Method for monitoring living cells
By capturing and analyzing non-fluorescent images of living cells and identifying cell nuclear pixels using computational models, the problem of time consuming evaluating cell signaling pathways in the prior art is solved, and efficient and accurate monitoring of live cell signaling pathways is achieved.
Patent Information
- Application Number
- CN202380035961.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-05-09
- Filing Date
- 2023-04-26
- Publication Date
- 2024-12-03
AI Technical Summary
The prior art requires cell lysis and time-consuming sample preparation and assay workflow when evaluating cellular signaling pathways, making it difficult to efficiently monitor changes in signaling pathways in living cells.
By capturing non-fluorescent images of living cells containing fluorescent protein nuclear translocation reporter genes, using computational models to identify cell nuclear pixels, and calculating the number of nuclear nuclear translocation genes based on pixel intensity, monitoring of signaling pathways in living cells is achieved without the need for separate fluorescent marker.
Simplified image analysis and released fluorescence channels, suitable for tracking the location and dynamics of cell components, improving the efficiency and accuracy of evaluating live cell signaling pathways.
Smart Images

Figure HDA0005098743380000011 
Figure HDA0005098743380000021 
Figure HDA0005098743380000031
Abstract
Description
[0001] CROSS-REFERENCE TO RELATED APPLICATIONS
[0002] This application is an international application claiming priority to U.S. application No. 17 / 740,313 filed on May 9, 2022, the contents of which are incorporated herein by reference. Background Art
[0003] Many signaling pathways promote cell survival and growth, and their dysregulation is associated, for example, with the initiation, progression, and recurrence of cancer. The standard approach to assessing signaling pathways in cells is endpoint assays, which require cell lysis and often involve time-consuming sample preparation and / or assay workflows. Summary of the invention
[0004] A first example includes a method for monitoring one or more living cells, the method comprising: (a) capturing a non-fluorescent image of a sample comprising one or more living cells, wherein the one or more living cells contain a fluorescent protein-based nuclear translocation reporter gene; (b) capturing a fluorescent image of the fluorescent protein-based nuclear translocation reporter gene within the one or more living cells in the sample; (c) identifying nuclear pixels of the non-fluorescent image corresponding to the nuclei of the one or more living cells through a computational model; (d) identifying a first pixel of the fluorescent image corresponding to the nucleus and a second pixel of the fluorescent image not corresponding to the nucleus based on the nuclear pixels; and (e) calculating a metric based on a first intensity of the first pixel and a second intensity of the second pixel, the metric representing a first quantity of the fluorescent protein-based nuclear translocation reporter gene located within the nuclei of the one or more living cells and a second quantity of the fluorescent protein-based nuclear translocation reporter gene not located within the nuclei of the one or more living cells.
[0005] A second example includes a non-transitory computer-readable medium storing instructions that, when executed by a computing device, cause the computing device to perform functions, the functions including: (a) capturing a non-fluorescent image of a sample including one or more living cells, wherein the one or more living cells contain a fluorescent protein-based nuclear translocation reporter gene; (b) capturing a fluorescent image of the fluorescent protein-based nuclear translocation reporter gene within the one or more living cells in the sample; (c) identifying nuclear pixels of the non-fluorescent image corresponding to the nuclei of the one or more living cells through a computational model; (d) identifying a first pixel of the fluorescent image corresponding to the nucleus and a second pixel of the fluorescent image not corresponding to the nucleus based on the nuclear pixels; and (e) calculating a metric based on a first intensity of the first pixel and a second intensity of the second pixel, the metric representing a first number of the fluorescent protein-based nuclear translocation reporter gene located within the nuclei of the one or more living cells and a second number of the fluorescent protein-based nuclear translocation reporter gene not located within the nuclei of the one or more living cells.
[0006] A third example includes a system comprising: an optical microscope; a fluorescence microscope; one or more processors; and a non-transitory computer-readable medium storing instructions, which, when executed by the one or more processors, causes the system to perform functions, the functions comprising: (a) capturing a non-fluorescent image of a sample comprising one or more living cells, wherein the one or more living cells contain a fluorescent protein-based nuclear translocation reporter gene; (b) capturing a fluorescent image of the fluorescent protein-based nuclear translocation reporter gene within the one or more living cells in the sample; (c) identifying nuclear pixels of the non-fluorescent image corresponding to the nuclei of the one or more living cells through a computational model; (d) identifying a first pixel of the fluorescent image corresponding to the nucleus and a second pixel of the fluorescent image not corresponding to the nucleus based on the nuclear pixels; and (e) calculating a metric based on a first intensity of the first pixel and a second intensity of the second pixel, the metric representing a first quantity of the fluorescent protein-based nuclear translocation reporter gene located within the nuclei of the one or more living cells and a second quantity of the fluorescent protein-based nuclear translocation reporter gene not located within the nuclei of the one or more living cells.
[0007] A fourth example includes a method for training a computational model, the method comprising: generating a first label for a first pixel of a fluorescent image of a sample, wherein the first label indicates whether the first pixel represents a cell nucleus within the sample; based on the first label, generating a second label for a second pixel of a first non-fluorescent image of the sample, wherein the second label indicates whether the second pixel represents a cell nucleus within the sample; and training the computational model to use the second label and the first non-fluorescent image to identify pixels of the second non-fluorescent image that represent cell nuclei.
[0008] A fifth example includes a non-transitory computer-readable medium storing instructions that, when executed by a computing device, cause the computing device to perform functions, the functions comprising: generating a first label for a first pixel of a fluorescent image of a sample, wherein the first label indicates whether the first pixel represents a cell nucleus within the sample; generating a second label for a second pixel of a first non-fluorescent image of the sample based on the first label, wherein the second label indicates whether the second pixel represents a cell nucleus within the sample; and training a computing model to identify pixels of the second non-fluorescent image that represent cell nuclei using the second label and the first non-fluorescent image.
[0009] A sixth example includes a system comprising: one or more processors; and a non-transitory computer-readable medium storing instructions, which, when executed by the one or more processors, causes the system to perform the following functions, including: generating a first label for a first pixel of a fluorescent image of a sample, wherein the first label indicates whether the first pixel represents a cell nucleus within the sample; based on the first label, generating a second label for a second pixel of a first non-fluorescent image of the sample, wherein the second label indicates whether the second pixel represents a cell nucleus within the sample; and training a computational model to use the second label and the first non-fluorescent image to identify pixels of the second non-fluorescent image that represent cell nuclei.
[0010] When the terms "substantially", "approximately" or "about" are used herein, it means that the recorded characteristics, parameters or values do not need to be achieved exactly, but deviations or changes (including, for example, tolerances, measurement errors, measurement precision limitations, and other factors known to those skilled in the art) may occur in an amount that does not interfere with the effect that the characteristics are intended to provide. In some examples disclosed herein, "substantially", "approximately" or "about" means within the range of + / - 0-5% of the recorded value.
[0011] By reading the following specific implementation and appropriately referring to the accompanying drawings, those skilled in the art will understand these and other aspects, advantages and alternatives. In addition, it should be understood that the present invention content and other descriptions and drawings provided herein are only intended to be illustrated by example, and therefore, there may be multiple changes. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] Figure 1 is a block diagram of an operating environment according to an example.
[0013] Figure 2 is a block diagram of a computing device according to an example.
[0014] Figure 3 is a fluorescence image according to the example.
[0015] Figure 4 is a binary map according to the example.
[0016] Figure 5 is a binary map according to the example.
[0017] Figure 6 is a non-fluorescent image according to the example.
[0018] Figure 7 is a block diagram of a method for training a computational model according to an example.
[0019] Figure 8 is a schematic diagram of a non-fluorescent image according to an example.
[0020] Fig. 9is a schematic diagram of a fluorescent image according to an example.
[0021] Fig.10 is a block diagram of a method for monitoring cells according to an example.
[0022] Fig.11 Phase images, expected nuclear maps, and predicted nuclear maps are shown according to the examples.
[0023] Fig.12 is a scatter plot of the predicted kernel area versus the target kernel area based on the examples.
[0024] Fig.13 is a scatter plot of the predicted kernel area versus the target kernel area based on the examples.
[0025] Fig.14 is a scatter plot of the predicted kernel ratios versus the target kernel ratios based on the examples.
[0026] Fig.15 is a scatter plot of the predicted kernel ratios versus the target kernel ratios based on the examples.
[0027] Fig.16 is a scatter plot of the predicted kernel ratios versus the target kernel ratios based on the examples.
[0028] Fig.17 is a scatter plot of the predicted kernel ratio and the target kernel ratio based on the example DETAILED DESCRIPTION
[0029] As described above, it is necessary to improve the technology for evaluating signal pathways in living cells. The present disclosure includes a method for monitoring one or more living cells. The method includes capturing a non-fluorescent image (e.g., a bright field image, a dark field image, or a phase contrast image) of a sample including one or more living cells. The one or more living cells contain a nuclear translocation reporter gene (FTR) based on a fluorescent protein (e.g., a cell membrane or cell wall of one or more living cells surrounds the FTR). The nuclear translocation reporter gene based on a fluorescent protein can be any such reporter gene that shuttles in and / or out of the nucleus in response to a stimulus of interest, as described in more detail below. The method also includes capturing a fluorescent image of a nuclear translocation reporter gene based on a fluorescent protein in one or more living cells in the sample. This generally involves irradiating one or more living cells and selectively detecting fluorescence emitted by a nuclear translocation reporter gene based on a protein. In one embodiment, a fluorescent image is captured using the same field of view as a non-fluorescent image. The method also includes identifying nuclear pixels corresponding to the nuclei of one or more living cells of a non-fluorescent image by a computational model. Any suitable computational model can be used, including but not limited to a visual transformer (ViT), a convolutional neural network, or another artificial neural network. The method also includes a first pixel corresponding to the cell nucleus of the fluorescence image and a second pixel not corresponding to the cell nucleus of the fluorescence image based on nuclear pixel identification. The method also includes calculating a metric based on a first intensity of the first pixel and a second intensity of the second pixel, the metric representing a first quantity of nuclear translocation reporter genes based on fluorescent proteins located in the cell nuclei of one or more living cells and a second quantity of nuclear translocation reporter genes based on fluorescent proteins not located in the cell nuclei of one or more living cells. Therefore, there is no need to use a separate fluorescent marker to mark the cell nucleus to identify the area related to the cell nucleus in the fluorescence image. This greatly simplifies image analysis and frees up the fluorescent channel to analyze another aspect of the cell, which is considered suitable for the intended use (e.g., tracking the location and dynamics of proteins, organelles, and other cellular components).
[0030] The computational model is typically trained to identify cell nuclei in a non-fluorescent image and then used to classify pixels of an unlabeled non-fluorescent image. Therefore, the method of training the computational model includes: generating a first label for a first pixel of a fluorescent image of a sample, wherein the first label indicates whether the first pixel represents a cell nucleus within the sample. The first label can be in the form of a binary map and / or can be generated, for example, by thresholding. The method also includes generating a second label for a second pixel of the first non-fluorescent image of the sample based on the first label, wherein the second label indicates whether the second pixel represents a cell nucleus within the sample. In one embodiment, this can be achieved by applying the binary map to the first non-fluorescent image. The method also includes training the computational model to identify pixels of the second non-fluorescent image that represent cell nuclei using the second label and the first non-fluorescent image.
[0031] Figure 1 1 is a block diagram illustrating an exemplary operating environment 100 of the present disclosure, which includes a system 10 and a sample 110. The system 10 includes a computing device 200a and an optical assembly 103, which includes an optical microscope 105 and a fluorescence microscope 107. Also shown are computing device 200 and a network 214, which will be described in more detail below.
[0032] Figure 2 is a block diagram illustrating an exemplary computing device 200 that is configured to interact directly or indirectly with the operating environment 100. Specifically, the computing device 200 can be configured to perform one or more functions, including an image generation function based in part on images obtained by the optical microscope 105 and / or the fluorescent microscope 107. The computing device 200 has a processor 202, as well as a communication interface 204, a data storage 206, an output interface 208, and a display 210, each of which is connected to a communication bus 212. The computing device 200 may also include hardware to enable communication within the computing device 200 and between the computing device 200 and other devices (e.g., not shown). For example, the hardware may include a transmitter, a receiver, and an antenna.
[0033] The communication interface 204 can be a wireless interface and / or one or more wired interfaces that allow both short-range and long-range communications with one or more networks 214 or one or more remote computing devices 216 (e.g., tablet computer 216a, personal computer 216b, laptop computer 216c, and mobile computing device 216d). Such wireless interfaces can provide communications under one or more wireless communication protocols, such as Bluetooth, WiFi (e.g., Institute of Electrical and Electronics Engineers (IEEE) 802.11 protocol), long term evolution (LTE), cellular communications, near field communications (NFC), and / or other wireless communication protocols. Such wired interfaces can include Ethernet interfaces, universal serial bus (USB) interfaces, or similar interfaces to communicate via wires, twisted pairs, coaxial cables, optical links, fiber optic links, or other physical connections to a wired network. Therefore, the communication interface 204 can be configured to receive input data from one or more devices and can also be configured to send output data to other devices.
[0034] The communication interface 204 may also include user input devices such as a keyboard, keypad, touch screen, touchpad, computer mouse, trackball, and / or other similar devices.
[0035] The data storage 206 may include or take the form of one or more computer-readable storage media that can be read or accessed by the processor 202. The computer-readable storage media may include volatile and / or non-volatile storage components, such as optical, magnetic, organic, or other memory or disk storage, which may be fully or partially integrated with the processor 202. The data storage 206 is considered to be a non-transitory computer-readable medium. In some examples, the data storage 206 may be implemented using a single physical device (e.g., one optical, magnetic, organic, or other memory or disk storage unit), while in other examples, the data storage 206 may be implemented using two or more physical devices.
[0036] Thus, the data memory 206 includes executable instructions 218 stored thereon. The instructions 218 include computer executable code. When the processor 202 executes the instructions 218, the processor 202 performs functions such as any of the functions described herein. The data memory 206 also includes a computational model 400 stored thereon.
[0037] The processor 202 may be a general purpose processor or a special purpose processor (e.g., a digital signal processor, an application specific integrated circuit, etc.). The processor 202 may receive input from the communication interface 204 and process the input to generate output that is stored in the data storage 206 and output to the display 210. The processor 202 may be configured to execute executable instructions 218 (e.g., computer readable program instructions) stored in the data storage 206 and executable to provide the computing device 200 functionality described herein.
[0038] Output interface 208 provides information to display 210 or other components. Thus, output interface 208 can be similar to communication interface 204 and can be a wireless interface (eg, a transmitter) or a wired interface. Output interface 208 can send commands to, for example, one or more controllable devices.
[0039] Figure 2 The computing device 200 shown in the figure can also represent a local computing device 200a in the operating environment 100, for example, in communication with the optical microscope 105 and / or the fluorescence microscope 107. The local computing device 200a can perform one or more steps of the method described below, can receive input from the user and / or can send image data and user input to the computing device 200 to perform all or part of the steps of the method. In addition, in an optional exemplary embodiment, The platform can be used to perform the method and includes the combined functionality of the computing device 200 , the optical microscope 105 , and the fluorescence microscope 107 .
[0040] Computational model training
[0041] The method of the present disclosure includes identifying nuclear pixels of a non-fluorescent image corresponding to the nuclei of one or more living cells by a computational model. The computational model is a model trained to identify the nuclei without using a nuclear label.
[0042] Figure 3 Exemplary functionality is shown with respect to training computational model 400 using fluorescent images to identify non-fluorescent image pixels corresponding to cell nuclei. Computational model 400 may be stored on data storage 206 of computing device 200, for example.
[0043] The computing device 200 may generate a first label 402 for a first pixel 404 of a fluorescent image 406 of a sample 408. The first label 402 indicates whether the first pixel 404 represents a cell nucleus within the sample 408. The fluorescent image 406 is typically captured after or while illuminating the sample 408, which causes a fluorescent nuclear marker within the sample 408 to emit light, thereby indicating the location of the cell nucleus within the sample 408. When capturing the fluorescent image 406, light that does not correspond to the emission wavelength range of the fluorescent marker is typically filtered out so that only the fluorescence emitted by the fluorescent marker (e.g., the cell nucleus) is depicted in the fluorescent image 406.
[0044] Figure 3 A fluorescent image 406 is shown showing a sample 408 including a plurality of cells (eg, living cells) and corresponding cell nuclei depicted by first pixels 404 surrounded by first labels 402 . Figure 3 406 is used as an example of many fluorescent images 406 of many samples 408. The fluorescent image 406 also includes some first pixels 404 located outside the first label 402. In this example, the user manually views the many fluorescent images 406 and manually generates the first labels 402 using the user interface (e.g., using a click and drag action to draw rectangles around some of the first pixels 404). In this way, the first labels 402 can be created in the form of metadata indicating the pixel locations corresponding to the nucleus of the cell. It can also be generally inferred that the first pixels 404 not marked by the first label 402 do not correspond to the nucleus of the cell. For example, these pixels can correspond to locations outside the cell or within the cytoplasm of the cell.
[0045] Figure 4 Additional exemplary functionality associated with training computational model 400 is shown. More specifically, Figure 4 The result of a more automated process of labeling the first pixel 404 of the fluorescent image 406 is shown in the form of a binary map 410. The binary map 410 is Figure 3 A compressed form of the fluorescent image 406 shown in FIG. 4 is described below. Figure 4The binary map 410 in FIG. 4 is used as an example of many binary maps 410 corresponding to the many fluorescence images 406 .
[0046] As shown, the computing device 200 generates a first label 402 for a first pixel 404 of a fluorescent image 406 of a sample 408. The first label 402 indicates whether the first pixel 404 represents a cell nucleus in the sample 408. That is, the first label 402 corresponds to the first pixel 404 representing a cell nucleus in the sample 408, while the other first pixels 404 do not represent a cell nucleus in the sample 408. In fact, the first label 402 has the same shape and position in the fluorescent image 406 and the corresponding binary map 410, but Figure 3 and Figure 4 can be used as a suitable example.
[0047] To generate the first label 402 of the binary map 410, the computing device 200 may perform thresholding on the intensity of the first pixel 404 of the fluorescent image 406. Regardless of which color model is used, there is a method for converting the pixel value defined by the color model into a grayscale that represents only the pixel intensity. Thus, performing thresholding includes classifying the first pixel 404 as a nuclear pixel (e.g., indicated by the first label 402 in the binary map 410) and a non-nuclear pixel based on whether the intensity of the first pixel 404 exceeds a threshold intensity. Because a fluorescent nuclear marker has been used to identify the nuclear region of the cell, the nuclear pixels are typically brighter than the non-nuclear pixels. (In Figure 3 and 4 4, for ease of illustration, core pixels are actually darker than non-core pixels. ) For example, pixel intensity can be defined as a scale from 0 to 1, and the threshold can be 0.8. Thus, a first pixel 404 with an intensity greater than 0.8 will be considered a core pixel, while a first pixel 404 with an intensity less than or equal to 0.8 will be considered a non-core pixel. The computing device 200 applies the first label 402 (e.g., only) to locations corresponding to core pixels. Other intensity scales and threshold intensities are also possible.
[0048] In addition, the threshold value can be redefined or adjusted according to manual inspection. For example, humans can view the result of thresholding, and can determine that different threshold strengths are used to perform thresholding and more accurately classify the first pixel 404 as a core pixel and a non-core pixel. In one embodiment, the computing device 200 requests an input indicating a second threshold strength through a user interface. For example, the request can take the form of a displayed user prompt. Next, the computing device 200 receives an input indicating (e.g., new) a second threshold strength, and reclassifies the first pixel 404 as a core pixel and a non-core pixel according to whether the intensity of the first pixel 404 exceeds the second threshold strength. When the user determines that the updated threshold strength has been optimized to accurately classify core pixels and non-core pixels, the user usually decides to end this process.
[0049] Figure 5 An exemplary binary map 410 is shown, and Figure 6 An exemplary non-fluorescent image 412 of sample 408 is shown. Figure 6 4 is an example of many non-fluorescent images 412 corresponding to many samples 408 .
[0050] In this context, computing device 200 generates second label 414 for second pixel 416 of non-fluorescent image 412 of sample 408 based on first label 402 (e.g., first label of binary map 410). Second label 414 indicates whether second pixel 416 represents a cell nucleus within sample 408. In practice, second label 414 will actually have the same shape and position as first label 402, but Figure 6 Thus, generating second label 414 may include applying the position and shape of first label 402 within binary map 410 to non-fluorescent image 412 .
[0051] Next, computing device 200 trains computing model 400 to use second label 414 and non-fluorescent image 412 to identify pixels of the non-fluorescent image that represent cell nuclei.
[0052] That is, the computational model 400 evaluates the second label 414 and the non-fluorescent image 412 (e.g., the second pixel 416) to identify the common attributes of the second pixel 416 marked as corresponding to the cell nucleus. The computing device 200 then uses these attributes to more accurately classify the unlabeled pixels of the non-fluorescent image as nuclear or non-nuclear. High pixel intensity makes it more likely that the computational model 400 will mark the unlabeled pixel as a nuclear. More generally, the computing device 200 will adjust various weighting factors corresponding to the algorithm of the computational model 400 based on the evaluation of the second label 414 and the non-fluorescent image 412 so that the computational model 400 more accurately classifies the unlabeled pixels as nuclear or non-nuclear.
[0053] Figure 7 4 is a block diagram of a method 300 for training a computational model 400. Figure 7 As shown, method 300 includes one or more operations, functions or actions, as shown in blocks 302, 304 and 306. Although these blocks are shown in sequence, these blocks can also be performed in parallel and / or in a different order than described herein. In addition, the various blocks can be combined into fewer blocks, divided into additional blocks and / or blocks can be deleted according to the desired implementation.
[0054] At block 302, method 300 includes generating a first label 402 for a first pixel 404 of a fluorescent image 406 of a sample 408, wherein the first label 402 indicates whether the first pixel 404 represents a cell nucleus within the sample 408. Figure 3 and Figure 4 Block 302 is described.
[0055] At block 304, method 300 includes generating a second label 414 for a second pixel 416 of non-fluorescent image 412 of sample 408 based on first label 402, wherein second label 414 indicates whether second pixel 416 represents a cell nucleus within sample 408. Figure 5 and Figure 6 Block 304 is described.
[0056] At block 306, method 300 includes training computational model 400 to identify pixels of the non-fluorescent image that represent cell nuclei using second label 414 and non-fluorescent image 412. Figure 6 Block 306 is described.
[0057] Using trained computational models
[0058] refer to Figure 8 , optical microscope 105 captures a non-fluorescent image 412 of sample 408 including living cells 502A and living cells 502B. Any suitable living nucleated cells may be used. In one embodiment, one or more cells are adherent cells. In another embodiment, one or more cells may be mammalian cells. In another embodiment, one or more mammalian cells are adherent mammalian cells. One or more cells may all be the same cell type or may include different cell types. This is referred to as step (a) in the claims below. Cell 502A includes a nucleus 504A and a cytoplasm 506A. Cell 502B includes a nucleus 504B and a cytoplasm 506B. Sample 408 also includes a background area 508, which includes any portion of sample 408 that does not belong to cell 502A or cell 502B.
[0059] Cells 502A (e.g., nucleus 504A and cytoplasm 506A) and cells 502B (e.g., nucleus 504B and cytoplasm 506B) of sample 408 include different concentrations of protein-based nuclear translocation reporters that can be used to monitor various cellular processes. As used herein, a "fluorescent protein-based nuclear translocation reporter" is a fusion protein that includes a fluorescent protein and a protein that shuttles in and / or out of the nucleus in response to a stimulus of interest. Any suitable fluorescent protein may be suitable for the intended use, including but not limited to green fluorescent protein, red fluorescent protein, yellow fluorescent protein, blue fluorescent protein, orange fluorescent protein, near-infrared fluorescent protein, and any derivatives thereof. For example, derivatives of green fluorescent protein (GFP) include but are not limited to EGFP, Emerald, SuperfolderGFP, AzamiGreen, mWasabi, TagGFP, TurboGFP, AcGFP, ZsGreen, and T-Sapphire. In an exemplary embodiment, a fluorescent protein-based nuclear translocation reporter (FTR) may include a protein kinase translocation reporter, which includes any reporter gene that moves into or out of the nucleus in response to protein kinase and phosphatase activity in a cell. In one embodiment, the protein kinase translocation reporter may include a human FoxO1 protein fused to a fluorescent protein, including but not limited to TagGFP2. In the embodiment, the reporter gene may be referred to as "KTR", and the monitored protein kinase is Akt because FoxO1 is phosphorylated by active Akt kinase. When Akt is active, KTR phosphorylation enhances nuclear export, and KTR is localized to the cytoplasm. When Akt is inactive, KTR dephosphorylation causes nuclear export to become weaker than nuclear import, and KTR is localized to the nucleus. Therefore, the nuclear: cytoplasmic (or nuclear: whole cell) fluorescence ratio can be used to quantify Akt activity.
[0060] Non-limiting examples of other FTRs include, but are not limited to, phosphatase translocation reporters (responsive to phosphatase activity in the cell), protease translocation reporters (responsive to protease activity in the cell), and analyte-responsive translocation reporters, where the analyte can be any analyte, including hydrogen ions, potassium ions, calcium ions, etc. FTRs are often difficult to discern in non-fluorescent image 412 because the fluorescence of the FTR is often dominated by other captured light.
[0061] FTR is expressed by cells and can be encoded by any vector capable of expressing FTR in cells for use in the methods of the present disclosure. The expression vector can be introduced into the cell using any suitable method. In one embodiment, the cell can be transiently transfected to introduce the expression vector. In other embodiments, the cell can be stably transfected (e.g., by viral infection, such as using a lentivirus) to allow FTR to be stably expressed in the cell.
[0062] Computing device 200 uses computational model 400 to identify nuclear pixels corresponding to cell nucleus 504A or cell nucleus 504B in non-fluorescent image 412. This is referred to as step (c) below. As described above, computational model 400 has been trained to identify nuclear pixels in unlabeled non-fluorescent images.
[0063] refer to Fig. 9 , the fluorescence microscope 107 captures a fluorescence image 406 of the FTR in the sample 408 (e.g., in the cell 502A and the cell 502B). This is referred to as step (b) hereinafter. Since the fluorescence imaging selectively captures the fluorescence emitted by the FTR in the sample 408, the regions of the sample 408 have different brightness levels due to the different concentrations of the FTR in the sample 408. Fig. 9 As shown in the different gray levels in FIG. 5 , the nucleus 504A and the cytoplasm 506B have high concentrations of FTR, the background region 508 has a near zero FTR level, and the cytoplasm 506A and the nucleus 504B have low levels of FTR. In other examples, high FTR intensity may be associated with lighter gray shades, but for ease of illustration, Fig. 9 Map low FTR intensities to lighter shades of grey.
[0064] Computing device 200 uses the nuclear pixels of non-fluorescent image 412 (e.g., pixels corresponding to cell nucleus 504A and cell nucleus 504B) to identify a first pixel in fluorescent image 406 that corresponds to cell nucleus 504A or cell nucleus 504B and a second pixel in fluorescent image 406 that does not correspond to cell nucleus 504A or cell nucleus 504B. This is hereinafter referred to as step (d). Since fluorescent image 406 and non-fluorescent image 412 have the same field of view, the binary map of nuclear pixels of non-fluorescent image 412 can be applied to fluorescent image 406 to identify a first pixel in fluorescent image 406 that corresponds to cell nucleus 504A or cell nucleus 504B.
[0065] Next, the computing device 200 calculates a metric based on a first intensity of a first pixel (e.g., a pixel corresponding to the cell nucleus 504A and / or the cell nucleus 504B in the fluorescent image 406) and a second intensity of a second pixel (a pixel corresponding to the background area 508, the cytoplasm 506A, and / or the cytoplasm 506B in the fluorescent image 406), the metric representing a first number of FTRs located within the cell nucleus 504A and / or the cell nucleus 504B and a second number of FTRs not located within the cell nucleus 504A or the cell nucleus 504B. This is hereinafter referred to as step (e).
[0066] Typically, the metric will be in the form of a ratio, but the metric may also be in the form of a difference. Other examples are possible. When a metric is described as representing a first number of FTRs located within the nucleus and a second number of FTRs not located within the nucleus, this may mean that the ratio of the first number of FTRs located within the nucleus to the second number of FTRs not located within the nucleus can be derived from the metric, even if not directly expressed by the metric.
[0067] Sample 408 can be completely characterized as the union of background area 508, cell 502A and cell 502B. In other examples, sample 408 may also include more cells. Therefore, sample 408 in this example can be expressed in mathematical form as:
[0068] Sample 408 = background 508 + cell nucleus 504A + cytoplasm 506A + cell nucleus 504B +
[0069] Cytoplasm 506B
[0070] Therefore, the metric R can take the following form:
[0071] R=G(F(504A,504B),F(508,506A,506B))
[0072] where F is a function that produces the sum, average (e.g., mean), or median of its independent variables, and G is a function that produces a ratio or difference of its independent variables. Typically, in a given example of a metric R, multiple instances of function F will take the same form (e.g., sum, average, or mean).
[0073] Thus, in this example, metric R can be the sum, mean, or median of the pixel intensities of nucleus 504A and nucleus 504B, divided or subtracted by the sum, mean, or median of the pixel intensities of background 508, cytoplasm 506A, and cytoplasm 506B. Based on these principles, metric R can also take many other forms:
[0074] R=G(F(504A+504B),F(508+506A+506B+504A+504B))
[0075] R=G(F(506A+506B),F(508+506A+506B+504A+504B))
[0076] In some examples, computing device 200 segments the background from the cells in non-fluorescent image 412 of sample 408 and excludes the second pixel that does not belong to the cell from the calculation of the second intensity of the second pixel. Therefore, metric R may also take the following form:
[0077] R=G(F(504A,504B),F(506A,506B))
[0078] R=G(F(504A+504B),F(506A+506B+504A+504B))
[0079] R=G(F(506A+506B),F(506A+506B+504A+504B))
[0080] In some examples, a single cell may be of interest. Thus, computing device 200 may identify (e.g., via a clustering algorithm) a nuclear pixel in non-fluorescent image 412 corresponding to cell nucleus 504A, and identify (e.g., via a clustering algorithm) a first pixel corresponding to cell nucleus 504A and a second pixel within cytoplasm 506A in fluorescent image 406. Thus, in some examples, metric R may only be relevant to a single cell:
[0081] R=G(F(504A),F(508,506A))
[0082] R=G(F(504A),F(508+506A+504A))
[0083] R=G(F(506A),F(508+506A+504A))
[0084] R=G(F(504A),F(506A))
[0085] R=G(F(504A),F(506A+504A))
[0086] R=G(F(506A),F(506A+504A))
[0087] Those skilled in the art will appreciate that the methods of the present disclosure can be used to assess the effect of a test compound on the activity to which FTR responds. Thus, in one embodiment, sample 408 is contacted with a test compound, and computing device 200 performs steps (a)-(e) multiple times (e.g., over a period of time) to determine the effect of the test compound on a first quantity of FTR located within nucleus 504A and / or nucleus 504B, and the effect of a second quantity of FTR not located within nucleus 504A and / or nucleus 504B. Any suitable test compound can be used, including but not limited to small molecules, proteins, peptides, nucleic acids, lipids, carbohydrates, and the like. The effect of a test compound on FTR localization provides a measurement of the effect of a test compound on the activity to which FTR responds.
[0088] In some examples, metric R provides a measure of kinase, phosphatase, or protease activity in cell 502A and / or cell 502B.
[0089] In some examples, metric R provides a measure of the analyte concentration in cell 502A and / or cell 502B.
[0090] Fig.10 is a block diagram of a method 600 for monitoring cells 502A and / or cells 502B. Fig.10 As shown, method 600 includes one or more operations, functions or actions, as shown in blocks 602, 604, 606, 608 and 610. Although these blocks are shown in sequence, these blocks can also be performed in parallel and / or in a different order than described herein. In addition, the various blocks can be combined into fewer blocks, divided into additional blocks and / or deleted according to the desired implementation.
[0091] At block 602, method 600 includes capturing a non-fluorescent image 412 of sample 408 comprising one or more living cells 502A and 502B. One or more living cells 502A and 502B may comprise a fluorescent protein-based nuclear translocation reporter gene. Figure 8 Block 602 is described.
[0092] At block 604, method 600 includes capturing a fluorescent image 406 of a fluorescent protein-based nuclear translocation reporter gene within one or more living cells 502A and 502B in sample 408. Fig. 9 Block 604 is described.
[0093] At block 606, method 600 includes identifying, by computational model 400, nuclear pixels in non-fluorescent image 412 corresponding to nuclei 504A and 504B of one or more living cells 502A and 502B. Figure 8 Block 606 is described.
[0094] At block 608, method 600 includes identifying, based on the nuclear pixels, a first pixel in fluorescent image 406 that corresponds to cell nuclei 504A and cell nuclei 504B and a second pixel in fluorescent image 406 that does not correspond to cell nuclei 504A and cell nuclei 504B. Fig. 9 Block 608 is described.
[0095] At block 610, the method 600 includes calculating a metric based on a first intensity of the first pixel and a second intensity of the second pixel, the metric representing a first quantity of the fluorescent protein-based nuclear translocation reporter gene located within the nuclei 504A and 504B of the one or more living cells 502A and 502B and a second quantity of the fluorescent protein-based nuclear translocation reporter gene not located within the nuclei 504A and 504B of the one or more living cells 502A and 502B. Fig. 9 Block 610 is described.
[0096] Experimental Results
[0097] A U-Net-based CNN was trained to detect nuclei in A549 and SK-MES-1 cells. The input of the model is phase contrast microscopy images and the output is a binary nucleus map (see Fig.11 ). Binary nuclear maps were obtained by pixel-wise thresholding of the fluorescence images of nuclear markers. The model was trained for 30 batches for each of the two cell lines using the Adam optimizer to maximize the Dice coefficient between the predicted and expected fluorescence maps with a learning rate of 10-4. The nuclear segmentation results were evaluated by comparing the image-wise total area of the labeled nuclei of the predicted and target images showing a strong correlation (R2 = 89% for A549 and R2 = 97% for SK-MES-1, see Fig.12 and 13 ).
[0098] Both nucleus segmentation models can slightly overestimate the nuclear area for two main reasons. First, the size of the predicted nuclei is on average slightly larger than the size of the expected nuclei. This issue can be addressed by fine-tuning the threshold selection, model training, and post-processing. Second, the efficiency of the nuclear markers is less than 100%, meaning that not all nuclei are labeled in the expected nuclear map. The CNN learns to predict nuclei based on their appearance in phase contrast images, which often means that more nuclei are predicted than are present in the target. This issue is particularly prominent in the A549 dataset used in this case study and can be mitigated by manual validation of the model predictions by cell biologists.
[0099] The output of each model is then used with the fluorescence image to calculate how much fluorescence is inside the nucleus compared to the total fluorescence. To obtain the predicted nucleus map, the output is converted to binary values (pixels belonging to the nucleus equal 1, pixels outside the nucleus equal 0). This map is multiplied pixel by pixel with the fluorescence image to obtain the fluorescence inside the nucleus. The intensities of the pixels predicted to be inside the nucleus are summed and then divided by the sum of the fluorescence intensities of all pixels to obtain the predicted ratio reading. The same process is performed on the target nucleus map to obtain the target ratio for comparison.
[0100] For both cell types, the correlation between the ratios obtained using applicant's invention and the comparative fluorescence-based ratios was strong (R2 = 97% for A549 and SK-MES-1, see Fig.14 and 15), meaning that KTR ratios can be reliably quantified. Note that the predicted KTR ratios are underestimated in both cell types. This is a direct result of the overestimation of nuclear area shown above and can be addressed in two different ways. First, nuclear size can be categorized as described. Second, using a separate calibration set, a multiplicative factor can be calculated and used to adjust the predicted KTR ratios.
[0101] The KTR ratio is also calculated on a cell-by-cell basis. In addition to nucleus segmentation, an instance segmentation model is trained to segment individual cells. The instance segmentation model is based on CenterMask and is trained to segment individual cells on a diverse dataset of multiple cell types. Each individual cell segmentation is then used to determine the cell area and ratio to calculate the individual cell KTR ratio, such as (fluorescence within its nucleus) / (total fluorescence within the cell). For this part, cells with a nuclear area below a certain threshold are excluded, which is mainly a problem in the target, where some cells have no nuclear markers or are much smaller than expected.
[0102] The single cell KTR ratio was positively correlated (R2 = 55% for A549 and 78% for SK-MES-1, see Fig.16 and 17 ), although the noise is larger than the whole image KTR ratio. Even though the noise level can be improved by fine-tuning the cell segmentation model and the nucleus segmentation model, the results also show that a single fluorescent KTR marker can be used to gain insight into the heterogeneity of KTR ratios without the need for nuclear or membrane markers to assist segmentation.
[0103] For purposes of illustration and description, the description of different advantageous arrangements is not intended to be exhaustive or limited to the examples in the disclosed form. Many modifications and variations will be apparent to those of ordinary skill in the art. Furthermore, different advantageous examples may describe different advantages over other advantageous examples. The examples or selected examples are chosen and described in order to best explain the principles of the examples, the practical application, and to enable others of ordinary skill in the art to understand the disclosure for the various examples with various modifications suitable for the particular use contemplated.
Claims
1. A method for monitoring one or more living cells, the method comprising: (a) capturing a non-fluorescent image of a sample comprising one or more living cells, wherein the one or more living cells comprise a fluorescent protein-based nuclear translocation reporter gene; (b) capturing a fluorescent image of the fluorescent protein-based nuclear translocation reporter gene in the one or more living cells in the sample; (c) identifying, by a computational model, nuclear pixels of the non-fluorescent image corresponding to the nuclei of the one or more living cells; (d) identifying, based on the nuclear pixels, a first pixel of the fluorescent image corresponding to the cell nucleus and a second pixel of the fluorescent image not corresponding to the cell nucleus; as well as (e) calculating a metric based on a first intensity of the first pixel and a second intensity of the second pixel, the metric representing a first quantity of the fluorescent protein-based nuclear translocation reporter gene located within the cell nucleus of the one or more living cells and a second quantity of the fluorescent protein-based nuclear translocation reporter gene not located within the cell nucleus of the one or more living cells. The method of claim 1 , wherein the metric is a ratio of the first quantity to the second quantity.
3. The method according to any one of claims 1 to 2, wherein the computational model is trained using non-fluorescent images having pixels labeled as corresponding to or not corresponding to a cell nucleus. The method of claim 3 , wherein the labels for the non-fluorescent images are generated by applying a thresholding algorithm to the fluorescent images.
5. The method according to any one of claims 1 to 4, wherein identifying the core pixels comprises: A binary map is generated that indicates whether each pixel in the non-fluorescent image represents a cell nucleus.
6. The method according to any one of claims 1 to 5, wherein the computational model is trained using fluorescent images with cell nuclei traced with fluorescent nuclear markers.
7. The method of any one of claims 1 to 6, wherein capturing the non-fluorescent image comprises: Capture brightfield, darkfield, or phase contrast images.
8. The method according to any one of claims 1 to 7, wherein the computational model is a visual transformer (ViT), a trained model, a convolutional model or an artificial neural network.
9. The method according to any one of claims 1 to 8, wherein calculating the metric comprises: The sum of the first intensities is calculated.
10. The method of claim 9, wherein the sum is a first sum and calculating the metric further comprises: calculating a second sum of the second intensities; as well as The first sum is compared to the second sum.
11. The method of claim 10, wherein comparing the first sum to the second sum comprises: A ratio of the first sum to the second sum is calculated.
12. The method of claim 10, wherein comparing the first sum to the second sum comprises: A difference between the first sum and the second sum is calculated.
13. The method according to any one of claims 1 to 12, wherein calculating the metric comprises: An average value of the first intensities is calculated.
14. The method of claim 13, wherein the average is a first average and calculating the metric further comprises: calculating a second average value of the second intensity; as well as The first average is compared to the second average.
15. The method of claim 14, wherein comparing the first average value to the second average value comprises: A ratio of the first average value to the second average value is calculated.
16. The method of claim 14, wherein comparing the first average value to the second average value comprises: A difference between the first average value and the second average value is calculated.
17. The method of any one of claims 1 to 16, wherein the second pixel corresponds to the cytoplasm of the one or more living cells, and wherein the calculating comprises: A metric is calculated based on the first intensity of the first pixel and the second intensity of the second pixel, the metric representing the first quantity of the fluorescent protein-based nuclear translocation reporter gene located in the cell nucleus and the second quantity of the fluorescent protein-based nuclear translocation reporter gene located in the cytoplasm of the one or more living cells.
18. The method according to any one of claims 1 to 16, wherein the calculating comprises: The metric is calculated based on the first intensity of the first pixel and the second intensity of the second pixel, and the metric represents the first quantity of the fluorescent protein-based nuclear translocation reporter gene located in the cell nucleus and the second quantity of the fluorescent protein-based nuclear translocation reporter gene located in the one or more living cells.
19. The method of any one of claims 1 to 16, wherein the second pixel corresponds to the cytoplasm of one or more of the living cells, and wherein the calculating comprises: The metric is calculated based on the first intensity of the first pixel and the second intensity of the second pixel, and the metric represents the second quantity of the fluorescent protein-based nuclear translocation reporter gene located in the cytoplasm and the third quantity of the fluorescent protein-based nuclear translocation reporter gene located in the one or more living cells.
20. The method of any one of claims 1 to 19, wherein the fluorescent protein-based nuclear translocation reporter gene is selected from the group consisting of a protein kinase translocation reporter gene, a phosphatase translocation reporter gene, a protease translocation reporter gene, and an analyte-responsive translocation reporter gene.
21. The method according to any one of claims 1 to 20, further comprising: Background and cells are segmented in the non-fluorescent image of the sample, and the second pixels that do not belong to cells are excluded from the calculation of the second intensity of the second pixels.
22. The method of any one of claims 1 to 21, wherein the one or more living cells are adherent mammalian cells.
23. The method according to any one of claims 1 to 22, wherein the method is performed to monitor a signaling pathway within the one or more living cells.
24. The method according to any one of claims 1 to 23, wherein the method further comprises: The sample is contacted with a test compound, and steps (a)-(e) are performed multiple times to determine the effect of the test compound on the first quantity of the fluorescent protein-based nuclear translocation reporter gene located in the cell nuclei of the one or more living cells, and the effect of the test compound on the second quantity of the fluorescent protein-based nuclear translocation reporter gene not located in the cell nuclei of the one or more living cells.
25. The method of any one of claims 1 to 24, wherein the metric provides a measure of kinase activity, phosphatase activity, or protease activity in the one or more living cells.
26. The method of any one of claims 1 to 25, wherein the metric provides a measure of the analyte concentration in the one or more living cells.
27. The method according to any one of claims 1 to 26, wherein: Identifying the nuclear pixel comprises: identifying the nuclear pixel of the non-fluorescent image corresponding to a single cell nucleus of a single cell of the one or more living cells, Identifying the first pixel and the second pixel includes: identifying the first pixel corresponding to the single cell nucleus and the second pixel located in the cytoplasm of the single cell, and Calculating the metric includes calculating the metric representing the first quantity of the fluorescent protein-based nuclear translocation reporter gene located in the cell nucleus and the second quantity of the fluorescent protein-based nuclear translocation reporter gene located in the cytoplasm of the single cell.
28. The method according to any one of claims 1 to 26, wherein: Identifying the nuclear pixel comprises: identifying the nuclear pixel of the non-fluorescent image corresponding to a single cell nucleus of a single cell of the one or more living cells, Identifying the first pixel and the second pixel includes: identifying the first pixel corresponding to the cell nucleus and the second pixel located in the cytoplasm of the single cell, and Calculating the metric includes calculating the metric representing the first quantity of the fluorescent protein-based nuclear translocation reporter gene located in the single cell nucleus and a third quantity of the fluorescent protein-based nuclear translocation reporter gene located in the single cell.
29. The method according to any one of claims 1 to 26, wherein: Identifying the nuclear pixel comprises: identifying the nuclear pixel of the non-fluorescent image corresponding to a single cell nucleus of a single cell of the one or more living cells, Identifying the first pixel and the second pixel includes: identifying the first pixel corresponding to the cell nucleus and the second pixel located in the cytoplasm of the single cell, and Calculating the metric includes calculating the metric representing the second quantity of the fluorescent protein-based nuclear translocation reporter gene located in the cytoplasm of the single cell and a third quantity of the fluorescent protein-based nuclear translocation reporter gene located in the single cell.
30. A non-transitory computer-readable medium storing instructions which, when executed by a computing device, cause the computing device to perform the method of any one of claims 1 to 29.
31. A system for monitoring one or more living cells, the system comprising: Optical microscope; Fluorescence microscopy; one or more processors; as well as A non-transitory computer readable medium storing instructions which, when executed by the one or more processors, cause the system to perform the method according to any one of claims 1 to 29.
32. A system for monitoring one or more living cells, the system comprising: Optical microscope; Fluorescence microscopy; one or more processors; as well as A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause the system to perform functions comprising: (a) capturing a non-fluorescent image of a sample comprising one or more living cells by the optical microscope, wherein the one or more living cells comprise a fluorescent protein-based nuclear translocation reporter gene; (b) capturing a fluorescent image of the fluorescent protein-based nuclear translocation reporter gene in the one or more living cells in the sample through the fluorescent microscope; (c) identifying, by a computational model, nuclear pixels of the non-fluorescent image corresponding to the nuclei of the one or more living cells; (d) identifying, based on the nuclear pixels, a first pixel of the fluorescent image corresponding to the cell nucleus and a second pixel of the fluorescent image not corresponding to the cell nucleus; and (e) calculating a metric based on a first intensity of the first pixel and a second intensity of the second pixel, the metric representing a first quantity of the fluorescent protein-based nuclear translocation reporter gene located within the cell nucleus of the one or more living cells and a second quantity of the fluorescent protein-based nuclear translocation reporter gene not located within the cell nucleus of the one or more living cells.
33. A method for training a computational model to identify pixels of a non-fluorescent image representing a cell nucleus, the method comprising: generating a first label for a first pixel of a fluorescent image of a sample, wherein the first label indicates whether the first pixel represents a cell nucleus within the sample; generating a second label for a second pixel of the first non-fluorescent image of the sample based on the first label, wherein the second label indicates whether the second pixel represents a cell nucleus within the sample; as well as The computational model is trained to identify pixels of the second non-fluorescent image that represent cell nuclei using the second label and the first non-fluorescent image.
34. The method of claim 33, wherein generating the first tag comprises: A threshold process is performed on the intensity of the first pixel.
35. The method of claim 34, wherein performing the thresholding comprises: The first pixel is classified into a core pixel and a non-core pixel according to whether the intensity of the first pixel exceeds a threshold intensity.
36. The method of claim 35, further comprising: receiving, via the user interface, input indicating a second threshold intensity; as well as The first pixel is reclassified into the core pixel and the non-core pixel according to whether the intensity of the first pixel exceeds the second threshold intensity.
37. The method of any one of claims 33 to 36, wherein generating the first tag comprises: A binary map is generated that indicates whether each of the first pixels represents a cell nucleus of the sample.
38. The method of claim 37, wherein generating the second tag comprises: The binary map is applied to the first non-fluorescent image.
39. The method of any one of claims 33 to 38, wherein training the computational model comprises: The errors between pixels of the second non-fluorescent image identified by the computational model and pixels identified by the second label are minimized.
40. A non-transitory computer-readable medium storing instructions which, when executed by a computing device, cause the computing device to perform the method of any one of claims 33 to 39.
41. A system for training a computational model to identify pixels of a non-fluorescent image representing a cell nucleus, the system comprising: one or more processors; as well as A non-transitory computer readable medium storing instructions which, when executed by the one or more processors, cause the system to perform the method of any one of claims 33 to 39.
42. The system of claim 41, further comprising: Optical microscope; and Fluorescence microscopy.
Citation Information
Patent Citations
Method for monitoring live cells
US20230360200A1
