A method and system for umbilical cord stem cell quality analysis using image processing

Abstract

The application discloses a kind of umbilical cord stem cell quality analysis method and system using image processing, it is related to cell quality analysis technical field, the method includes: the first microscopic image of the umbilical cord stem cell of cultivation is collected, the stem cell morphology identification is carried out to first microscopic image, obtain multiple stem cell morphology information, quality identification is carried out, and obtain first quality information;Carry out beta-galactosidase dyeing image processing collection, obtain second microscopic image, and obtain multiple blue morphological information and second quality information by identification;The similarity of multiple blue morphological information and multiple stem cell morphology information is analyzed, and first confidence is obtained, and the precipitation crystallization analysis of multiple blue morphological information is carried out, and second confidence is obtained by processing;According to first confidence and second confidence, in combination with the first quality information and second quality information are calculated, and quality information is obtained.The application solves the technical problem that stem cell quality analysis is not accurate in the prior art.

Description

Technical Field

[0001] This invention relates to the field of cell quality analysis technology, and specifically to a method and system for quality analysis of umbilical cord stem cells using image processing. Background Technology

[0002] Umbilical cord stem cells (UC-C) possess significant value in regenerative medicine and disease treatment due to their robust proliferative and differentiation capabilities and immunomodulatory properties. Stem cell quality directly determines their usability, with cell viability, senescence status, and morphological integrity being core evaluation indicators. Current techniques primarily assess stem cell quality through a combination of microscopic imaging and manual observation. Operators must perform morphological interpretation on unstained samples and identify senescent cells within the blue-stained regions of β-galactosidase-stained samples. However, this method has significant drawbacks: First, manual interpretation relies heavily on operator experience, is highly subjective, and time-consuming, making it unsuitable for large-scale cell culture quality monitoring. Second, non-specific precipitates and crystals are easily generated during β-galactosidase staining, their morphology closely resembling the blue-stained regions of senescent cells, leading to a significantly increased misjudgment rate. Furthermore, traditional methods struggle to establish multi-parameter fusion quality assessment models, resulting in inaccurate evaluation results. These shortcomings compromise the reliability of the assessment results, directly impacting the accuracy of stem cell quality assessment. Summary of the Invention

[0003] This application provides a method and system for quality analysis of umbilical cord stem cells using image processing, which addresses the technical problems of poor accuracy and low efficiency in stem cell quality assessment in the prior art.

[0004] In view of the above problems, this application provides a method and system for quality analysis of umbilical cord stem cells using image processing.

[0005] In a first aspect, this application provides a method for quality analysis of umbilical cord stem cells using image processing, the method comprising:

[0006] A first microscopic image of cultured umbilical cord stem cells is collected. Stem cell morphology is identified from the first microscopic image to obtain multiple stem cell morphology information. Quality is then identified to obtain first quality information.

[0007] β-galactosidase staining image processing was performed to acquire a second microscopic image, and multiple blue morphological and second quality information were identified.

[0008] The similarity between the multiple blue morphological information and the multiple stem cell morphological information is analyzed to obtain a first confidence level. The multiple blue morphological information is then subjected to precipitation and crystallization analysis to obtain a second confidence level.

[0009] Based on the first confidence level and the second confidence level, and combined with the first quality information and the second quality information, the quality information is calculated to obtain the quality information.

[0010] Secondly, this application provides an image processing-based umbilical cord stem cell quality analysis system, comprising:

[0011] The morphology recognition module is used to acquire a first microscopic image of cultured umbilical cord stem cells, perform stem cell morphology recognition on the first microscopic image to obtain multiple stem cell morphology information, and perform quality recognition to obtain first quality information.

[0012] The staining recognition module is used to perform β-galactosidase staining image processing and acquisition, obtain a second microscopic image, and identify and obtain multiple blue morphological information and second quality information.

[0013] The confidence analysis module is used to analyze the similarity between the multiple blue morphological information and the multiple stem cell morphological information to obtain a first confidence level, and to perform precipitation and crystallization analysis on the multiple blue morphological information to obtain a second confidence level.

[0014] The result calculation module is used to calculate the quality information based on the first confidence level and the second confidence level, combined with the first quality information and the second quality information.

[0015] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0016] This application proposes an image processing-based method and system for umbilical cord stem cell quality analysis. By integrating morphological feature analysis of unstained samples with a dual-confidence verification mechanism for β-galactosidase-stained samples, the objectivity and accuracy of stem cell quality assessment are significantly improved. Compared with traditional methods, the technical solution provided in this application significantly reduces the subjective bias of manual interpretation and the risk of misjudgment caused by staining interference: First, a deep learning-based cell morphology recognizer can automatically extract key morphological features such as stem cell size and outline, replacing experience-dependent manual observation and standardizing the quality assessment process; Second, by calculating the similarity between the stained area and the original cell morphology through coordinate matching and simultaneously analyzing the precipitation and crystallization interference features of the stained area, the true aging signal and staining artifacts are effectively distinguished, solving the misjudgment problem caused by non-specific crystallization in traditional methods; Finally, by integrating the morphological quality coefficient, stained area quality parameters, and dual-confidence calculations, a multi-dimensional quality assessment model is constructed, overcoming the limitations of single-parameter assessment.

[0017] This application achieves the technical effect of improving the reliability of stem cell quality assessment and analysis while ensuring assessment efficiency, thus providing a more reliable quality basis for stem cell applications. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A schematic flowchart illustrating an image processing-based method for quality analysis of umbilical cord stem cells provided in this application embodiment;

[0020] Figure 2 This is a schematic diagram of an image processing-based umbilical cord stem cell quality analysis system provided in an embodiment of this application.

[0021] The components represented by each number in the attached diagram are explained below:

[0022] The module includes a morphology recognition module 100, a staining recognition module 200, a confidence analysis module 300, and a result calculation module 400. Detailed Implementation

[0023] This application provides a method and system for quality analysis of umbilical cord stem cells using image processing, which addresses the technical problems of poor accuracy and low efficiency in stem cell quality assessment in the prior art.

[0024] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0025] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0026] Example 1, as Figure 1 As shown, this application provides a method for quality analysis of umbilical cord stem cells using image processing, wherein the method includes:

[0027] S10: Collect the first microscopic image of the cultured umbilical cord stem cells, perform stem cell morphology identification on the first microscopic image to obtain multiple stem cell morphology information, perform quality identification to obtain the first quality information.

[0028] In traditional umbilical cord stem cell quality assessment, manual microscopic observation of unstained samples suffers from serious subjectivity and inefficiency. Manual observation relies on experience to visually interpret morphological characteristics such as cell size and contour uniformity, making it difficult to standardize the calculation of morphological quality coefficients for different batches of cells. Furthermore, it cannot quickly process microscopic image data in large-scale culture scenarios, resulting in large fluctuations and low reproducibility of cell quality assessment results.

[0029] Step S10 in the method provided in this application embodiment includes:

[0030] A first microscopic image of cultured umbilical cord stem cells is collected, wherein the first microscopic image includes microscopic images of multiple umbilical cord stem cells;

[0031] The first microscopic image is input into the cell morphology recognizer, and the recognition output obtains multiple stem cell morphology information;

[0032] The training steps for the cell morphology recognizer include:

[0033] Based on historical detection data of umbilical cord stem cells, a set of first microscopic images of the samples was collected. The stem cell morphology information of multiple umbilical cord stem cells in the first microscopic image of each sample was labeled to obtain a set of stem cell morphology information of the sample. The stem cell morphology information includes the size information of the stem cells.

[0034] A cell morphology recognizer was constructed based on a convolutional neural network.

[0035] Using the first set of microscopic images of the sample and the set of stem cell morphology information of the sample, the cell morphology recognizer is iteratively trained until the accuracy meets the preset requirements;

[0036] The multiple stem cell morphology information is input into a cell morphology quality classification table, and multiple cell morphology quality coefficients are obtained by classification. The cell morphology quality classification table is constructed based on the mapping relationship between the sample stem cell morphology information set and the sample cell morphology quality coefficient set.

[0037] The average of the multiple cell morphology quality grades is calculated to obtain the first quality information.

[0038] In this embodiment of the application, a microscope is used to acquire a first microscopic image of cultured umbilical cord stem cells, wherein the first microscopic image includes microscopic images of multiple umbilical cord stem cells.

[0039] Based on historical detection data of umbilical cord stem cells, a set of first microscopic images of the samples was collected. The morphological information of multiple umbilical cord stem cells in the first microscopic image of each sample was labeled. The labeling information included the size information of the stem cells, such as length (in micrometers), width (in micrometers), and aspect ratio, thus obtaining a set of morphological information of the sample stem cells.

[0040] A cell morphology recognizer is constructed based on a convolutional neural network. A four-layer structure is adopted, where the input layer uses one node to receive microscopic images of stem cells, the first convolutional layer uses 32 3×3 convolutional kernels activated by the ReLU function, the second convolutional layer uses 64 3×3 convolutional kernels activated by the ReLU function, and the output layer uses three nodes to output three types of stem cell morphology information. The loss function is the mean squared error.

[0041] The cell morphology recognizer was trained in a supervised manner using the first set of microscopic images of the samples and the set of stem cell morphology information of the samples. The model parameters were iteratively adjusted until the recognition accuracy of the cell recognizer met the preset requirements. The preset requirements were set to the accuracy of stem cell morphology information output by the cell recognizer being above 90%, which means that the training of the cell recognizer was completed.

[0042] Multiple identified stem cell morphology information entries are input into a cell morphology quality classification table. This table categorizes and generates multiple cell morphology quality coefficients. The classification table is constructed based on a mapping relationship between the set of sample stem cell morphology information and the set of sample cell morphology quality coefficients. This pre-built mapping relationship between stem cell morphology information and cell morphology quality coefficients provides a more intuitive reflection of stem cell quality; a higher coefficient indicates better stem cell morphology. For example, cells with an aspect ratio of 1.0-1.2, a length range of 12-18 μm, and a width range of 10-18 μm have a corresponding cell morphology quality coefficient of 0.9 in the classification table. A higher cell morphology quality coefficient indicates better cell quality.

[0043] Calculate the arithmetic mean of multiple cell morphology quality grades as the first quality information.

[0044] This application utilizes a deep learning-driven automated morphology recognition mechanism to achieve rapid analysis and quality coefficient mapping of umbilical cord stem cell microscopic images. The cell morphology recognizer can accurately extract key features such as size and contour, and combined with a pre-constructed morphological quality classification table, outputs standardized quality coefficients, avoiding human experience bias and significantly improving the objectivity and processing efficiency of morphological quality assessment, thus laying a data foundation for subsequent processing.

[0045] S20: Perform β-galactosidase staining image processing and acquisition to obtain a second microscopic image, and identify and obtain multiple blue morphological information and second quality information.

[0046] β-galactosidase staining is a core method for detecting stem cell aging. Currently, the identification of the blue stained area in the stained sample relies entirely on manual observation. Manually observing and calculating the percentage of blue area is labor-intensive, difficult to accurately identify, and yields vague results, leading to a high misjudgment rate of aging status and poor reliability of quality assessment.

[0047] Step S20 in the method provided in this application embodiment includes:

[0048] Cell samples were stained with β-galactosidase and a second microscopic image was acquired.

[0049] The second microscopic image is input into the staining recognition device, and multiple blue morphological information is obtained from the recognition output. The staining recognition device is constructed based on a convolutional neural network and trained using a sample set of second microscopic images and a sample set of blue morphological information.

[0050] Based on the multiple blue morphology information, the second quality information is calculated and obtained;

[0051] The second quality information is calculated based on the multiple blue morphology information, including:

[0052] Based on the multiple blue morphological information, the area ratio of the blue region is calculated by subtracting 1 from the total area to obtain the second quality information.

[0053] Cell samples were stained with β-galactosidase, and a second microscopic image was acquired. Senescent cells could be visually identified by observing the blue tinge of the cells after β-galactosidase staining.

[0054] Based on the historical detection data of stem cells after staining, a second set of microscopic images of the samples was collected. The blue morphological information of multiple umbilical cord stem cells in the second microscopic image of each sample was labeled. The labeled information is the morphological information of the blue area, including shape features and the number of pixels in the blue area, thus obtaining a set of blue morphological information of the sample.

[0055] A staining recognition device is constructed based on a convolutional neural network. A four-layer structure is adopted, where the input layer uses one node to receive the microscopic image of the stained stem cells; the first convolutional layer uses 16 3×3 convolutional kernels activated by the ReLU function; the second convolutional layer uses 32 3×3 convolutional kernels activated by the ReLU function; and the output layer uses two nodes to output two types of blue morphology information. The cross-entropy loss function is used.

[0056] The constructed staining recognizer is trained in a supervised manner using the second set of microscopic images of the samples and the set of blue morphological information of the samples. The model parameters are iteratively adjusted until the accuracy of the blue morphological information output by the staining recognizer reaches more than 90%, which means that the staining recognizer training is complete.

[0057] The second microscopic image is input into the trained staining recognizer, and the recognition output obtains multiple blue morphological information.

[0058] Based on multiple blue morphological information, the second quality information is obtained by subtracting the blue area percentage from 1. The blue area percentage of each cell is calculated as follows: Blue area percentage of a single cell = Number of pixels in the blue area / Total number of pixels in the cell. For example, if the number of pixels in the blue area is 120 and the total number of pixels in the cell is 256, then the cell's blue area percentage = 120 / 256 = 0.47. A larger blue area percentage indicates a greater degree of cell aging and poorer cell quality. The second quality information is calculated as 1 - Cell's blue area percentage. For example, if the cell's blue area percentage is 0.47, then the second quality information = 1 - 0.47 = 0.53. A larger second quality information indicates better cell activity, less aging, and better cell quality.

[0059] This application employs a staining identifier based on a convolutional neural network to automatically extract blue morphological information from stained samples and calculate the area proportion. This method not only avoids errors from manual statistical analysis but also significantly improves the stability and accuracy of aging state assessment by precisely locating the coordinates and morphological features of blue regions.

[0060] S30: Analyze the similarity between the multiple blue morphological information and the multiple stem cell morphological information to obtain a first confidence level, and perform precipitation and crystallization analysis on the multiple blue morphological information to obtain a second confidence level.

[0061] The β-galactosidase staining method has limitations; it easily produces precipitates and crystals that resemble the morphology of senescent cells after staining, affecting the interpretation results. Traditional methods lack quantitative verification mechanisms and cannot identify such interference. When manually comparing cell morphology before and after staining, it is difficult to match corresponding cells and even more difficult to quantitatively analyze the characteristics of precipitates and crystals, leading to false results and misjudging stem cell quality.

[0062] Step S30 in the method provided in this application embodiment includes:

[0063] Obtain multiple second coordinates of the multiple blue morphological information in the second microscopic image, and obtain multiple first coordinates of the multiple stem cell morphological information in the first microscopic image;

[0064] Based on multiple first coordinates and multiple second coordinates, multiple blue morphological information is matched with multiple stem cell morphological information, and the similarity between the matched blue morphological information and the stem cell morphological information is calculated to obtain multiple similarities. The average value is calculated to obtain the first confidence level.

[0065] Precipitation and crystallization analysis was performed on the multiple blue morphological information to obtain a second confidence level;

[0066] Among these methods, precipitation and crystallization analysis is performed on the multiple blue morphological information to obtain a second confidence level, including:

[0067] Obtain information on the morphology of the precipitate crystals;

[0068] Calculate the similarity between the multiple blue morphological information and the precipitate crystal morphological information, and calculate 1 minus the mean to obtain the second confidence level.

[0069] In this embodiment, multiple blue morphological information is obtained at multiple second coordinates within a second microscopic image, where the second coordinates are the centroid coordinates of the blue cell morphology. Multiple stem cell morphological information is also obtained at multiple first coordinates within a first microscopic image, where the first coordinates are the centroid coordinates of the stem cell.

[0070] Based on multiple first coordinates and multiple second coordinates, multiple blue morphological information is matched with multiple stem cell morphological information. The blue morphological information with the closest first and second coordinates is matched with the multiple stem cell morphological information. The similarity between the matched blue morphological information and the stem cell morphological information is calculated to obtain multiple similarity scores. For example, the staining length similarity = 1 - |blue region length - stem cell length| / ((blue region length + stem cell length) / 2). For example, if the blue region length is 10 μm and the stem cell length is 12 μm, then the staining length similarity = 1 - |10 - 12| / ((10 + 12) / 2) = 0.819. Similarly, the staining width similarity and staining aspect ratio similarity between the matched blue morphological information and the stem cell morphological information are calculated.

[0071] The detection of blue staining within the cytoplasm is reliable, but the staining material may precipitate and crystallize outside the cell, failing to reflect cellular aging. Higher similarity indicates a greater likelihood that the blue area is within the cytoplasm, leading to a more accurate reflection of cell quality. The arithmetic mean of staining length similarity, staining width similarity, and staining aspect ratio similarity is calculated as the first confidence level. First confidence level = (staining length similarity + staining width similarity + staining aspect ratio similarity) / 3. For example, if the staining length similarity is 0.819, the staining width similarity is 0.8, and the staining aspect ratio similarity is 0.901, then the first confidence level = (0.819 + 0.8 + 0.901) / 3 = 0.84. A higher first confidence level indicates a more accurate and reliable quality assessment result.

[0072] Microscopic observation is used to obtain information on the morphology of precipitates and crystals. This information includes the shape characteristics of the precipitates and crystals outside the cells after staining, as well as the number of pixels in the precipitate and crystal regions. Precipitates and crystals do not reflect the cell state.

[0073] Calculate the similarity between multiple blue morphological information and precipitate crystal morphological information. For example, the precipitate length similarity = 1 - |blue region length - precipitate crystal length| / ((blue region length + precipitate crystal length) / 2). For instance, if the blue region length is 10 μm and the precipitate crystal length is 4 μm, then the precipitate length similarity = 1 - |10-4| / ((10+4) / 2) = 0.143. Similarly, calculate the precipitate width similarity and precipitate aspect ratio similarity between the blue morphological information and the precipitate morphological information. The higher the similarity, the more extracellular precipitate crystals are present in the blue morphology, which cannot accurately reflect the cell state and is invalid information. Calculate the second confidence level: Second confidence level = 1 - the arithmetic mean of the precipitate length similarity, precipitate width similarity, and precipitate aspect ratio similarity. The second confidence level is calculated as (settling length similarity + settling width similarity + settling aspect ratio similarity) / 3. For example, if the settling length similarity is 0.143, the settling width similarity is 0.2, and the settling aspect ratio similarity is 0.131, then the second confidence level is 1 - (0.143 + 0.2 + 0.131) / 3 = 0.842. A higher second confidence level indicates a more accurate and reliable quality assessment result.

[0074] This application introduces a dual-confidence verification mechanism. It calculates the similarity between the stained area and the original cell morphology through coordinate matching (first confidence level) and analyzes the difference between the blue area and the precipitated crystal morphology (second confidence level), forming a dual-confidence verification mechanism for judging the stained area. This achieves quantitative verification of the authenticity of the staining signal and suppresses the risk of false results affecting cell quality assessment.

[0075] S40: Based on the first confidence level and the second confidence level, and combined with the first quality information and the second quality information, calculate to obtain quality information.

[0076] Current stem cell quality assessment methods often rely on single parameters, such as morphology or staining results alone. However, key indicators such as morphological integrity and aging status are intrinsically related, and relying on a single parameter may lead to inaccurate results. Traditional methods cannot reconcile morphological and staining quality, resulting in a one-sided assessment dimension and inaccurate overall quality interpretation.

[0077] Step S40 in the method provided in this application embodiment includes:

[0078] The confidence level is calculated based on the first confidence level and the second confidence level.

[0079] The quality information is calculated based on the confidence level, the first quality information, and the second quality information.

[0080] In this embodiment of the application, the confidence level is calculated based on the first confidence level and the second confidence level. The confidence level is calculated as (first confidence level + second confidence level) / 2. For example, if the first confidence level is 0.8 and the second confidence level is 0.9, then the confidence level is (0.8 + 0.9) / 2 = 0.85.

[0081] The quality information is calculated based on the confidence level, the first quality information, and the second quality information. Quality information = first quality information + second quality information × confidence level. For example, if the first quality information is 0.9, the second quality information is 0.53, and the confidence level is 0.85, then the quality information = 0.9 + 0.53 × 0.85 = 1.35. The higher the quality information, the better the cell quality.

[0082] This application constructs a multi-dimensional quality assessment by integrating dual confidence levels, morphological quality information, and staining quality information. This enables the final quality information to simultaneously reflect the physiological state of cells and the reliability of detection, and outputs stem cell quality evaluation results with strong anti-interference capabilities and more comprehensive coverage of indicators.

[0083] Example 2, as Figure 2 As shown, based on the same inventive concept as the umbilical cord stem cell quality analysis method using image processing provided in Embodiment 1, this embodiment of the invention also provides an umbilical cord stem cell quality analysis system using image processing, comprising:

[0084] The morphology recognition module 100 is used to acquire a first microscopic image of cultured umbilical cord stem cells, perform stem cell morphology recognition on the first microscopic image to obtain multiple stem cell morphology information, and perform quality recognition to obtain first quality information.

[0085] The staining recognition module 200 is used to perform β-galactosidase staining image processing and acquisition, obtain a second microscopic image, and identify and obtain multiple blue morphological information and second quality information.

[0086] The confidence analysis module 300 is used to analyze the similarity between the multiple blue morphological information and the multiple stem cell morphological information to obtain a first confidence level, and to perform precipitation and crystallization analysis on the multiple blue morphological information to obtain a second confidence level.

[0087] The result calculation module 400 is used to calculate and obtain quality information based on the first confidence level and the second confidence level, combined with the first quality information and the second quality information.

[0088] In one embodiment, the shape recognition module 100 is further configured to:

[0089] A first microscopic image of cultured umbilical cord stem cells is collected, wherein the first microscopic image includes microscopic images of multiple umbilical cord stem cells;

[0090] The first microscopic image is input into the cell morphology recognizer, and the recognition output obtains multiple stem cell morphology information;

[0091] The training steps for the cell morphology recognizer include:

[0092] Based on historical detection data of umbilical cord stem cells, a set of first microscopic images of the samples was collected. The stem cell morphology information of multiple umbilical cord stem cells in the first microscopic image of each sample was labeled to obtain a set of stem cell morphology information of the sample. The stem cell morphology information includes the size information of the stem cells.

[0093] A cell morphology recognizer was constructed based on a convolutional neural network.

[0094] Using the first set of microscopic images of the sample and the set of stem cell morphology information of the sample, the cell morphology recognizer is iteratively trained until the accuracy meets the preset requirements;

[0095] The multiple stem cell morphology information is input into a cell morphology quality classification table, and multiple cell morphology quality coefficients are obtained by classification. The cell morphology quality classification table is constructed based on the mapping relationship between the sample stem cell morphology information set and the sample cell morphology quality coefficient set.

[0096] The average of the multiple cell morphology quality grades is calculated to obtain the first quality information.

[0097] In one embodiment, the staining recognition module 200 is further configured to:

[0098] Cell samples were stained with β-galactosidase and a second microscopic image was acquired.

[0099] The second microscopic image is input into the staining recognition device, and multiple blue morphological information is obtained from the recognition output. The staining recognition device is constructed based on a convolutional neural network and trained using a sample set of second microscopic images and a sample set of blue morphological information.

[0100] Based on the multiple blue morphology information, the second quality information is calculated and obtained;

[0101] The second quality information is calculated based on the multiple blue morphology information, including:

[0102] Based on the multiple blue morphological information, the area ratio of the blue region is calculated by subtracting 1 from the total area to obtain the second quality information.

[0103] In one embodiment, the confidence analysis module 300 is further configured to:

[0104] Obtain multiple second coordinates of the multiple blue morphological information in the second microscopic image, and obtain multiple first coordinates of the multiple stem cell morphological information in the first microscopic image;

[0105] Based on multiple first coordinates and multiple second coordinates, multiple blue morphological information is matched with multiple stem cell morphological information, and the similarity between the matched blue morphological information and the stem cell morphological information is calculated to obtain multiple similarities. The average value is calculated to obtain the first confidence level.

[0106] Precipitation and crystallization analysis was performed on the multiple blue morphological information to obtain a second confidence level;

[0107] Among these methods, precipitation and crystallization analysis is performed on the multiple blue morphological information to obtain a second confidence level, including:

[0108] Obtain information on the morphology of the precipitate crystals;

[0109] Calculate the similarity between the multiple blue morphological information and the precipitate crystal morphological information, and calculate 1 minus the mean to obtain the second confidence level.

[0110] In one embodiment, the result calculation module 400 is further configured to:

[0111] The confidence level is calculated based on the first confidence level and the second confidence level.

[0112] The quality information is calculated based on the confidence level, the first quality information, and the second quality information.

[0113] In summary, the embodiments of this application have at least the following technical effects:

[0114] This application proposes an image processing-based method and system for umbilical cord stem cell quality analysis. By integrating morphological feature analysis of unstained samples with a dual-confidence verification mechanism for β-galactosidase-stained samples, the objectivity and accuracy of stem cell quality assessment are significantly improved. Compared with traditional methods, the technical solution provided in this application significantly reduces the subjective bias of manual interpretation and the risk of misjudgment caused by staining interference: First, a deep learning-based cell morphology recognizer can automatically extract key morphological features such as stem cell size and outline, replacing experience-dependent manual observation and standardizing the quality assessment process; Second, by calculating the similarity between the stained area and the original cell morphology through coordinate matching and simultaneously analyzing the precipitation and crystallization interference features of the stained area, the true aging signal and staining artifacts are effectively distinguished, solving the misjudgment problem caused by non-specific crystallization in traditional methods; Finally, by integrating the morphological quality coefficient, stained area quality parameters, and dual-confidence calculations, a multi-dimensional quality assessment model is constructed, overcoming the limitations of single-parameter assessment.

[0115] This application achieves the technical effect of improving the reliability of stem cell quality assessment and analysis while ensuring assessment efficiency, thus providing a more reliable quality basis for stem cell applications.

[0116] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0117] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0118] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for quality analysis of umbilical cord stem cells using image processing, characterized in that, The method includes: A first microscopic image of cultured umbilical cord stem cells is collected. Stem cell morphology is identified from the first microscopic image to obtain multiple stem cell morphology information. Quality is then identified to obtain first quality information. β-galactosidase staining image processing was performed to acquire a second microscopic image, and multiple blue morphological and second quality information were identified. The similarity between the multiple blue morphological information and the multiple stem cell morphological information is analyzed to obtain a first confidence level. The multiple blue morphological information is then subjected to precipitation and crystallization analysis to obtain a second confidence level. Based on the first confidence level and the second confidence level, and combined with the first quality information and the second quality information, the quality information is calculated to obtain the quality information. A first microscopic image of cultured umbilical cord stem cells was collected. Stem cell morphology was identified from the first microscopic image to obtain multiple stem cell morphological information. Quality was then assessed to obtain first quality information, including: A first microscopic image of cultured umbilical cord stem cells is collected, wherein the first microscopic image includes microscopic images of multiple umbilical cord stem cells; The first microscopic image is input into the cell morphology recognizer, and the recognition output obtains multiple stem cell morphology information; The multiple stem cell morphology information is input into a cell morphology quality classification table, and multiple cell morphology quality coefficients are obtained by classification. The cell morphology quality classification table is constructed based on the mapping relationship between the sample stem cell morphology information set and the sample cell morphology quality coefficient set. Calculate the average of the multiple cell morphology quality grades to obtain the first quality information; Image processing was performed using β-galactosidase staining to acquire a second microscopic image. Multiple blue morphological and secondary quality information was identified, including: Cell samples were stained with β-galactosidase and a second microscopic image was acquired. The second microscopic image is input into the staining recognition device, and multiple blue morphological information is obtained from the recognition output. The staining recognition device is constructed based on a convolutional neural network and trained using a sample set of second microscopic images and a sample set of blue morphological information. Based on the multiple blue morphology information, the second quality information is calculated and obtained; Analyzing the similarity between the multiple blue morphological information and the multiple stem cell morphological information to obtain a first confidence level, and performing precipitation and crystallization rate analysis on the multiple blue morphological information to obtain a second confidence level, including: Obtain multiple second coordinates of the multiple blue morphological information in the second microscopic image, and obtain multiple first coordinates of the multiple stem cell morphological information in the first microscopic image; Based on multiple first coordinates and multiple second coordinates, multiple blue morphological information is matched with multiple stem cell morphological information, and the similarity between the matched blue morphological information and the stem cell morphological information is calculated to obtain multiple similarities. The average value is calculated to obtain the first confidence level. Precipitation and crystallization analysis was performed on the multiple blue morphological information to obtain a second confidence level.

2. The method for quality analysis of umbilical cord stem cells using image processing according to claim 1, characterized in that, The training steps for the cell morphology recognizer include: Based on historical detection data of umbilical cord stem cells, a set of first microscopic images of the samples was collected. The stem cell morphology information of multiple umbilical cord stem cells in the first microscopic image of each sample was labeled to obtain a set of stem cell morphology information of the sample. The stem cell morphology information includes the size information of the stem cells. A cell morphology recognizer was constructed based on a convolutional neural network. Using the first set of microscopic images of the sample and the set of stem cell morphology information of the sample, the cell morphology recognizer is iteratively trained until the accuracy meets the preset requirements.

3. The method for quality analysis of umbilical cord stem cells using image processing according to claim 1, characterized in that, Based on the multiple blue morphological information, a second quality information is calculated, including: Based on the multiple blue morphological information, the area ratio of the blue region is calculated by subtracting 1 from the total area to obtain the second quality information.

4. The method for quality analysis of umbilical cord stem cells using image processing according to claim 1, characterized in that, Precipitation and crystallization analysis was performed on the multiple blue morphological information, and a second confidence level was obtained, including: Obtain information on the morphology of the precipitate crystals; Calculate the similarity between the multiple blue morphological information and the precipitate crystal morphological information, and calculate 1 minus the mean to obtain the second confidence level.

5. The method for quality analysis of umbilical cord stem cells using image processing according to claim 1, characterized in that, Based on the first confidence level and the second confidence level, and combining the first quality information and the second quality information, quality information is calculated to obtain the following: The confidence level is calculated based on the first confidence level and the second confidence level. The quality information is calculated based on the confidence level, the first quality information, and the second quality information.

6. A quality analysis system for umbilical cord stem cells using image processing, characterized in that, The system for implementing the image processing-based umbilical cord stem cell quality analysis method according to any one of claims 1-5 comprises: The morphology recognition module is used to acquire a first microscopic image of cultured umbilical cord stem cells, perform stem cell morphology recognition on the first microscopic image to obtain multiple stem cell morphology information, and perform quality recognition to obtain first quality information. The staining recognition module is used to perform β-galactosidase staining image processing and acquisition, obtain a second microscopic image, and identify and obtain multiple blue morphological information and second quality information. The confidence analysis module is used to analyze the similarity between the multiple blue morphological information and the multiple stem cell morphological information to obtain a first confidence level, and to perform precipitation and crystallization analysis on the multiple blue morphological information to obtain a second confidence level. The result calculation module is used to calculate the quality information based on the first confidence level and the second confidence level, combined with the first quality information and the second quality information.

Images