Information processing device, information processing method, and computer program

Abstract

To evaluate the function of suspension animal cells with high accuracy, quickly, and easily. [Solution] The information processing device comprises a model acquisition unit, a feature acquisition unit, and a prediction execution unit. The model acquisition unit acquires a trained model. The trained model is a model that outputs a functional index value relating to the function of a planktonic animal cell when a feature is input that includes at least one of the following: a luminance index value relating to the luminance value in a cell region on multiple cell images taken at different time points of a stimulated planktonic animal cell, and a texture index value relating to the texture in a cell region. The feature acquisition unit acquires features of the planktonic animal cell to be evaluated. The prediction execution unit predicts the functional index value of the planktonic animal cell to be evaluated by inputting the features of the planktonic animal cell to be evaluated into the trained model, and outputs the prediction result to an output device.

Description

Technical Field

[0001] The technology disclosed in this specification relates to information processing for evaluating the functions of suspension animal cells.

Background Art

[0002] In recent years, a treatment method using CAR-T cells has been established as a next-generation cell therapy that can be expected to have extremely high therapeutic effects. CAR-T cells are T cells that express a chimeric antigen receptor (CAR). In the treatment method using CAR-T cells, for example, a CAR gene is introduced into T cells collected from a patient to obtain CAR-T cells, and after the CAR-T cells are expanded and cultured in vitro, they are returned to the patient's body by intravenous drip. The CAR-T cells returned to the patient's body recognize a cancer antigen that serves as a target for cancer cells and attack it. Hematopoietic suspension cells such as CAR-T cells are suitable for scale-up that does not require adherent culture or cell detachment treatment from the perspective of cell production, and can be used for treatment that is completed by intravenous drip without requiring advanced transplantation techniques from the perspective of treatment. Therefore, it is a highly promising cell therapy.

[0003] CAR cell engineering, which can control various cell fates (proliferation, migration, differentiation, death, etc.) by freely designing CAR, is an important technology for enhancing cell control ability in cell research. In order to produce suspension animal cells such as CAR-T cells more cheaply and freely, a technology for quickly and easily evaluating the functions of these cells is required.

[0004] Conventionally, as a technology for quickly and easily evaluating the functions of adherent cells, a technology for analyzing cell morphology information using images has been proposed (see, for example, Non-Patent Documents 1 and 2).

Prior Art Documents

Non-Patent Documents

[0005]

Non-Patent Document 1

[0006] Unlike adherent cells that adhere to a culture vessel, suspension cells float within the culture vessel, resulting in images of suspension cells that are less clear. Therefore, applying conventional cell morphology analysis techniques used for adherent cells directly to suspension cells does not yield sufficient accuracy. Thus, conventional methods have presented challenges in evaluating the function of suspension animal cells with high accuracy, quickly, and easily.

[0007] This specification discloses a technology capable of solving the above-mentioned problems. [Means for solving the problem]

[0008] The technologies disclosed herein can be implemented, for example, in the following forms:

[0009] (1) The information processing device disclosed herein comprises a model acquisition unit, a feature acquisition unit, and a prediction execution unit. The model acquisition unit acquires a trained model. The trained model is a model that outputs a functional index value relating to the function of a planktonic animal cell when a feature is input that includes at least one of the following: a luminance index value relating to the luminance value in a cell region on a plurality of cell images taken at different timings of a stimulated planktonic animal cell, and a texture index value relating to the texture in the cell region. The feature acquisition unit acquires the feature for the planktonic animal cell to be evaluated. The prediction execution unit predicts the functional index value for the planktonic animal cell to be evaluated by inputting the feature for the planktonic animal cell to be evaluated into the trained model, and outputs the prediction result to an output device. According to this information processing device, the function of a planktonic animal cell can be evaluated with high accuracy, quickly and easily using the trained model.

[0010] (2) In the above-mentioned information processing device, the feature quantities may not include index values ​​calculated based on the contours of the planktonic animal cells. With this configuration, the function of planktonic animal cells can be evaluated with higher accuracy using a trained model.

[0011] (3) In the above-mentioned information processing device, the feature quantities may include the brightness index values. With this configuration, the function of planktonic animal cells can be evaluated with higher accuracy using a trained model.

[0012] (4) In the above-mentioned information processing device, the functional index value may include a response index value that indicates the responsiveness of the planktonic animal cells to the stimulus. With this configuration, the responsiveness of planktonic animal cells to the stimulus can be evaluated quickly and easily with high accuracy using a trained model.

[0013] (5) In the above-mentioned information processing device, the stimulus may include the addition of a ligand. With this configuration, the responsiveness of suspension animal cells to ligand addition can be evaluated quickly and easily with high accuracy using a trained model.

[0014] (6) In the above-mentioned information processing device, the feature quantities may be composed of index values ​​that are inversely correlated with the concentration of the ligand. With this configuration, the function of planktonic animal cells can be evaluated with higher accuracy using a trained model.

[0015] (7) The above-described information processing device may further include an image acquisition unit that acquires a target image which is a cell image of the planktonic animal cell to be evaluated, and the feature acquisition unit may be configured to analyze the target image and acquire the feature quantities of the planktonic animal cell to be evaluated. With this configuration, the function of the planktonic animal cell can be evaluated even more quickly and easily.

[0016] (8) In the above-mentioned information processing device, the suspension animal cells may include CAR-T cells. With this configuration, the function of CAR-T cells can be evaluated quickly and easily with high accuracy using a trained model.

[0017] (9) Other information processing devices disclosed herein include a model acquisition unit, a feature acquisition unit, and a prediction execution unit. The model acquisition unit acquires a trained model. The trained model is a model that outputs a functional index value relating to the function of a planktonic animal cell when a feature is input, which is composed of feature quantities of multiple cell images of planktonic animal cells to which a ligand has been added, taken at different times, and which are index values ​​inversely correlated with the concentration of the ligand. The feature acquisition unit acquires the feature quantities relating to the planktonic animal cell to be evaluated. The prediction execution unit predicts the functional index value relating to the planktonic animal cell to be evaluated by inputting the feature quantities relating to the planktonic animal cell to be evaluated into the trained model, and outputs the prediction result to an output device. According to this information processing device, the function of planktonic animal cells can be evaluated with high accuracy, quickly and easily using the trained model.

[0018] (10) Other information processing devices disclosed herein include a model acquisition unit. The model acquisition unit generates a trained model by performing training. The trained model is a model that, when a feature quantity including a brightness index value relating to the brightness value in a cell region on multiple cell images taken at different time points of a stimulated planktonic animal cell is input, outputs a function index value relating to the function of the planktonic animal cell. According to this information processing device, it is possible to provide a trained model that can evaluate the function of planktonic animal cells quickly and easily with high accuracy.

[0019] The technologies disclosed herein can be implemented in various forms, for example, in the form of an information processing device and method, a cell function evaluation device and method, an information processing system including an information processing device, a computer program for implementing the above method, and a non-temporary recording medium on which the computer program is stored. [Brief explanation of the drawing]

[0020] [Figure 1] Conceptual diagram illustrating the cell response prediction model MO in this embodiment. [Figure 2] Explanatory diagram showing the schematic configuration of the information processing apparatus 100 [Figure 3] Flowchart showing the cell response prediction model acquisition process [Figure 4] Explanatory diagram showing an example of an index value for a cell region obtained from the cell image I0 [Figure 5] Explanatory diagram showing the relationship between each index value for a cell region obtained from the cell image I0 and the depletion time [Figure 6] Explanatory diagram showing the relationship between each index value for a cell region obtained from the cell image I0 and the depletion time [Figure 7] Explanatory diagram showing the relationship between each index value for a cell region obtained from the cell image I0 and the imaging time [Figure 8] Explanatory diagram showing the relationship between each index value for a cell region obtained from the cell image I0 and the imaging time [Figure 9] Flowchart showing the prediction process [Figure 10] Explanatory diagram showing the prediction accuracy using the cell response prediction model MO [Figure 11] Explanatory diagram showing the prediction accuracy using the cell response prediction model MO

Mode for Carrying Out the Invention

[0021] (Regarding the cell response prediction model MO) FIG. 1 is an explanatory diagram conceptually showing the cell response prediction model MO in the present embodiment. The cell response prediction model MO is a learned model used for evaluating the function of floating animal cells Ce such as CAR-T cells. Note that the floating animal cells Ce are merely cells and do not include microorganisms or bacteria.

[0022] Examples of evaluating the function of suspension animal cells Ce include assessing how well the receptors expressed in CAR-T cells function, and evaluating how CAR-T cells are regulated by drugs when used in combination with CAR-T cells. In this embodiment, the cell response prediction model MO is used to predict the response of suspension animal cells Ce when a ligand is added to them.

[0023] The cell response prediction model MO receives feature vector V as input. Feature vector V includes luminance index values ​​related to the brightness values ​​in cell regions on cell image I0, which consists of multiple images of suspension animal cells Ce with ligand added, taken at different time points. As described later, cell image I0 is acquired by the information processing device 100 and subjected to image analysis by the information processing device 100. The information processing device 100 recognizes cell regions in cell image I0 and acquires feature vector V, which includes luminance index values ​​related to the brightness values ​​in those cell regions.

[0024] When the feature quantity V of suspension animal cell Ce is input to the cell response prediction model MO, the cell response prediction model MO outputs a functional index value related to the function of suspension animal cell Ce. The functional index value is, for example, a response index value that shows the responsiveness of suspension animal cell Ce to ligand addition, and includes the degree of cell proliferation and differentiation, and the concentration of ligand in solution. By using such a cell response prediction model MO, the function (performance) of suspension animal cell Ce can be evaluated with high accuracy, simply and quickly.

[0025] (Configuration of the information processing device 100) Next, the configuration of the information processing device 100 for creating a cell response prediction model MO and performing predictions using the cell response prediction model MO will be described. Figure 2 is an explanatory diagram showing the schematic configuration of the information processing device 100. The information processing device 100 is composed of, for example, a computer (PC, server, smartphone, tablet terminal, etc.).

[0026] The information processing device 100 comprises a control unit 110, a storage unit 120, a display unit 130, an operation input unit 140, and an interface unit 150. Each of these units is connected to the others via a bus 190 so as to be able to communicate with each other.

[0027] The display unit 130 of the information processing device 100 is composed of, for example, a liquid crystal display or an organic EL display, and displays various images and information. The display unit 130 is an example of an output device and display device. The operation input unit 140 is composed of, for example, a keyboard, mouse, buttons, microphone, trackpad, etc., and receives operations and instructions from the administrator. The display unit 130 may also function as the operation input unit 140 by being equipped with a touch panel. The interface unit 150 is composed of, for example, a LAN interface or a USB interface, and communicates with other devices by wired or wireless connection. The information processing device 100 may also be equipped with other output devices (for example, a speaker).

[0028] The storage unit 120 of the information processing device 100 is composed of, for example, ROM, RAM, a hard disk drive (HDD), and stores various programs and data, and is used as a workspace and temporary storage area for data when executing various programs. For example, the storage unit 120 stores a prediction processing program CP, which is a computer program for executing various processes described later. The prediction processing program CP is provided, for example, stored on a computer-readable recording medium (not shown) such as a CD-ROM, DVD-ROM, or USB memory, or is provided in a state that can be obtained from an external device (a server on a network or other terminal device) via the interface unit 150, and is stored in the storage unit 120 in a state that can be operated on the information processing device 100. The storage unit 120 of the information processing device 100 stores a cell response prediction model MO in advance, or during the execution of various processes described later.

[0029] The control unit 110 of the information processing device 100 is composed of, for example, a CPU, and controls the operation of the information processing device 100 by executing a computer program read from the storage unit 120. For example, the control unit 110 functions as a prediction processing unit 111 for executing various processes described later by reading and executing a prediction processing program CP from the storage unit 120. The prediction processing unit 111 includes an image acquisition unit 112, a model acquisition unit 113, a feature acquisition unit 114, and a prediction execution unit 115. The functions of each of these units will be explained in accordance with the descriptions of the various processes described later.

[0030] (Process for obtaining cell response prediction models) Next, the cell response prediction model acquisition process performed by the information processing device 100 of this embodiment will be described. Figure 3 is a flowchart of the cell response prediction model acquisition process. The cell response prediction model acquisition process is the process of acquiring a cell response prediction model MO. In this embodiment, the information processing device 100 acquires the cell response prediction model MO by generating the cell response prediction model MO itself through predetermined machine learning. The cell response prediction model acquisition process is started when a user operates the operation input unit 140 of the information processing device 100 and inputs a start command.

[0031] First, the image acquisition unit 112 (Figure 2) of the information processing device 100 acquires cell images I0 (S110). As shown in Figure 1, cell images I0 are multiple images of suspension animal cells Ce to which ligand has been added, taken at different times. The image acquisition unit 112 acquires a large number of cell images I0 under different conditions. These conditions include, for example, the type and concentration of ligand to be added, the depletion (initialization) time before ligand addition, and the timing of the images (date and time of acquisition).

[0032] Next, the feature acquisition unit 114 (Figure 2) of the information processing device 100 acquires feature quantities V of planktonic animal cells Ce by analyzing the cell image I0 (S120). For each cell image I0, the feature acquisition unit 114 recognizes the region representing each cell and acquires feature quantities V including a brightness index value related to the brightness value in that region.

[0033] Figure 4 is an explanatory diagram showing an example of index values ​​for a cell region obtained from a cell image I0. The feature quantity V of the planktonic animal cell Ce used in this embodiment includes the mean value of the luminance value of the cell region (Intensity_mean) and the standard deviation of the luminance value of the cell region (Intensity_sd), which are index values ​​for the cell region. The mean value of the luminance value of the cell region is an index value that indicates the degree of brightness of the cell region, and the standard deviation of the luminance value of the cell region is an index value that indicates the variability of the brightness of the cell region. These index values ​​are luminance index values ​​related to the luminance value of the cell region.

[0034] The feature vector V of the planktonic animal cell Ce may further include the sum of squares of the elements of the gray-level co-occurrence matrix (GLCM) (Energy), the distribution (homogeneity) of the GLCM elements (Homogeneity), and the differences between the components of the GLCM (Contrast). These index values ​​are texture index values ​​that represent the texture, such as the pattern of the cell region.

[0035] Thus, the feature quantity V of the planktonic animal cell Ce used in this embodiment consists of index values ​​belonging to the "A" classification in Figure 4 and does not include index values ​​belonging to the "B" classification. For example, the feature quantity V of the planktonic animal cell Ce does not include the following index values, which are calculated based on the contour of the planktonic animal cell Ce. • Number of pixels in the cell region (Area): Area of ​​the cell region • Number of pixels in the longest direction of the cell region (Length): Size of the cell region in the longest direction. • Number of pixels in the short-side direction of the cell region (Width): Size of the cell region in the short-side direction. • The ratio of the number of pixels in the long side to the number of pixels in the short side of the cell region (Length / width ratio): the narrowness of the cell region. • Number of pixels at the perimeter of the cell region (Perimeter): Perimeter of the cell region • Ratio of the squared number of pixels in the periphery to the number of pixels in the cell region (Compactness): Roundness of the cell region

[0036] Figures 5 and 6 are explanatory diagrams showing the relationship between various index values ​​for cell regions obtained from cell image I0 and depletion time. As shown in Figure 5, suspension animal cells Ce were cultured in three culture media CM1 to CM3, each containing 24 well plates WP, and the relationship between various index values ​​for cell regions obtained from cell image I0 of suspension animal cells Ce and depletion time was investigated. Images were taken in bright-field mode every 5 minutes for 2 hours. Ba / F3 cells were used as the suspension animal cells Ce. Different concentrations of ligand (IL-3) were added to each of the 24 well plates WP. In Figure 5, the ligand concentrations in each well plate WP of culture media CM1 to CM2 are shown in grayscale. For convenience, the ligand concentration for culture media CM3 is not shown in grayscale, but the ligand concentration in each well plate WP of culture media CM3 is the same as the ligand concentration in each well plate WP of culture media CM1 to CM2. The same points apply to Figures 7 and 8, which will be described later.

[0037] Figure 5 shows a heatmap representing the levels of each index value obtained from cell image I0 immediately after ligand addition (0 minutes) for three different culture media CM1-CM3 (depletion times: 6 hours, 6 hours 30 minutes, and 7 hours, respectively). In this heatmap, one well plate WP was selected from each of the three culture media CM1-CM3 in order of decreasing ligand concentration, and the levels of each index value obtained from cell image I0 of the well plate WP are shown from left to right. In this heatmap, the levels of the five index values ​​belonging to the "A" classification (Intensity_mean, Intensity_sd, Energy, Homogeneity, Contrast) change smoothly from the left edge to the right edge of the heatmap. Therefore, these index values ​​can be said to be index values ​​that are not affected by depletion time. On the other hand, the levels of the other index values ​​(the eight index values ​​belonging to the "B" classification) do not change smoothly from the left edge to the right edge of the heatmap. Therefore, these index values ​​can be said to be index values ​​that are affected by depletion time. Referring to Figure 5, the feature quantity V of the suspension animal cell Ce used in this embodiment can be said to be composed of index values ​​that are not affected by depletion time.

[0038] Figure 6 shows bar graphs representing the relationship between ligand concentration and the level of indicator values ​​for two indicator values ​​belonging to category "B" (Area, Length) and two indicator values ​​belonging to category "A" (Intensity_mean, Contrast), based on the heatmap in Figure 5. Referring to Figure 6, the indicator values ​​belonging to category "A" are inversely correlated with ligand concentration, with the level of the indicator value increasing as the ligand concentration decreases. On the other hand, the indicator values ​​belonging to category "B" are inversely correlated with ligand concentration, with the level of the indicator value decreasing as the ligand concentration decreases. Referring to Figure 6, it can be said that the feature quantity V of the suspension animal cell Ce used in this embodiment is composed of indicator values ​​that are inversely correlated with ligand concentration.

[0039] Figures 7 and 8 are explanatory diagrams showing the relationship between each index value for a cell region obtained from cell image I0 and the timing of imaging. Figure 7 shows a heat map representing the levels of each index value obtained from cell image I0 immediately after ligand addition for three culture media CM1-CM3 (all with a depletion time of 6 hours) with different imaging times. In this heat map, similar to Figure 5, one well plate WP is selected from each of the three culture media CM1-CM3 in order of decreasing ligand concentration, and the levels of each index value obtained from cell image I0 of the well plate WP are shown from left to right. In this heat map, the levels of the five index values ​​belonging to the "A" classification described above change smoothly from the left edge to the right edge of the heat map. Therefore, it can be said that these index values ​​are not affected by the timing of imaging. On the other hand, the level changes from the left edge to the right edge of the heat map are not smooth for the other index values ​​(the eight index values ​​belonging to the "B" classification described above). Therefore, it can be said that these index values ​​are affected by the timing of imaging. Referring to Figure 7, it can be said that the feature quantity V of the suspension animal cell Ce used in this embodiment is composed of index values ​​that are not affected by the timing of the image acquisition.

[0040] Figure 8 shows bar graphs illustrating the relationship between ligand concentration and the level of indicator values ​​for two indicator values ​​belonging to category "B" (Area, Length) and two indicator values ​​belonging to category "A" (Intensity_mean, Contrast), based on the heatmap in Figure 7. Referring to Figure 8, the indicator values ​​belonging to category "A" are inversely correlated with ligand concentration, with the indicator value level increasing as the ligand concentration decreases. On the other hand, the indicator values ​​belonging to category "B" are inversely correlated with ligand concentration, with the indicator value level decreasing as the ligand concentration decreases. Referring to Figure 8, it can be said that the feature quantity V of the suspension animal cell Ce used in this embodiment is composed of indicator values ​​that are inversely correlated with ligand concentration.

[0041] Let's return to the explanation of the cell response prediction model acquisition process (Figure 3). Next, the model acquisition unit 113 (Figure 2) of the information processing device 100 creates a cell response prediction model MO by performing machine learning (S130). As described above, the cell response prediction model MO is a trained model that, when feature quantities V of suspension animal cells Ce are input, outputs functional index values ​​related to the function of suspension animal cells Ce. Functional index values ​​are, for example, response index values ​​that show the responsiveness of suspension animal cells Ce to ligand addition, such as the degree of cell proliferation and differentiation, and the concentration of ligand in solution. The model acquisition unit 113 prepares training data to which the feature quantities V acquired in S120 and the functional index values ​​are associated, and creates the cell response prediction model MO by performing machine learning using the training data. Various known machine learning algorithms can be used for machine learning to create the cell response prediction model MO. The created cell response prediction model MO is stored in the storage unit 120 of the information processing device 100. With this, the cell response prediction model acquisition process is completed.

[0042] (Predictive processing) Next, the prediction process performed by the information processing device 100 of this embodiment will be described. Figure 9 is a flowchart of the prediction process. The prediction process is a process that predicts the function of suspension animal cells Ce using the cell response prediction model MO. The prediction process may be performed by the same information processing device 100 in which the cell response prediction model acquisition process described above is performed, or it may be performed by a different information processing device 100. The prediction process is started when a user operates the operation input unit 140 of the information processing device 100 and inputs a start command.

[0043] First, the image acquisition unit 112 (Figure 2) of the information processing device 100 acquires cell images I0 of the suspension animal cells Ce to be evaluated via the interface unit 150 (S310). As described above, cell images I0 are multiple images of suspension animal cells Ce to which ligand has been added, taken at different time points.

[0044] Next, the feature acquisition unit 114 (Figure 2) of the information processing device 100 acquires feature quantities V by analyzing the cell image I0 of the target floating animal cell Ce (S320).

[0045] Next, the prediction execution unit 115 (Figure 2) of the information processing device 100 inputs the feature quantity V of the suspension animal cell Ce to be evaluated into the cell response prediction model MO, thereby predicting functional index values ​​related to the function of the suspension animal cell Ce (S330). For example, the prediction execution unit 115 predicts a response index value that indicates the responsiveness of the suspension animal cell Ce to ligand addition.

[0046] Next, the prediction execution unit 115 outputs the prediction result (S340). For example, the prediction execution unit 115 displays the prediction result on the display unit 130. With this, the prediction process is completed.

[0047] (Examples) An example of the cell response prediction model MO described above will be explained below. Figure 10 is an explanatory diagram showing the prediction accuracy using the cell response prediction model MO. On the right side of Figure 10, as in the embodiment described above, the prediction accuracy of the cell response prediction model MO created by machine learning using the five index values ​​belonging to classification "A" in Figure 4 as feature quantities V is shown. On the left side of Figure 10, as a comparative example, the prediction accuracy of a trained model created by machine learning using the eight index values ​​belonging to classification "B" in Figure 4 as feature quantities is shown. For each of the embodiment and the comparative example, the evaluation was performed using the sum of the index values ​​obtained by taking the mean and standard deviation of the above five or eight index values ​​over seven successive periods. Lasso regression was used as the machine learning algorithm. As shown in Figure 10, the cell response prediction model MO of this embodiment has significantly higher prediction accuracy compared to the trained model of the comparative example. Thus, it can be said that by using the cell response prediction model MO of this embodiment, the function of suspension animal cells Ce can be evaluated quickly and easily with high accuracy.

[0048] Figure 11 is an explanatory diagram illustrating an example of the relationship between cell image features and functional index values ​​of cells. Figure 11 shows the frequency distribution of the feature Contrast in the cell image. The left column of Figure 11 is for samples to which the ligand (IL-3) was added, and the right column is for samples to which the ligand (IL-3) was not added. In each case, the frequency distribution of Contrast is shown at predetermined times (0 minutes, 30 minutes, 120 minutes) after the start of the test (ligand addition timing in the left column). Figure 11 also schematically shows a portion of the cell image at each time point. "All cells" are shown as all cells in the cell image in white, and "specific cells" are shown as specific cells in white, with the remaining cells in gray. The specific cells are cells that have a Contrast value within a predetermined range R1 (in this example, the range of mean ± standard deviation) set in the frequency distribution of the feature Contrast at 0 minutes from the start for the ligand-added sample.

[0049] In the right column, which includes samples without ligand addition, the frequency distribution of the feature contrast remained almost constant from immediately after the start of the test until 120 minutes had passed. There were only a few specific cells in the cell images. On the other hand, in the left column, which includes samples with ligand addition, the distribution of the feature contrast was skewed towards the low value side immediately after the start of the test (immediately after ligand addition), and there were many specific cells in the cell images. Subsequently, as time progressed, the distribution of the feature contrast shifted towards the high value side, and the number of specific cells decreased accordingly.

[0050] These findings suggest that in ligand-added samples, the number of cells with low Contrast values ​​increases in response to ligand addition, and then returns to the pre-ligand addition state over time. Therefore, using the Contrast feature allows for rapid and simple evaluation of ligand detection ability by comparing the presence or absence of ligand addition, and also allows for rapid and simple evaluation of the cell's response to ligand addition from the change in frequency distribution over time. This point applies not only to the Contrast feature but also to the other feature V mentioned above.

[0051] (Effects of this embodiment) As described above, the information processing device 100 of this embodiment comprises a model acquisition unit 113, a feature acquisition unit 114, and a prediction execution unit 115. The model acquisition unit 113 acquires a cell response prediction model MO. The cell response prediction model MO is a trained model that, when a feature quantity V including a brightness index value relating to the brightness value in a cell region on multiple cell images I0 taken at different timings of a stimulated planktonic animal cell Ce is input, outputs a functional index value relating to the function of the planktonic animal cell Ce. The feature acquisition unit 114 acquires the feature quantity V for the planktonic animal cell Ce to be evaluated. The prediction execution unit 115 predicts the functional index value for the planktonic animal cell Ce to be evaluated by inputting the feature quantity V for the planktonic animal cell Ce to the cell response prediction model MO, and outputs the prediction result to the output device. According to this embodiment, the function of planktonic animal cell Ce can be evaluated quickly and easily with high accuracy using the cell response prediction model MO.

[0052] In this embodiment, feature quantity V does not include index values ​​calculated based on the contour of the suspension animal cell Ce. Therefore, the function of the suspension animal cell Ce can be evaluated with higher accuracy using the cell response prediction model MO.

[0053] In this embodiment, feature quantity V may include texture index values ​​related to the texture in the cellular region. By adopting such a configuration, the function of suspension animal cells Ce can be evaluated with higher accuracy using the cell response prediction model MO.

[0054] In this embodiment, the functional index values ​​may include response index values ​​that indicate the responsiveness of suspension animal cells Ce to stimulation. By adopting such a configuration, the responsiveness of suspension animal cells Ce to stimulation can be evaluated quickly, easily, and with high accuracy using the cell response prediction model MO.

[0055] In this embodiment, stimulation of suspension animal cells Ce may include the addition of a ligand. By adopting such a configuration, the responsiveness of suspension animal cells Ce to ligand addition can be evaluated quickly, easily, and with high accuracy using the cell response prediction model MO.

[0056] In this embodiment, feature quantity V may be composed of an index value that is inversely correlated with the ligand concentration. By adopting such a configuration, the function of suspension animal cells Ce can be evaluated with higher accuracy using the cell response prediction model MO.

[0057] The information processing device 100 of this embodiment further includes an image acquisition unit 112 that acquires a target image, which is a cell image I0 of the planktonic animal cell Ce to be evaluated, and a feature acquisition unit 114 analyzes the target image to acquire feature quantities V of the planktonic animal cell Ce to be evaluated. According to the information processing device 100 of this embodiment, the function of the planktonic animal cell Ce can be evaluated more quickly and easily.

[0058] (modified version) The technologies disclosed herein are not limited to the embodiments described above and can be modified in various forms without departing from their essence, for example, the following modifications are possible.

[0059] The configuration of the information processing device 100 in the above embodiment is merely an example and can be modified in various ways. Furthermore, the content of the cell response prediction model acquisition process and prediction process in the above embodiment is merely an example and can be modified in various ways. For example, in the above embodiment, the information processing device 100 acquires the cell response prediction model MO by generating the cell response prediction model MO itself, but the information processing device 100 may acquire the cell response prediction model MO generated by another device.

[0060] In the above embodiment, the explanatory variables used as input to the cell response prediction model MO are merely examples and can be modified in various ways. Furthermore, the form of each explanatory variable can be arbitrarily modified.

[0061] In the above embodiment, the feature quantity V of the planktonic animal cell Ce may not include a luminance index value relating to the brightness value in the cell region, but may include a texture index value representing the texture of the cell region. Even with such a configuration, the function of the planktonic animal cell Ce can be evaluated with high accuracy using the cell response prediction model MO. In other words, the feature quantity V of the planktonic animal cell Ce only needs to include at least one of the luminance index value and the texture index value.

[0062] In the above embodiment, the feature quantity V of the suspension animal cell Ce may be composed of an index value that is inversely correlated with the ligand concentration. The index value that is inversely correlated with the ligand concentration may be the brightness index value or the texture index value mentioned above. The index value that is inversely correlated with the ligand concentration may be an index value other than the brightness index value or the texture index value mentioned above. An example of such an index value is the contrast-signal intensity ratio. Even with such a configuration, the function of the suspension animal cell Ce can be evaluated with high accuracy using the cell response prediction model MO.

[0063] In the above embodiment, at least one of the functional units included in the control unit 110 of the information processing device 100 may be included in another device instead of the control unit 110 of the information processing device 100. The cell response prediction model acquisition process and the prediction process in the above embodiment do not necessarily have to be performed by a single device, but may be performed by separate devices. Each step of the cell response prediction model acquisition process in the above embodiment does not necessarily have to be performed by a single device, but may be performed by different devices. Similarly, each step of the prediction process in the above embodiment does not necessarily have to be performed by a single device, but may be performed by different devices. In the above embodiment, some of the configurations implemented by hardware may be replaced with software, and conversely, some of the configurations implemented by software may be replaced with hardware. [Explanation of symbols]

[0064] 100: Information processing unit 110: Control unit 111: Prediction processing unit 112: Image acquisition unit 113: Model acquisition unit 114: Feature acquisition unit 115: Prediction execution unit 120: Storage unit 130: Display unit 140: Operation input unit 150: Interface unit 190: Bus CM1~CM3: Culture medium CP: Prediction processing program Ce: Suspended animal cells I0: Cell image MO: Cell response prediction model V: Features WP: Well plate

Claims

1. A model acquisition unit acquires a trained model that outputs a functional index value relating to the function of the planktonic animal cells when a feature quantity is input that includes at least one of the following: a luminance index value relating to the luminance value in a cell region on multiple cell images of stimulated planktonic animal cells taken at different time points, and a texture index value relating to the texture in the cell region. A feature acquisition unit that acquires the aforementioned feature quantities for the suspension animal cells to be evaluated, A prediction execution unit that inputs the feature quantities of the planktonic animal cells to be evaluated into the trained model to predict the functional index values ​​of the planktonic animal cells to be evaluated and outputs the prediction results to an output device, An information processing device equipped with the following features.

2. An information processing apparatus according to claim 1, The aforementioned feature quantities do not include index values ​​calculated based on the contours of the planktonic animal cells.

3. An information processing apparatus according to claim 1, The aforementioned feature quantity includes the luminance index value, and is an information processing device.

4. An information processing apparatus according to claim 1 or claim 2, An information processing device comprising the aforementioned functional index value, which includes a response index value indicating the responsiveness of the suspension animal cells to the aforementioned stimulus.

5. An information processing apparatus according to claim 4, The aforementioned stimulus includes the addition of a ligand, and is an information processing device.

6. An information processing device according to claim 5, The aforementioned feature quantity is composed of index values ​​that are inversely correlated with the concentration of the ligand, in an information processing device.

7. An information processing apparatus according to claim 1 or claim 2, further, The system includes an image acquisition unit that acquires a target image, which is a cell image of the floating animal cells to be evaluated. The feature acquisition unit is an information processing device that analyzes the target image and acquires the feature quantities of the floating animal cells to be evaluated.

8. An information processing apparatus according to claim 1 or claim 2, The aforementioned suspension-type animal cells include CAR-T cells, and the information processing device is also used.

9. A model acquisition unit acquires a trained model that outputs a functional index value related to the function of the suspension animal cells when it receives feature quantities from multiple cell images of suspension animal cells to which a ligand has been added, taken at different time points, and which are composed of index values ​​that are inversely correlated with the concentration of the ligand. A feature acquisition unit that acquires the aforementioned feature quantities for the suspension animal cells to be evaluated, A prediction execution unit that inputs the feature quantities of the planktonic animal cells to be evaluated into the trained model to predict the functional index values ​​of the planktonic animal cells to be evaluated and outputs the prediction results to an output device, An information processing device equipped with the following features.

10. An information processing device comprising a model acquisition unit that generates a trained model by performing training, which outputs a functional index value relating to the function of a planktonic animal cell when a feature quantity is input that includes at least one of the following: a luminance index value relating to the luminance value in a cell region on multiple cell images of a planktonic animal cell taken at different time points after the cell cell has been stimulated, and a texture index value relating to the texture in the cell region.

11. The process involves obtaining a trained model that outputs a functional index value relating to the function of a planktonic animal cell when a feature quantity is input that includes at least one of the following: a luminance index value relating to the luminance value in a cell region on multiple cell images of a planktonic animal cell taken at different time points after the cell cell has been stimulated, and a texture index value relating to the texture in the cell region. A step of obtaining the aforementioned feature quantities for the suspension animal cells to be evaluated, The process involves inputting the feature quantities of the planktonic animal cells to be evaluated into the trained model to predict the functional index values ​​of the planktonic animal cells to be evaluated, and outputting the prediction results to an output device. An information processing method comprising:

12. On the computer, A process to obtain a trained model that outputs a function index value relating to the function of the planktonic animal cell when a feature quantity including at least one of the following is input: a brightness index value relating to the brightness value in a cell region on multiple cell images of a planktonic animal cell taken at different time points after the cell cell has been stimulated, and a texture index value relating to the texture in the cell region. A process for obtaining the aforementioned feature quantities for the floating animal cells to be evaluated, The process involves inputting the feature quantities of the planktonic animal cells to be evaluated into the trained model to predict the functional index values ​​of the planktonic animal cells to be evaluated, and outputting the prediction results to an output device. A computer program that executes something.

Images