in step S503 (step S505). Next, the neurite extraction unit 45 binarizes this difference image (step S506).

FIGS. 18A, 18B, 18D, and 18E show the results of image processing for closing and binarization in steps S505 and S506. That is, the neurite extracting unit 45 obtains the difference of the original image shown in FIG. 18(a) with respect to the image shown in FIG. 18(b) closed in step S503 in step S505. is obtained. The differential image generated in this way is, for example, an image from which the dark masses corresponding to the living cells 250 in the original image have been removed. Then, the neurite extracting unit 45 binarizes this differential image in step S506 to obtain the image shown in FIG. 18(e). The white portion in the image shown in FIG. 18( e ) is an object showing a dark and thin portion in the original image as the original image, and is highly likely to be the neurite 300 .

In addition, the neurite extraction unit 45 binarizes the original image input in step S501 when extracting the living cell 250 (step S507). Then, the neurite extracting unit 45 extracts an image portion (product of objects) in which objects are common between the image binarized in step S507 and the image binarized in step S504 (step S508).

FIG. 19 shows an image example corresponding to the image processing in steps S507 and S508. FIG. 19(a) is an example of the original image input in step S501. FIG. 19(b) shows a state in which the binarized image of the original image of FIG. 19(a) in step S507 and the binarized image in step S504 are superimposed.

The portion shown in white in FIG. 19B is the portion classified into white gradation in the image binarized in step S507. Also, in the same FIG. 19B, the black part with the lowest brightness is the object corresponding to the part classified into black gradation in the image binarized in step S504. This object shown in black corresponds to a dark mass area in the original image.

Then, the neurite extracting unit 45 executes the process of step S508, and among the objects in the image binarized in step S504 as shown in FIG. are excluded. FIG. 19(c) shows the processing result of this step S508. Specifically, the four objects extracted in the area enclosed by the dashed lines in FIG. 19(b) are excluded in FIG. 19(c) because their surroundings are not bright in the original image. ing. As a result of the processing in step S508, only dark clusters in the original image with bright borders are extracted as objects. This means that objects corresponding to the living cell 250 have been narrowed down from the binarized image in step S504, for example. In other words, depending on step S508, objects corresponding to living cells 250 are extracted with high accuracy.

Further, the neurite extraction unit 45 extracts an object corresponding to the neurite 300 from the image binarized in step S506 as follows. As described with reference to FIG. 18(e), in the image binarized in step S506, the object in the portion classified into white gradation is a dark and thin portion in the original image, and is therefore likely to be the neurite 300. is high. However, at the stage of the image binarized in step S506, there is still a high possibility that relatively large amounts of noise components such as dead cells 240 and live cells 250 are mixed in the object.

Therefore, the neurite extraction unit 45 generates a difference image by subtracting the image processed by the soft matching in step S502 from the image binarized in step S506 in order to extract an object corresponding to the neurite 300. (step S509).

In the image processed by the soft matching in step S502, image regions having features corresponding to dead cells 240 are extracted as objects, as described above. Therefore, the difference image generated in step S509 is an image having the contents of the object corresponding to the dead cell 240 excluded from the objects of the image binarized in step S506.

Further, the neurite extraction unit 45 further generates a difference image by subtracting the image of the object extracted in step S508 from the difference image generated in step S509 (step S510). The object extracted by step S508 corresponds to the living cell 250 as described above. Therefore, objects corresponding to living cells 250 are further excluded from the difference image generated in step S510. In other words, the difference image generated in step S510 has the content of objects that are highly likely to be dead cells 240 and living cells 250 excluded from objects that are candidates for neurites 300 .

FIG. 20(a) shows an example of the image binarized in step S506. FIG. 20(b) shows an image obtained by performing soft matching in step S502 using the original image from which the image in FIG. 20(a) is generated. FIG. 20(c) is an image showing the object extracted in step S508.
For example, the processing of steps S509 and S510 corresponds to processing of subtracting both the image shown in FIG. 20(b) and the image shown in FIG. 20(c) from the image shown in FIG. 20(a).

Next, the neurite extracting unit 45 performs processing for further narrowing down objects corresponding to the neurite 300 from the difference image generated in step S510 as follows.
As described above, the neurites 300 are observed as thin and dark portions in the original image as the phase-contrast image. likely to be. Therefore, the neurites 300 can be extracted with higher accuracy by excluding objects whose length is less than a certain length among the objects shown in the differential image generated in step S510.

Therefore, in other words, the neurite extraction unit 45 performs thinning processing on the candidate object of the neurite 300 shown in the differential image generated in step S510 (step S511). Note that the thinning processing method is not particularly limited here.

FIG. 21(a) shows an example of the contents of the difference image generated in step S510. In this figure, the black part with the lowest luminance is the object extracted as the neurite 300 candidate.
The image shown in FIG. 21(b) is obtained by the neurite extracting unit 45 thinning the object in the image of FIG. 21(a). As can be seen from FIG. 21(b), the line thinning in step S511 transforms the object, which had a certain thickness in FIG. 21(a), into a linear shape. By linearizing the object in this way, the object can be quantified by its length.

Therefore, the neurite extraction unit 45 selects only objects whose length is equal to or greater than a certain length from the objects thinned in step S511 (step S512). Specifically, the neurite extracting unit 45 deletes, for example, thinned objects whose length is less than a preset threshold, and leaves only those whose length is equal to or greater than the threshold.

Next, the neurite extraction unit 45 restores the original thickness of the object selected in step S512 (step S513).
FIG. 21(c) shows an example image showing the object restored in step S512. 21(a), the object existing in the area enclosed by the dashed line was deleted in step S512 because its length was less than a fixed length. does not exist. In this way, in the image at the stage restored in step S513, noises corresponding to dead cells 240 and live cells 250 are removed, and objects corresponding to neurites 300 are extracted with high precision.
Then, the neurite extracting unit 45 outputs the image representing the object restored in step S513 as the result of extracting the neurite 300 (step S514).

As described above, according to the soft matching in step S502, an image is obtained in which the portion corresponding to the dead cell 240, the portion surrounded by a bright outline, is extracted as an object. After that, the neurite extraction unit 45 executes the following processing to extract dead cells 240 with higher accuracy.

That is, the neurite extraction unit 45 dilates the original image input in step S501 (step S515), and converts the image dilated in step S515 to the original image input in step S501. is subtracted to generate a difference image (step S516).

Next, the neurite extraction unit 45 binarizes the difference image generated in step S516 (step S517). Then, the neurite extracting unit 45 extracts, as an object, a common part between the object in the binarized image in step S517 and the object extracted by the soft matching in step S502 (step S518).

FIG. 22 shows an image example corresponding to the processing in steps S515 and S516. FIG. 22(a) is an example of the original image input at step S501, and FIG. 22(b) is an image obtained by dilating the original image of FIG. 22(a) at step S516. FIG. 22(c) is an image obtained by subtracting the image of FIG. 22(a) from the image of FIG. 22(a) in step S516. In the image shown in FIG. 22(c), a portion with a large density gradient is emphasized.

Also, FIG. 23 shows an image example corresponding to the processing of steps S517 and S518. FIG. 23(a) shows an example in which the differential image shown in FIG. 22(c) is binarized in step S517, and portions classified into white gradation are extracted as objects.
FIG. 23(c) is an image showing an object extracted as a common part between the object shown in FIG. 23(a) and the object shown in FIG. 23(b) in step S508. As can be seen by comparing FIGS. 23(b) and 23(c), at the stage of step S518, objects as dead cells 240 are further narrowed down from the objects extracted by soft matching.

As can be understood from the description so far, the object extracted in step S508 corresponds to the living cell 250. FIG. However, at the stage of step S508, the objects corresponding to the plurality of living cells 250 are extracted as one connected object because they are close to each other even though they are actually separate. may be In this regard, the same can be said for the dead cell 240 object extracted in step S518.

For example, in determining the state of the neurite 300, the number of live cells 250 and dead cells 240 is an effective determination factor. For this reason, it is preferable that the number of objects corresponding to each of live cells 250 and dead cells 240 in the extraction result is as accurate as possible.
Therefore, the neurite extraction unit 45 performs image processing for segmentation on the object of the living cell 250 extracted in step S508 (step S519). In other words, the neurite extracting unit 45 divides an object estimated as a plurality of living cells 250 redundantly extracted as one, corresponding to each of the plurality of living cells 250 . As a result, the objects of the living cell 250 that are in contact with each other and extracted as one object are divided into a plurality of objects. Also, the neurite extracting unit 45 performs image processing for segmentation on the dead cell 240 object extracted in step S518 (step S521). As a result, the objects of the dead cells 240 that are in contact with each other and extracted as one object are divided into a plurality of objects.

Algorithms for segmentation in step S519 or S521 are not particularly limited, but a water boundary separation algorithm is known as one of the effective ones.
FIGS. 24A and 24B schematically show region division by water boundary separation. For example, FIG. 24A shows a state in which two circular objects OBJ1 and OBJ2 are in contact with each other so as to partially overlap each other. In this case, the object OBJ1 and the object OBJ2 are separated as shown in FIG.

In addition, live cells 250 are larger than dead cells 240 and have relatively diverse shapes. On the other hand, dead cells 240 are relatively small and exist in a nearly circular shape on the original image.
Therefore, in steps S519 and S521, the neurite extracting unit 45 performs segmentation suitable for each of the live cells 250 and the dead cells 240 as follows.

Specifically, FIGS. 24(c) and 24(d) show an example of segmentation of the object of the living cell 250 in step S519. FIG. 24(a) is an image before segmentation (that is, an image showing the object extracted in step S508), and FIG. 24(b) is an image after segmentation.

In FIG. 24( a ), the estimated area where two living cell 250 objects may be in contact (watershed) is the area enclosed by dashed lines A and B, for example. Here, if two objects overlap at the portion indicated by the dashed line A, the dividing line (watershed) when dividing these objects will be somewhat long. On the other hand, the length of the watershed between the objects, assuming that the two objects overlap at the portion indicated by the dashed line B, is considerably shorter than the portion indicated by the dashed line A. FIG.

The living cells 250 are characterized by having various shapes, as described above. From this point of view, there is a high possibility that the object including the portion enclosed by the dashed line A is one living cell 250 in reality. On the other hand, there is a high possibility that the portion enclosed by the dashed line B is a portion where two objects that actually exist are overlapped. Therefore, in segmenting the region corresponding to the living cell 250 in step S519, the neurite extracting unit 45 selects only when the length of the watershed (dividing line) when dividing the overlapping portion is equal to or less than a certain value. Try to split. As a result, when the image shown in FIG. 24(c) is subjected to area division, the image shown in FIG. 24(d) is obtained. That is, the area indicated by dashed line A is not divided, and the area indicated by dashed line B is divided.

On the other hand, the dead cells 240 are, as described above, relatively small and nearly circular in shape. Therefore, in segmenting the region corresponding to the living cell 250 in step S519, the neurite extracting unit 45 segments all sites detected as, for example, watersheds regardless of their length.
FIG. 24E is an example of an image showing the object extracted in step S518. In this image, the portion indicated by the dashed line C is presumed to be the watershed. Therefore, as shown in FIG. 24(f), the neurite extracting unit 45 performs region division so as to divide the portion indicated by the dashed line C in step S521.

Then, the neurite extracting unit 45 outputs the extraction result of the object after the division region is performed in step S519 as the extraction result of the living cell 250 (step S520). Further, the neurite extracting unit 45 outputs the extraction result of the object after the division region is performed in step S521 as the extraction result of the dead cell 240 (step S522).

The processing shown in the flowchart of FIG. 16 is also performed on the original image as a single focus image generated corresponding to one imaging time in FIG. Therefore, the neurite extracting unit 45 executes the process of FIG. 16 for each single focus image forming the time-lapse image.

Thus, in this embodiment, the rosette 100 and the neurite 300 are extracted from the image, and the states of the rosette 100 and the neurite 300 are determined based on the extraction result. That is, in this embodiment, the incubator 11 automatically extracts the rosette 100 and the neurite 300 and makes various determinations based on the extraction results. As a result, in the present embodiment, human labor can be reduced.
In addition, in the present embodiment, since the state of Rosetta is determined based on the content of the image of the cell, the differentiated state of nerve cells can be stably evaluated in a non-invasive manner. In the present embodiment, a phase-contrast image is used as an image format that can effectively extract the rosette 100 and the neurite 300 noninvasively.
In addition, in this embodiment, since the cells in the environment being cultured in the incubator 11 are imaged, there is no need to interrupt the culture and remove the cells from the incubator for imaging. This stabilizes the quality of the cells during the culture process without deteriorating.

In this embodiment, the neurites 300 are extracted together with the rosette 100, and the cell state is comprehensively determined based on the state of these structures. You may However, if the result of extracting the neurite 300 is also used, a more reliable determination result can be expected.
Also, the structure to be extracted together with the rosette 100 may be other than the neurite 300, such as the neural tube. Moreover, after extracting the rosette 100 and the neurite 300, other structures may also be extracted, and the state of the cell may be determined based on these extraction results.

Also, a program for realizing the function of each part in FIG. A state determination or the like may be performed. It should be noted that the "computer system" referred to here includes hardware such as an OS and peripheral devices.

The "computer system" also includes the home page providing environment (or display environment) if the WWW system is used.
The term "computer-readable recording medium" refers to portable media such as flexible discs, magneto-optical discs, ROMs and CD-ROMs, and storage devices such as hard discs incorporated in computer systems. In addition, "computer-readable recording medium" means a volatile memory (RAM) inside a computer system that acts as a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. , includes those that hold the program for a certain period of time. Further, the program may be for realizing part of the functions described above, or may be capable of realizing the functions described above in combination with a program already recorded in the computer system.

Although the embodiment of the present invention has been described in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and design and the like are included within the scope of the gist of the present invention.

It should be noted that the inventions described in the above embodiments are arranged and disclosed as appendices.

(Appendix 1)
an image input unit for inputting a first captured image of cells in the process of neuronal differentiation;
an omnifocal image generation unit that generates an omnifocal image that is in focus at a position in the cell thickness direction as a first original image based on at least the first captured image;
A rosette extraction unit for extracting, as a rosette appearing in a differentiation process, a common area between an area having a luminance distribution of a certain value or less in the first original image and an area having a density difference of a certain value or less in the first original image. and,
a rosette correspondence determination unit that determines the state of the extracted rosette;
A cell evaluation device comprising:

(Appendix 2)
The Rosetta correspondence determination unit
Determining a process or future prediction related to differentiation of the cell based on the determined Rosetta state;
The cell evaluation device according to Appendix 1.

(Appendix 3)
The image input unit
inputting a second captured image, which is a single focus image of a cell in the process of differentiation in which a predetermined structure other than the rosette appears,
a structure extraction unit that extracts the predetermined structure from a second original image based on the second captured image;
a structure correspondence determination unit that determines the extracted state of the structure,
The cell evaluation device according to Supplementary Note 1 or Supplementary Note 2.

(Appendix 4)
The structure correspondence determination unit
Determining the differentiation state of the structure or the quality of the structure related to differentiation induction of the cell based on the determined state of the structure;
The cell evaluation device according to appendix 3.

(Appendix 5)
Further comprising a comprehensive determination unit that determines a differentiation state or differentiation efficiency related to differentiation culture of the cell based on the determination result of the Rosetta correspondence determination unit and the determination result of the structure correspondence determination unit;
The cell evaluation device according to appendix 3 or appendix 4.

(Appendix 6)
The image input unit
inputting the first captured image obtained by capturing a phase contrast image of the cell observed with a phase contrast microscope;
The cell evaluation device according to any one of appendices 1 to 5.

(Appendix 7)
The omnifocal image generation unit is
Acquiring the first captured image as a plurality of single-focus images captured by changing the focal position along the depth direction of the cell and changing the imaging position on the predetermined xy plane of the cell, The cell evaluation device according to appendix 6, wherein, by synthesizing each of a plurality of monofocal images, an omnifocal image in which the position of the cell in the thickness direction is in focus is generated as at least the first original image. .

(Appendix 8)
an image input step of inputting a captured image of cells in the process of neuronal differentiation;
an omnifocal image generating step of generating an omnifocal image in focus at a position in the cell thickness direction as a first original image based on at least the captured image;
A rosette extracting step of extracting, as a rosette appearing in the differentiation process, a common area between an area having a luminance distribution of a certain value or less in the first original image and an area having a density difference of a certain value or less in the first original image. and,
a rosette correspondence determination step of determining the state of the extracted rosette;
A cell evaluation method comprising:

(Appendix 9)
to the computer,
an image input step of inputting a captured image of cells in the process of neuronal differentiation;
an omnifocal image generating step of generating an omnifocal image in focus at a position in the cell thickness direction as a first original image based on at least the captured image;
A rosette extracting step of extracting, as a rosette appearing in the differentiation process, a common area between an area having a luminance distribution of a certain value or less in the first original image and an area having a density difference of a certain value or less in the first original image. ,
a rosette correspondence determination step of determining the state of the extracted rosette;
program to run the

REFERENCE SIGNS LIST 11 incubator 12 upper casing 13 lower casing 14 base plate 15 thermostatic chamber 15a temperature control device 15b humidity control device 16 large door 17 middle door 18 small door 19 culture container 21 stocker 22 observation unit 23 container transport device 24 transport stand 31 sample stand 32 stand Arm 33 Body portion 33a Imaging device 33b LED light source 34 Vertical robot 35 Rotation stage 35a Rotation shaft 36 Mini stage 37 Arm unit 40 Control device 41 Storage unit 42 Image input unit 43 Omnifocal image generation unit 44 Rosetta extraction unit 45 Neurite extraction unit 46 Rosetta correspondence determination unit 47 Neurite correspondence determination unit 48 Comprehensive determination unit 49 Device control unit 51 Recovery device 52 Seeding device 53 Medium exchange device 100 Rosetta 200 Nerve cell 210 Cell body 220 Axon 230 Dendrite 240 Dead cell 250 Viable cell 300 neurites

Claims (7)

an image acquisition unit that acquires an image of nerve cells that is transilluminated and non-invasively captured;
An image region having a luminance of a certain value or less or a certain value or more is extracted from the obtained image, and a shape formed by a change in luminance of the image region in the single-focus image is a thin line having a luminance lower than that of the surroundings of the image region. part is extracted as a neurite of the nerve cell, a granular part with a brightness gradient of the edge of the image area in the single-focal image is extracted as a dead cell, and in the single-focal image, the area is above a certain level and the a first extraction unit that extracts , as a living cell, a portion of the edge of the image region where the brightness is equal to or higher than a certain level ;
a first determination unit that determines the growth state of the nerve cell based on the number of the live cells or the dead cells extracted by the first extraction unit;
Cell evaluation device. The image acquisition unit acquires a phase-contrast image of the nerve cell as the image.
The cell evaluation device according to claim 1. The first extraction unit extracts the image region having the brightness information of a certain value or less by closing image processing based on the brightness information of the image, 3. The cell evaluation apparatus according to claim 2, wherein a portion of which the brightness information of the edge is equal to or higher than a certain level and has a bright edge is extracted as a living cell other than the neurite. The first extracting unit extracts the image region having the brightness information of a certain level or more by soft matching image processing based on the brightness information of the image, and extracts the image area having a brightness gradient of a certain level or more at the edge of the image region. The cell evaluation device according to claim 2, wherein the part is extracted as dead cells other than the neurites. Based on the determination result, the first determination unit outputs a control signal to the control unit that transports the culture vessel for the nerve cells when it is determined to discard the nerve cells, or outputs the culture medium for the nerve cells The cell evaluation device according to claim 1, wherein when the exchange is determined, a control signal is output to the control unit that executes the medium exchange. a cell evaluation device according to any one of claims 1 to 5;
A temperature-controlled room for culturing cells,
and an observation unit for capturing said image of cells. to the computer,
an image acquisition step of acquiring an image of the neuron transilluminated and non-invasively imaged;
An image region having a luminance of a certain value or less or a certain value or more is extracted from the obtained image, and a shape formed by a change in luminance of the image region in the single-focus image is a thin line having a luminance lower than that of the surroundings of the image region. part is extracted as a neurite of the nerve cell, a granular part with a brightness gradient of the edge of the image area in the single-focal image is extracted as a dead cell, and in the single-focal image, the area is above a certain level and the a first extraction step of extracting a portion of the edge of the image region where the luminance is above a certain level as a living cell ;
and a first determination step of determining the growth state of the nerve cell based on the number of the live cells or the dead cells extracted by the first extraction step.

Images