A stem cell quality evaluation method and system based on image recognition

Abstract

The present invention relates to the field of image processing technology, and in particular to a method and system for evaluating stem cell quality based on image recognition. The method comprises: collecting a stem cell sample and an original image of the sample, determining a first quality evaluation index of the stem cell sample according to the location of the stem cell sample in a donor and the age of the donor, accurately extracting a background area and multiple stem cell areas of the original image, determining a second quality evaluation index of each stem cell area according to a distance variation trend of edge pixels of each stem cell area, taking the survival rate of the stem cell sample in a culture stage as a third quality evaluation index of the stem cell sample, obtaining a total quality evaluation index of the stem cell sample by combining the above three quality evaluation indexes, and evaluating the quality of the stem cell sample using the total quality evaluation index, which not only improves the accuracy of stem cell area segmentation, but also ensures the accuracy and reliability of quality evaluation.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and more particularly to a method and system for evaluating stem cell quality based on image recognition. Background Art

[0002] With their unique self-renewal and multidirectional differentiation potential, stem cells have become a key factor in repairing and replacing damaged tissues or organs. For example, in the treatment scenario of myocardial infarction, stem cells can differentiate into cardiomyocytes to assist in the repair process of damaged heart tissue. Usually, stem cells are collected from the patient's body, cultured, and induced to differentiate into target cell types. Therefore, it is of great significance to evaluate the quality of stem cells. In the quality control section of my country's first national standard for "human stem cells", "Technical Specifications for the Management of Pluripotent Stem Cells in Biobanks", it is mentioned that important factors affecting the quality of stem cells include cell source, cell activity, cell survival rate, etc.

[0003] The existing Chinese patent document with announcement number CN118674716B discloses a stem cell quality evaluation method and system, which includes: collecting stem cell morphology images, and performing bilateral filtering on the stem cell morphology images to obtain preprocessed images; performing threshold segmentation on the preprocessed images through a multi-level Otsu algorithm and a local adaptive factor to obtain a threshold segmented image; performing multi-directional Sobel operator and gradient covariance matrix analysis on the threshold segmented image to obtain multi-dimensional morphological features; performing multi-scale non-maximum suppression and morphological feature description on the multi-dimensional morphological features to obtain a target feature point set; and inputting the multi-dimensional morphological features and the target feature point set into a multi-branch deep neural network to evaluate stem cell quality.

[0004] However, stem cells cultured in vitro often have dense stacking, blurred edge boundaries and a bird's nest-like appearance. This means that in actual operations, if we rely solely on traditional threshold segmentation algorithms, it is very easy to have problems with incomplete segmentation of complete stem cell areas. In addition, the above-mentioned technical solutions do not comprehensively consider the factors that affect stem cell quality in the process of evaluating cell quality, which leads to deviations in the assessment of stem cell quality and affects the accuracy and reliability of stem cell quality evaluation. Summary of the invention

[0005] In order to solve the problem that the traditional cell quality evaluation method is difficult to completely segment the stem cell area and the factors affecting the stem cell quality are not fully considered in the process of evaluating the cell quality, which affects the accuracy and reliability of the stem cell quality evaluation, the present invention proposes a stem cell quality evaluation method based on image recognition, comprising:

[0006] Collecting stem cell samples and original images of the stem cell samples, determining a first quality evaluation index of the stem cell samples according to the location of the stem cell samples in the donor and the age of the donor; identifying two boundary points in the horizontal direction of each pixel point according to the grayscale fluctuation characteristics of the pixel points in the horizontal direction of each pixel point in the original image;

[0007] The probability that each pixel belongs to the stem cell region is calculated based on the average gradient difference between each pixel and its two horizontal boundary points, the gray value variance of all pixels passing from each pixel to its two horizontal boundary points, and the difference between the gray value of each pixel and the maximum gray value of the original image, and the background region and stem cell region of the original image are determined according to the probability.

[0008] Obtain edge pixel points of each stem cell region, determine a second quality evaluation index for characterizing that the stem cell region presents a spindle-shaped feature according to a distance variation trend of the edge pixel points, and use the survival rate of the stem cell sample in the culture stage as a third quality evaluation index of the stem cell sample;

[0009] The first quality evaluation index, the average of the second quality evaluation index and the cumulative value of the third quality evaluation index are normalized to obtain a total quality evaluation index of the stem cell sample, and the quality of the stem cell sample is evaluated using the total quality evaluation index.

[0010] The above technical solution first obtains the first quality evaluation index of stem cell samples from the macro level, combined with industry norms and guidelines, and can quickly make a preliminary quantitative judgment on the overall quality of stem cell samples. And further, from the micro level, the grayscale change characteristics and gradient change characteristics of each pixel in the horizontal direction are analyzed, the probability of each pixel belonging to the stem cell area is calculated, and the background area and stem cell area of ​​the original image are accurately extracted, which can accurately separate the research object from the complex image background, improve the accuracy of stem cell area extraction or segmentation, and allow subsequent operations such as feature extraction and morphological analysis of stem cells to focus on truly meaningful stem cell areas, avoiding interference from irrelevant information. And further, the edge pixels of the stem cell area are analyzed. The edge pixels contain key morphological information such as the shape and contour of the stem cell area. By analyzing the distance change trend of the edge pixels, the morphological characteristics of the stem cell area are reflected. The activity of stem cells is closely related to the morphology. The spindle-shaped features are often more active. In this way, the second quality evaluation index is determined, and the quality of the stem cell area can be accurately quantified from the microscopic morphological level. And further, the survival rate of stem cell samples during the culture process is included in the quality evaluation system, which supplements the consideration of quality from the perspective of cell survival status. Furthermore, multiple aspects of information such as macroscopic sample standards, microscopic cell morphology, and cell survival status of stem cell samples are integrated into a final overall quality evaluation index, thereby improving the accuracy and reliability of stem cell sample quality evaluation.

[0011] Preferably, the first quality evaluation index of the stem cell sample is determined by:

[0012]

[0013] In the formula, is the first quality evaluation indicator of the stem cell sample, is the age of the donor of the stem cell sample, is the natural exponential function, A score for the location of the stem cell sample in the donor, wherein the score is determined according to a pre-acquired scoring rule, It is a normalization operation.

[0014] This technical solution improves the reliability of the first quality evaluation indicator by comprehensively considering the age of the stem cell sample donor and the sample location, using the natural exponential function to quantify the impact of age, and combining the location score to refine the differences.

[0015] Preferably, the second quality evaluation index is determined by the following method:

[0016] The two edge pixels corresponding to the maximum width of each stem cell area and The line connecting the two is the boundary, dividing the stem cell area into Partial and part, with and Traverse the starting point separately part, part, obtained for characterization part, The Euclidean distance sequence of the distance change trend of some edge pixels is recorded as part, The partial Euclidean distance sequence is used to calculate the second quality evaluation index of the stem cell area:

[0017] , For the The second quality evaluation index of the stem cell area, For the The maximum width of the stem cell region, , They are part, The average value of the Euclidean distance sequence of the parts, , They are part, The entropy value of the partial Euclidean distance sequence, is the natural exponential function, It is a normalization operation.

[0018] This technical solution precisely captures the dynamic changes in the distance between edge pixels, so that the second quality evaluation indicator of the stem cell area is accurately associated with the spindle-shaped characteristics, accurately reflects the individual activity of stem cells, improves the accuracy of microscopic morphology quantification, and makes the morphological analysis of stem cells more precise.

[0019] Preferably, part, The method for obtaining the partial Euclidean distance sequence is:

[0020] by and As the starting point, along the perpendicular direction of their respective gradient directions, part, The edge pixels of some parts are traversed, and each traversal will locate part, The Euclidean distance of a pair of edge pixels is calculated, and the traversal operation is continued until the Euclidean distance of a pair of edge pixels located by a certain traversal is 0. The traversal is stopped, and the Euclidean distances calculated in sequence during the traversal process are used to form part, Partial Euclidean distance sequence.

[0021] This technical solution can transform the abstract changes in stem cell edge morphology into intuitive numerical sequences. Euclidean distance, as a classic indicator for measuring the spatial distance between two points, accurately quantifies the relative position relationship between edge pixels, reflecting the subtle changes in the edge morphology of parts A and B. These subtle changes are closely related to the activity and quality of stem cells.

[0022] Preferably, the method for determining the background area and stem cell area of ​​the original image according to the magnitude of the probability is:

[0023] A probability threshold is preset; pixel points in the original image whose probability of belonging to the stem cell region is greater than the probability threshold are used as pixel points of the stem cell region to constitute multiple stem cell regions in the original image; other pixel points in the original image except the stem cell region are used as pixel points of the background region to constitute the background region of the original image.

[0024] Preferably, the calculation formula for the probability that each pixel belongs to the stem cell region is:

[0025]

[0026] In the formula, It is The probability that a pixel belongs to the cell region is It is The gray value of a pixel, is the maximum grayscale value of the original image, is the absolute value symbol, and Respectively The horizontal boundary points to the right and left of pixels. It is from The variance of the grayscale values ​​of all pixels from a pixel to its horizontal right boundary point, It is from The variance of the grayscale values ​​of all pixels from a pixel to its horizontal left boundary point is It is The average gradient difference between a pixel point and its horizontal boundary points to the right and left, are weights, and , It is a normalization operation.

[0027] This technical solution comprehensively considers multiple key image features such as the grayscale and gradient of pixels, quantifies and integrates multiple key image features, and accurately measures the possibility that each pixel belongs to the stem cell area, providing an accurate basis for accurate image segmentation and extraction of stem cell areas.

[0028] Preferably, the method for identifying two boundary points in the horizontal direction of each pixel is:

[0029] In the image Take a pixel point as the center, and start traversal operation in the horizontal right and left directions at the same time, and only traverse to one pixel point in the horizontal right and left directions each time;

[0030] During the traversal process, the current horizontal right pixel point, horizontal left pixel point and the first pixel point are calculated in real time. The average difference between the gradient values ​​of pixels; continue traversing until the average difference value obtained in a certain traversal exceeds half of the maximum gradient difference value of the original image, then stop traversing and record the horizontal right pixel point traversed at this time as , the horizontal left pixel is recorded as ,Will and As the Two boundary points in the horizontal direction of pixels.

[0031] This technical solution takes a single pixel as the core and traverses point by point in both horizontal directions. This microscopic operation can deeply explore the subtle feature changes of the image in the horizontal dimension, accurately capture the gradient information around the pixel, improve the accuracy of local feature extraction of the image, and ensure that key information around each pixel is not missed. Based on these subtle features, the boundary contour of the stem cell area can be accurately defined.

[0032] Preferably, the survival rate of the stem cell sample during the culture stage is obtained by:

[0033] The number of live cells is obtained by a live cell detection method, the total number of cells in the stem cell sample is obtained by an automated cell counting device, and the ratio of the number of live cells to the total number of cells is used as the survival rate of the stem cell sample during the culture stage.

[0034] This technical solution introduces the survival rate of stem cell samples during the culture stage to evaluate the quality of stem cell samples based on the dimension of cell activity, further increasing the authenticity and reliability of the quality assessment results.

[0035] Preferably, the method for evaluating the quality of the stem cell sample using the total quality evaluation index is:

[0036] A first total quality evaluation index threshold and a second total quality evaluation index threshold are preset, and the first total quality evaluation index threshold is greater than the second total quality evaluation index threshold; when the total quality evaluation index of the stem cell sample is greater than the first total quality evaluation index threshold, the quality of the stem cell sample is excellent; when the total quality evaluation index of the stem cell sample is less than the second total quality evaluation index threshold, the quality of the stem cell sample is inferior; when the total quality evaluation index of the stem cell sample is equal to the first total quality evaluation index threshold or equal to the second total quality evaluation index threshold, or between the first total quality evaluation index threshold and the second total quality evaluation index threshold, the quality of the stem cell sample is ordinary.

[0037] The present invention also provides a stem cell quality evaluation system based on image recognition, the stem cell quality evaluation system comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps of the stem cell quality evaluation method.

[0038] The present invention has the following effects:

[0039] Starting from the macro sample standard, the present invention combines the micro cell morphology analysis to accurately extract the image area and analyze the edge pixels, so as to realize the multi-dimensional rapid preliminary quantification of the overall quality of the sample. At the same time, the survival rate of the stem cell culture stage is incorporated into the system, and the overall quality evaluation index is obtained by integrating various information. This not only improves the accuracy of stem cell region segmentation, but also ensures the accuracy and reliability of quality evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0041] Figure 1 It is a schematic flow chart of the method of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0043] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0044] Reference Figure 1The method for evaluating stem cell quality based on image recognition provided by the present invention comprises steps S1 to S6:

[0045] S1: Collect stem cell samples and analyze them at a macro level to determine the first quality evaluation indicator of the stem cell samples.

[0046] In one embodiment, when collecting stem cell samples, the patient is used as the donor, and the determination of the collection location needs to comprehensively consider factors such as the patient's overall physical condition and age. Professional knowledge and clinical experience are used to determine the appropriate location for collecting stem cell samples from the donor to ensure that the collected stem cell samples can meet the quality requirements and meet the treatment needs to the greatest extent possible based on the patient's condition. If the patient has myocardial infarction, priority is given to collecting from areas rich in myocardial stem cells around the heart, such as the subepicardial tissue, where stem cells may have higher potential for differentiating into myocardial cells to repair damaged heart tissue; if the patient has a neurological disease, such as Parkinson's disease, the focus is on collecting stem cells from specific areas of the brain, such as the substantia nigra in the midbrain, which are closely related to the function of the nervous system, and ultimately a stem cell sample composed of multiple stem cells is collected.

[0047] In one embodiment, for the collected stem cell sample, a first quality evaluation index of the stem cell sample is determined according to the location of the stem cell sample in the donor and the age of the donor.

[0048] The determination process of the first quality evaluation indicator is mainly based on the following two dimensions:

[0049] Considering the age dimension, a large number of scientific studies and clinical practices have shown that as the human body ages, cell vitality tends to gradually decline. The stem cells in young people's bodies not only have an advantage in quantity, but also have a stronger proliferation ability than the elderly. They can self-renew and differentiate more efficiently, and thus play a more significant role in tissue repair, organ regeneration, etc. Therefore, the age factor plays a key negative correlation role in the stem cell sample quality evaluation system, that is, the older the age, the lower the baseline expectation of stem cell sample quality in theory.

[0050] Considering the location of the stem cell sample in the donor, based on existing medical research results and clinical application feedback, differentiated scoring standards are set for different locations. The specific details are as follows:

[0051] Umbilical cord blood: 1 point. As an important source of stem cells, umbilical cord blood is usually rich in various types of high-quality stem cells. These stem cells have extremely strong original activity, high proliferation potential and strong differentiation ability. Many breakthroughs have been achieved in the fields of blood system disease treatment and immune regulation. Bone marrow: 0.8 points. Bone marrow has long been regarded as a high-quality mesenchymal stem cell. One of the main sources of fat, the stem cells therein have outstanding performance in supporting hematopoiesis, repairing bones and cartilage tissues, etc., with high stability and good differentiation adaptability, providing a solid guarantee for the treatment of various diseases. Adipose tissue: Assigned 0.6 points. Adipose-derived stem cells have gradually attracted attention in recent years. Although its stem cell content is slightly less than that of bone marrow, umbilical cord blood and other parts, it has the advantages of convenient material collection and little damage to the donor. It also shows certain application potential in the field of soft tissue repair and regenerative medicine, and can provide effective cell support for specific treatment needs. Skin: Assigned 0.4 points. Compared with other major sources, stem cells obtained from the skin are usually relatively limited in terms of quantity, scale, proliferation and differentiation potential. However, in sub-fields such as skin trauma repair and local tissue regeneration, skin stem cells can still play a certain role and have a local application value that cannot be ignored. Other locations: 0.2 points are assigned. Due to many factors such as the complex local microenvironment and difficulty in maintaining cell stemness, the proliferation ability and differentiation potential of stem cells contained in other areas of the human body other than the above-mentioned typical parts are generally more restricted, and their universality in clinical applications is relatively low.

[0052] By organically combining the age factor with the donor location score of the stem cell sample, the first quality evaluation index of the stem cell sample is calculated according to the following formula:

[0053]

[0054] In the formula, It is the first quality evaluation indicator of stem cell samples. is the age of the donor of the stem cell sample, is the natural exponential function, is a score for the location of the stem cell sample in the donor, wherein the score is determined according to a pre-acquired scoring rule, by The function performs normalization operation, limiting the first quality evaluation index, which is affected by complex factors and has a variable value range, to The range of 1:1 makes different batches of stem cell samples comparable.

[0055] In this formula, incorporating the age of the stem cell sample donor into the calculation of the first quality evaluation index has important biological significance. With age, the functions and characteristics of human cells, including stem cells, change. Generally speaking, stem cells from older donors may be relatively weak in terms of proliferation ability and differentiation potential. Through the natural exponential function, the impact of age on quality evaluation is quantitatively reflected exponentially. The older the age, the smaller the value of this part, and the proportion of this part in the overall index changes, which reasonably reflects the potential negative impact of age factors on stem cell quality.

[0056] In this formula, the position of the stem cell sample in the donor is taken into account and a score is given according to the pre-acquired scoring rules. This is considered from the perspective of the microscopic anatomical position of the sample source. Differences in the microenvironment of different parts of the body will affect the initial state, gene expression and other characteristics of stem cells. For example, stem cells in certain tissues may have higher activity and quality potential because they are in a more active metabolic environment for a long time or are stimulated by more paracrine signals. It broadens the breadth of consideration of the first quality evaluation indicator, makes up for the shortcomings of evaluating only from a single macro age factor, and makes the evaluation results more nuanced. It can not only distinguish stem cells from donors of different ages, but also accurately judge quality differences between different samples from donors of the same age based on location factors, thereby enhancing the adaptability of the evaluation system to the diversity of stem cell sample sources.

[0057] In summary, this scheme comprehensively considers the donor age and sample location. From a biological perspective, the age factor reflects the changes in functional characteristics of stem cells as the donor ages, and the location score takes into account the impact of the microenvironment of different body parts on the initial state of stem cells. The combination of the two comprehensively covers the key sample source information, making the first quality evaluation indicator more reliable.

[0058] S2: Collect microscope images of stem cell samples and analyze the grayscale value characteristics and gradient value characteristics of each pixel at a microscopic level to accurately extract multiple stem cell areas.

[0059] In one embodiment, a professional microscope device with high resolution and low chromatic aberration characteristics is selected to ensure that the subtle features of the stem cell sample can be clearly captured. During the operation, the collected stem cell sample is first carefully and evenly spread on a special slide, and a pipette is used to control the amount of sample dripping to avoid sample accumulation or uneven distribution affecting the imaging effect. Subsequently, the slide is placed steadily on the stage of the microscope, and the parameters such as the objective lens magnification, focal length, and light source brightness of the microscope are reasonably adjusted according to factors such as the type of sample and the estimated cell density. For example, for samples with low cell density, the objective lens magnification is appropriately increased to more clearly observe the morphology of a single stem cell; for samples with high cell activity and more sensitive to light, the light source brightness is accurately weakened to prevent strong light from damaging the cells and changing their original morphological characteristics. Finally, the microscope imaging function is started to obtain a clear, accurate microscope image that can truly reflect the original state of the stem cell sample as the original image of the stem cell sample.

[0060] In one embodiment, since the original image of the stem cell sample may have problems such as noise, poor contrast, and unclear details, a series of optimization processes need to be performed on it, including:

[0061] First, the median filter is used to remove the noise in the original image. This filter can effectively filter out the noise while retaining important details such as the edge of the original image by selecting the median of the pixel neighborhood to replace the central pixel value, ensuring that the clarity of the original image is not greatly affected. Then, the histogram equalization algorithm is used to enhance the contrast of the original image. This algorithm remaps the pixel grayscale value based on the grayscale histogram distribution of the original image, so that the details of the darker or brighter areas in the original image can be highlighted, making the layering of the entire original image clearer, which is convenient for the subsequent accurate observation and analysis of stem cell characteristics. Subsequently, the Laplace filter is used to sharpen the original image. It enhances the edge and detail information by highlighting the high-frequency components in the original image, making the contour, internal structure and other subtle features of the stem cells clearer and more recognizable. Finally, the original image processed as above is grayed.

[0062] In one embodiment, after obtaining the original image of the stem cell sample, the background area and stem cell area of ​​the image are accurately extracted according to the following method:

[0063] First, the probability that each pixel in the original image belongs to the stem cell region is calculated, and the probability threshold is preset to 0.6, which is an empirical value. Then, the pixels with a probability of belonging to the stem cell region greater than 0.6 are regarded as the pixels belonging to the stem cell region in the original image, and the other pixels except all the pixels belonging to the stem cell region in the original image are regarded as the pixels belonging to the background region. Finally, all the pixels belonging to the stem cell region constitute multiple stem cell regions in the original image, and all the pixels belonging to the background region constitute the background region in the original image.

[0064] In one embodiment, a method for obtaining the probability that each pixel of the original image of the stem cell sample belongs to the stem cell region includes:

[0065] First, take the pixel point in the upper left corner of the original image as the coordinate origin, and horizontally to the right Axis, vertically downward Axis, and obtain the grayscale value and gradient value corresponding to each pixel in the original image. The grayscale value can reflect the brightness of the pixel and the brightness information of the original image at each pixel. For example, the stem cell area has a lower grayscale value (close to black) because the cell nucleus is dense and opaque, and the cytoplasm is relatively transparent and has a lighter grayscale value. The background area has no cell structure and high light transmittance, and usually has a brighter grayscale value (close to white or light gray). The gradient value can characterize the speed of change of the pixel in the original image. For example, at the edge of the stem cell area and the background area, the gradient value of the pixel is often larger.

[0066] The first The pixel point is taken as the center, and the traversal operation is carried out in the right and left directions respectively. Each time, the traversal operation will be carried out to the first pixel point. There are 1 pixel point in the horizontal right direction and 1 pixel point in the horizontal left direction. In one example, the first traversal, traversing to the The pixel point is between the first pixel point and the The first pixel in the horizontal left direction; the second traversal, traversing to the The pixel point is between the second pixel point and the The second pixel to the left of the pixel is located horizontally, and so on.

[0067] In each traversal process, the gradient value of the horizontal right pixel point currently traversed and the The difference between the gradient values ​​of the pixels (the absolute value of the difference), and the gradient value of the pixel currently traversed in the horizontal left direction and the The difference of the gradient values ​​of the pixels is calculated to get the average difference of the gradient values; for example, the second traversal to the The gradient value of the second pixel point in the horizontal right direction is 20, the gradient value of the second pixel point in the horizontal left direction is 30, and the The gradient value of each pixel is 25, and the average difference of the gradient values ​​is .

[0068] Here, by traversing the pixels in the horizontal direction and calculating the average difference value, the dynamic changes of the pixels in the horizontal direction around each pixel at the gradient level are essentially accurately measured. Due to the different transmittances of the nucleus and cytoplasm contained in the stem cells, the intensity of the light undergoes a complex change process when penetrating different parts of the stem cells. Mapped to the original image, the gradient value in the horizontal direction of each pixel will fluctuate, that is, the grayscale change rate between adjacent pixels is unstable, and at the boundary of the stem cell area, this gradient change reaches the most significant degree. Therefore, by continuously updating the average difference value, the mutation points in the horizontal direction around each pixel can be tracked in real time and accurately, and then the boundary between the cell area and the background area can be accurately identified. Compared with the traditional method of using a fixed threshold to determine the boundary, this mode of judging based on the relative degree of change fully considers the complex and dynamic characteristics of cells and backgrounds in actual images, and reduces the probability of misjudgment.

[0069] Continue traversing until a certain time when the gradient value of the pixel point in the horizontal right direction and the gradient value of the pixel point in the horizontal left direction that are currently traversed are the same as the gradient value of the pixel point in the horizontal right direction. When the average difference between the gradient values ​​of the pixels exceeds half of the maximum gradient difference value of the original image (the difference between the maximum gradient value and the minimum gradient value of the original image), the traversal stops. This means that the traversal has reached the boundary between the cell area and the background area, and the horizontal right pixel traversed at this time is recorded as , the horizontal left pixel is recorded as ,Will and As the Two boundary points in the horizontal direction of pixels.

[0070] Next, calculate from The variance of the grayscale values ​​of all pixels from a pixel to its horizontal right boundary point is recorded as , calculated from the The variance of the grayscale values ​​of all pixels from the pixel point to the boundary point in the horizontal left direction is recorded as .

[0071] By calculating the grayscale variance here, the feature consistency of the pixel in the horizontal direction can be judged more accurately, and the grayscale variance reflects the fluctuation degree of the grayscale value in the horizontal direction of the pixel. The different transmittances of the nucleus and cytoplasm in the stem cell cause the grayscale value of the pixel in the stem cell area to fluctuate greatly in the horizontal direction. The greater the grayscale value fluctuation, the more likely the pixel is to be located in the stem cell area, while the grayscale value of the pixel in the background area is relatively stable and will not fluctuate greatly.

[0072] Finally, the probability that each pixel in the original image belongs to the stem cell region is calculated:

[0073]

[0074] In the formula, It is The probability that a pixel belongs to the cell region is It is The gray value of a pixel, is the maximum grayscale value of the original image, is the absolute value symbol, and Respectively The horizontal boundary points to the right and left of pixels. It is from The variance of the grayscale values ​​of all pixels from a pixel to its horizontal right boundary point, It is from The variance of the grayscale values ​​of all pixels from a pixel to its horizontal left boundary point is It is The average gradient difference between a pixel point and its horizontal boundary points to the right and left, are weights, and , It is a normalization operation.

[0075] In this formula, For the The difference between the gray value of the pixel and the maximum gray value of the original image is divided by 255. Normalization operation is performed. Since the cell nuclei in the stem cell area are dense and opaque, the gray value is low, and there is a significant difference from the background area. Some can highlight the deviation between the gray value of the pixel and the background area. The bigger, the The more pixels there are, the more likely they are in the stem cell area. Represents the average of the two variances. The larger the average, the The more pixels there are, the more likely they are to be located in the stem cell area, because the grayscale values ​​of the pixels in the stem cell area fluctuate greatly in the horizontal direction, while the grayscale values ​​of the pixels in the background area are relatively stable. Indicates The average gradient difference between the pixel point and its horizontal right and left boundary points. The larger the average gradient difference, the The more likely a pixel point is to be located in the stem cell region, the greater the difference in gradient value between the pixel points inside the stem cell region and the pixel points at the boundary of the stem cell region.

[0076] In this formula, the weights are set based on the characteristics of the stem cell area and the background area. , are all empirical values, because the most important and intuitive way to distinguish whether each pixel belongs to the cell area or the background area is the difference in grayscale value, followed by the fluctuation degree of the grayscale value of the pixels traversed by the pixel in the horizontal direction, and finally the difference in gradient value. The reason why the average difference in gradient value has the smallest weight is that the boundary of the stem cell area is between the stem cell area and the background area. If a pixel point is located in the background area, the gradient value difference between it and the pixel point at the stem cell boundary will be large, so it needs to be given a smaller weight. By setting the weight in this way, the calculation result can better reflect the actual situation and accurately determine the probability that the pixel point belongs to the stem cell area.

[0077] In summary, this formula comprehensively calculates the probability of each pixel belonging to the stem cell area from multiple perspectives, avoiding the one-sidedness of single factor judgment. The greater the difference between the gray value of the pixel and the maximum gray value of the original image, the The greater the fluctuation of the grayscale value between the pixels traversed by the pixel point in the horizontal direction, the The larger the average difference in gradient values ​​between the pixel point and the boundary pixel point of the stem cell area, the The larger the pixel belongs to the cell area, the greater the probability is, and vice versa.

[0078] In one embodiment, after acquiring a plurality of stem cell regions and background regions in the original image, the following operations are further performed:

[0079] First, through the binarization operation, the stem cell area is set to white (pixel value is 255) and the background area is set to black (pixel value is 0), so as to construct a binary image. Then, the erosion method in morphological operation is used to process the binary image, and the size is 3 3 square core, the stem cell area is reduced by this corrosion operation, the purpose of which is to remove the relatively small connecting parts between stem cells, so as to achieve effective separation of adherent cells. After the corrosion operation is completed, the expansion operation is used to restore the size of the cell area. The relatively small connecting parts that were previously corroded will not be restored during the expansion process.

[0080] Next, the binary image is further converted into a grayscale image. During the conversion process, there is an important calculation step, which is to determine the distance from each pixel point (white pixel point) of the stem cell area in the binary image to the background pixel point (black pixel point) closest to it. This step can provide key spatial position information for subsequent segmentation operations.

[0081] Finally, after completing the distance calculation, the watershed algorithm is used to perform image segmentation based on the converted grayscale image. The principle of the watershed algorithm is similar to the concept of dividing terrain such as mountains and valleys in geography. In the image, it can divide the original image into multiple different areas based on the similarity between pixels and the distance calculated previously. After being processed by the watershed algorithm, the above binary image is divided into multiple white areas, each of which corresponds to a stem cell area. The total number of stem cell areas is N. Through this operation, multiple stem cell areas are segmented from the original image.

[0082] S3: Analyze the morphological characteristics of each stem cell region to determine a second quality evaluation index for each stem cell region.

[0083] The quality of stem cells is closely related to their morphological characteristics. Usually, stem cells of good quality show significant regularity in morphology, with plump and round cell bodies, and most typically showing a distinct spindle-shaped appearance. This morphology implies that the internal structure of the cell is complete, the function is sound, and it has stronger vitality and differentiation potential. On the contrary, the morphology of stem cells of poor quality is more chaotic, often more irregular, with wide and flat cell bodies, lacking the proper three-dimensional sense and compactness. This not only reflects that the cells may have encountered interference from adverse factors during their growth and development, resulting in deviations in their morphological construction, but also means that their functional activity may be greatly reduced, making it difficult to play the key roles of efficient repair and regeneration of high-quality stem cells. In the stem cell quality evaluation system, morphological characteristics can be used as an intuitive and important basis for judgment.

[0084] Therefore, here, the morphological characteristics of each stem cell region are analyzed according to the distance change trend of the edge pixel points of each stem cell region, specifically, whether the spindle-shaped feature of each stem cell region is obvious is judged, and reflected by the second quality evaluation index of each stem cell region. The larger the value of the second quality evaluation index, the more obvious the spindle-shaped feature of the stem cell region is, and the better the quality of the stem cell region, and vice versa.

[0085] In one embodiment, the method for determining the second quality evaluation index of each stem cell region is:

[0086] First, obtain the Euclidean distance sequence that can reflect the morphological characteristics of each stem cell region: the two edge pixel points corresponding to the maximum width of each stem cell region and The line connecting the two is the boundary, dividing the stem cell area into Partial and part, with and Traverse the starting point separately part, Part, get part, The Euclidean distance sequence of the part, where The partial Euclidean distance sequence is used to characterize The distance change trend of some edge pixels, The partial Euclidean distance sequence is used to characterize The distance change trend of some edge pixels. Here, the distance change trend of edge pixels is used to characterize the morphological characteristics of the stem cell area, which is consistent with the actual morphology and structural characteristics of stem cells. The morphological characteristics of the stem cell area are of great significance for understanding the growth status and structural integrity of stem cells. The Euclidean distance sequence can better capture this information. The Euclidean distance sequence represents the morphological characteristics of the stem cell area in a quantifiable way. The area is divided by the edge pixel line corresponding to the maximum width and the Euclidean distance sequence is obtained respectively, which can describe the morphological characteristics of the stem cell area in detail from different angles.

[0087] Next, the second quality evaluation index of the stem cell region is calculated:

[0088]

[0089] In this formula, For the The second quality evaluation index of the stem cell area, For the The maximum width of the stem cell region, , They are part, The average value of the Euclidean distance sequence of the parts, , They are part, The entropy value of the partial Euclidean distance sequence, is the natural exponential function, It is a normalization operation.

[0090] In this formula, It is a key indicator to describe the overall size of the stem cell area. It intuitively reflects the scale of the stem cell area, which is a basic and important reference factor when evaluating cell morphology, because cell size and morphology are usually interrelated.

[0091] In this formula, and Reflects , The average distance between the edge pixels of two parts. and Used to measure , The degree of disorder or uncertainty of the Euclidean distance sequence of the two parts, a lower entropy value means that the Euclidean distance sequence is relatively stable and regular, that is, the distance change of edge pixels is more uniform, indicating that the smoother and more regular the edge of the stem cell area, the more it conforms to the spindle-shaped characteristics. It reflects , The difference in the degree of disorder of the Euclidean distance sequence between the two parts is that the smaller the value, the more similar the distance change trends of the two parts are, indicating that the stem cell region , The more symmetrical the shapes of the two parts are.

[0092] In this formula, When the average value of the Euclidean distance sequence of the A and B parts of the stem cell region is , is small, and the difference in entropy between the two parts of the Euclidean distance sequence is When smaller, The larger the value, the greater the maximum width of several cell regions. The larger it is, the more the stem cell area conforms to the spindle-shaped characteristics of being wide in the middle and narrow at both ends, with the two ends being relatively similar, and the quality of the stem cell area is higher.

[0093] In one embodiment, part / The method for obtaining the partial Euclidean distance sequence is:

[0094] by and As the starting point, along the perpendicular direction of their respective gradient directions, part, The edge pixels of some parts are traversed, and each traversal will locate part, The Euclidean distance of a pair of edge pixels is calculated, and the traversal operation is continued until the Euclidean distance of a pair of edge pixels located by a certain traversal is 0. The traversal is stopped, and the Euclidean distances calculated in sequence during the traversal process are used to form part, The Euclidean distance sequence of the part. part, Some Euclidean distance sequences can accurately capture subtle changes in the edge of the stem cell region, because the vertical direction of the gradient can reflect the edge changes to the greatest extent. Each traversal calculation of the Euclidean distance can accurately record the distance changes between edge pixels. When the Euclidean distance is 0, the traversal is stopped, and the traversal boundary can be reasonably determined, ensuring that the obtained Euclidean distance sequence completely and accurately reflects the edge morphology of the stem cell region.

[0095] Since good quality stem cells have a spindle-shaped feature, that is, they are wide in the middle and narrow at both ends, and the degree of narrowness at both ends is similar, and Starting from the edge pixels, the edges are traversed in the direction perpendicular to their respective gradients and the Euclidean distance is calculated. This method is consistent with the judgment criteria of the spindle-shaped feature and can effectively capture the morphological change information of the stem cell region from the middle to the two ends. In particular, the Euclidean distance sequence can intuitively reflect the degree of narrowness at the two ends and its changes, which helps to judge whether the stem cell region meets the high-quality morphological requirements of the spindle-shaped feature.

[0096] S4: The survival rate of stem cell samples during the culture stage is used as the third quality evaluation indicator of stem cell samples.

[0097] Cell viability is one of the key indicators for measuring stem cell quality, and is usually presented as the percentage of live cells to total cells. The determination of live cell counts can be achieved with the help of a variety of mature live cell detection methods, including: trypan blue exclusion test, MTT / CCK-8 test, etc., while the total cell count can be completed by traditional cell counting plate counting or by using more advanced and efficient automated cell counting equipment.

[0098] In one embodiment, the number of viable cells is obtained based on a trypan blue exclusion assay:

[0099] When the stem cell sample reaches the final culture stage, according to the principle of the classic trypan blue exclusion experiment, since living cells have intact cell membranes and can exclude trypan blue dye, this characteristic can be used to accurately identify living cells, and then the number of living cells in the sample can be obtained, which is recorded as d; at the same time, the automated cell counting equipment, with its high-precision optical recognition and rapid counting functions, can be used to count the total number of cells in the sample, which is recorded as D. In this way, the third quality evaluation index of the stem cell sample is obtained. for:

[0100]

[0101] When the proportion of living cells in the stem cell sample is larger, it means that the cell survival condition of the stem cell sample during the culture stage is better. Accordingly, the calculated third quality evaluation index is The larger the value, the more intuitive and accurate it is, which reflects the survival vitality of the stem cell sample during the culture process, and provides a key and reliable basis for the comprehensive evaluation of stem cell quality.

[0102] S5: Determine the overall quality evaluation index of stem cell samples.

[0103] In one embodiment, the average value of the second quality evaluation index of all stem cell regions included in the stem cell sample is first calculated, and the normalized value of the first quality evaluation index, the third quality evaluation index and the accumulated value of the average value of the stem cell sample is used as the total quality evaluation index of the stem cell sample. Specifically, it is expressed as:

[0104]

[0105] In the formula, It is the overall quality evaluation index of stem cell samples. It is the first quality evaluation indicator of stem cell samples, reflecting the quality of their cell sources. The stem cell sample contains The second quality evaluation index of the stem cell area, is the total number of stem cell regions. By calculating the average value of the second quality evaluation index of all stem cell regions, it reflects the quality of the stem cell sample in terms of cell activity. It is the third quality evaluation indicator of stem cell samples, reflecting the quality of cell survival rate. is the normalization function.

[0106] S6: Using the total quality evaluation index to evaluate the quality of the stem cell sample.

[0107] In one embodiment, the method for evaluating the quality of the stem cell sample using the total quality evaluation index is as follows: the first total quality evaluation index threshold is set to 0.9, and the second total quality evaluation index threshold is set to 0.5. When the total quality evaluation index of the stem cell sample is greater than 0.9, the quality of the stem cell sample is determined to be excellent; when the total quality evaluation index of the stem cell sample is less than 0.5, the quality of the stem cell sample is determined to be inferior; when the total quality evaluation index of the stem cell sample is between 0.5 and 1.5, the quality of the stem cell sample is determined to be poor; When the sample is within the range of 0.1 to 0.2, the quality of the stem cell sample is judged to be average.

[0108] In addition, the present invention also provides a stem cell quality evaluation system based on image recognition, the stem cell quality evaluation system includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps of the stem cell quality evaluation method.

[0109] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0110] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A method for evaluating stem cell quality based on image recognition, characterized in that: include: Collecting stem cell samples and original images of the stem cell samples, determining a first quality evaluation index of the stem cell samples according to the location of the stem cell samples in the donor and the age of the donor; identifying two boundary points in the horizontal direction of each pixel point according to the grayscale fluctuation characteristics of the pixel points in the horizontal direction of each pixel point in the original image; According to the average gradient difference between each pixel and its two horizontal boundary points, the gray value variance of all pixels passing from each pixel to its two horizontal boundary points, and the difference between the gray value of each pixel and the maximum gray value of the original image, the probability that each pixel belongs to the stem cell area is calculated, and the background area and stem cell area of ​​the original image are determined according to the size of the probability; Obtain the edge pixel points of each stem cell region, and determine the second quality evaluation index for characterizing the spindle-shaped feature of the stem cell region according to the distance change trend of the edge pixel points, in the following manner: and The line connecting the two is the boundary, dividing the stem cell area into Partial and part, with and Traverse the starting point separately part, part, obtained for characterization part, The Euclidean distance sequence of the distance change trend of some edge pixels is recorded as part, The partial Euclidean distance sequence is used to calculate the second quality evaluation index of the stem cell area: , For the The second quality evaluation index of the stem cell area, For the The maximum width of the stem cell region, , They are part, The average value of the Euclidean distance sequence of the parts, , They are part, The entropy of the partial Euclidean distance sequence, is the natural exponential function, It is a normalization operation; and the survival rate of stem cell samples during the culture stage is used as the third quality evaluation index of stem cell samples; The first quality evaluation index, the average of the second quality evaluation index and the cumulative value of the third quality evaluation index are normalized to obtain a total quality evaluation index of the stem cell sample, and the quality of the stem cell sample is evaluated using the total quality evaluation index.

2. The method for evaluating stem cell quality based on image recognition according to claim 1, characterized in that: The first quality evaluation index of the stem cell sample is determined in the following manner: ； In the formula, is the first quality evaluation indicator of the stem cell sample, is the age of the donor of the stem cell sample, is the natural exponential function, A score for the location of the stem cell sample in the donor, wherein the score is determined according to a pre-acquired scoring rule, It is a normalization operation.

3. The method for evaluating stem cell quality based on image recognition according to claim 1, characterized in that: part, The method for obtaining the partial Euclidean distance sequence is: by and As the starting point, along the perpendicular direction of their respective gradient directions, part, The edge pixels of some parts are traversed, and each traversal will locate part, The Euclidean distance of a pair of edge pixels located in the part is calculated, and the traversal operation is continued until the Euclidean distance of a pair of edge pixels located in a certain traversal is 0, then the traversal is stopped, and the Euclidean distances calculated in sequence during the traversal process are used to form part, Partial Euclidean distance sequence.

4. The method for evaluating stem cell quality based on image recognition according to claim 1, characterized in that: The method for determining the background area and stem cell area of ​​the original image according to the size of the probability is: A probability threshold is preset; pixel points in the original image whose probability of belonging to the stem cell area is greater than the probability threshold are used as pixel points of the stem cell area to constitute multiple stem cell areas in the image; other pixel points in the original image except the stem cell area are used as pixel points of the background area to constitute the background area of ​​the original image.

5. The method for evaluating stem cell quality based on image recognition according to claim 4, characterized in that: The calculation formula for the probability that each pixel belongs to the stem cell area is: ； In the formula, It is The probability that a pixel belongs to the stem cell region is It is The gray value of a pixel, is the maximum grayscale value of the original image, is the absolute value symbol, and Respectively The horizontal boundary points to the right and left of pixels. It is from The variance of the grayscale values ​​of all pixels from a pixel to its horizontal right boundary point, It is from The variance of the grayscale values ​​of all pixels from a pixel to its horizontal left boundary point is It is The average gradient difference between a pixel point and its horizontal boundary points to the right and left, are weights, and , It is a normalization operation.

6. The method for evaluating stem cell quality based on image recognition according to claim 5, characterized in that: The method for identifying the two boundary points in the horizontal direction of each pixel is: In the original image Take a pixel point as the center, and start traversal operation in the horizontal right and left directions at the same time, and only traverse to one pixel point in the horizontal right and left directions each time; During the traversal process, the current horizontal right pixel point, horizontal left pixel point and the first pixel point are calculated in real time. The average difference between the gradient values ​​of pixels; continue traversing until the average difference value obtained in a certain traversal exceeds half of the maximum gradient difference value of the original image, then stop traversing and record the horizontal right pixel point traversed at this time as , the horizontal left pixel is recorded as ,Will and As the Two boundary points in the horizontal direction of pixels.

7. The method for evaluating stem cell quality based on image recognition according to claim 1, characterized in that: The survival rate of stem cell samples during the culture stage is obtained by: The number of live cells is obtained by a live cell detection method, the total number of cells in the stem cell sample is obtained by an automated cell counting device, and the ratio of the number of live cells to the total number of cells is used as the survival rate of the stem cell sample during the culture stage.

8. The method for evaluating stem cell quality based on image recognition according to claim 1, characterized in that: The method for evaluating the quality of the stem cell sample using the total quality evaluation index is: A first total quality evaluation index threshold and a second total quality evaluation index threshold are preset, and the first total quality evaluation index threshold is greater than the second total quality evaluation index threshold; when the total quality evaluation index of the stem cell sample is greater than the first total quality evaluation index threshold, the quality of the stem cell sample is excellent; when the total quality evaluation index of the stem cell sample is less than the second total quality evaluation index threshold, the quality of the stem cell sample is inferior; when the total quality evaluation index of the stem cell sample is equal to the first total quality evaluation index threshold or equal to the second total quality evaluation index threshold, or between the first total quality evaluation index threshold and the second total quality evaluation index threshold, the quality of the stem cell sample is ordinary.

9. A stem cell quality evaluation system based on image recognition, characterized in that: The stem cell quality evaluation system comprises a memory and a processor, wherein a computer program is stored in the memory, and the processor executes the computer program to implement the steps of the stem cell quality evaluation method according to any one of claims 1 to 8.

Images