Rat mood cycle judgment method based on image recognition

Abstract

The invention relates to the technical field of image recognition and judgment, in particular to a rat mood cycle judgment method based on image recognition. According to the technical scheme, the method comprises the steps of sample collection and preparation, image collection and optimization, cell segmentation and positioning, feature extraction and analysis, cell classification and recognition, and result verification and feedback. Through systematic and fine operation, the efficiency and accuracy of rat vagina exfoliated cell morphology recognition are remarkably improved, the scientificity of recognition is comprehensively guaranteed from accurate sample collection, image collection optimization, multi-dimensional feature extraction and integrated learning classification to result verification feedback, the detection time is greatly shortened, and the detection efficiency is improved. Manual errors are reduced, the performance can be continuously improved through dynamic optimization, and efficient and reliable technical support is provided for research of the rat mood cycle.

Description

Technical Field

[0001] The invention relates to the technical field of image recognition and judgment, and in particular to a method for judging the estrous cycle of rats based on image recognition. Background Art

[0002] In the field of biomedical research, rats are commonly used experimental animals, and accurate monitoring of their estrous cycle is of vital importance for studying estrogen-related physiological mechanisms, drug development, and reproductive biology. Estrogen can affect the estrous cycle of rats, and its mechanism of action is mainly to promote the proliferation of vaginal mucosal cells, thereby accelerating the shedding of vaginal mucosal cells.

[0003] The traditional method of monitoring the estrous cycle of rats mainly relies on the daily examination of vaginal exfoliated cells. The specific operation process is to first feed the rats for one week, and then collect vaginal exfoliated cells from 8 to 9 am every day for the next 10 days. During the collection process, the staff needs to use a pipette to drop 200 μL of sterile saline into each well of the 48-well plate, and match the well number with the rat number one by one. After that, fix the rat with the left hand, pull the rat tail to expose the vulva, hold the pipette with the right hand to absorb the saline in the corresponding well number, gently place the tip of the pipette 2-4 mm into the vaginal opening of the rat, slowly inject and aspirate, repeat 3 times, and inject 80 μL of the aspirate into the original corresponding well, repeatedly blow and beat to make the vaginal exfoliated cells enter the saline, and each rat needs to replace a new pipette.

[0004] After the sample collection is completed, the 48-well plate is left to stand for 10 minutes, and the cells are allowed to settle to the bottom of the wells before being placed under an inverted microscope for observation. The staff needs to determine the estrous cycle of the rat based on the dynamic changes in the types and proportions of the three types of white blood cells, nucleated epithelial cells, and non-nucleated epithelial cells observed under the microscope. This method of judgment not only requires the staff to have rich experience and be able to accurately identify the morphological characteristics of different cells, but also requires a deep understanding of the relationship between changes in cell proportions and the estrous cycle.

[0005] However, this traditional method of judging the estrous cycle of rats through manual microscopy has many disadvantages. First, the operation process is extremely cumbersome, and samples need to be collected at a fixed time every day. In addition, the collection process involves multiple delicate operation steps, which requires high technical skills of the operator. Secondly, microscopic examination is time-consuming and labor-intensive. Dozens of films need to be processed for one microscopic examination, and it usually takes a whole day to complete the observation and analysis of all samples. Furthermore, the accuracy of manual judgment depends to a large extent on the professional level and working status of the operator, and is easily affected by subjective factors. There may be differences in judgment between different operators, which makes it difficult to ensure the reliability and consistency of the results.

[0006] In summary, the present application proposes a method for determining the estrous cycle of rats based on image recognition. Summary of the invention

[0007] The purpose of the present invention is to address the problem of low efficiency in artificially identifying cell morphological characteristics in the background technology and to propose a method for determining the estrous cycle of rats based on image recognition.

[0008] The technical solution of the present invention: a method for determining the estrous cycle of rats based on image recognition, comprising the following steps: Sample collection and preparation: Rat vaginal exfoliated cell samples were collected, processed and preserved according to specific methods; Image acquisition and optimization: Use professional equipment to acquire high-resolution images and perform geometric correction, color correction, and image denoising preprocessing; Cell segmentation and positioning: A semantic segmentation model based on a convolutional neural network is used to segment cells, and the segmented cells are accurately positioned and marked; Feature extraction and analysis: extract the shape, texture and optical features of cells, and perform feature selection and dimensionality reduction; Cell classification and identification: Use ensemble learning classifiers to classify cells and optimize the classification results; Result verification and feedback: Verify the classification results through multiple indicators, and provide feedback and optimization for each link based on the verification results.

[0009] Optionally, when collecting rat vaginal exfoliated cell samples, use a high-precision pipette, and draw and inject 200 μL of sterile saline each time. After collection, store the samples at an ambient temperature of 25-28°C. Before microscopic observation, add 3-4% paraformaldehyde solution as a cell fixative. The ratio of fixative to sample is 1:4, and the fixation time is 15-30 minutes.

[0010] Optionally, in the image acquisition and optimization, the geometric correction adopts a method based on polynomial fitting. For the quadratic polynomial distortion model, the formula is:

[0011] in are the original coordinates, are the corrected coordinates, and is the correction coefficient, and the distortion model is established by collecting images of a known standard grid.

[0012] Optionally, in the image acquisition and optimization, the color correction adopts a white balance algorithm, selects a white reference area in the image, calculates its average RGB value, and then normalizes the RGB value of the entire image, and the formula is:

[0013] in is the original RGB value, is the average RGB value of the white reference area, is the corrected RGB value.

[0014] Optionally, in the image acquisition and optimization, the image denoising adopts the wavelet transform denoising method, selects the db4 wavelet to perform multi-layer decomposition on the image, removes the high-frequency noise component by threshold processing, and then performs wavelet reconstruction, and the threshold adopts a general threshold according to the noise level. Determine, among which, is the noise standard deviation, is the total number of image pixels.

[0015] Optionally, in the cell segmentation and positioning, the semantic segmentation model based on the convolutional neural network adopts U-Net, the network structure includes a contraction path and an expansion path, a large number of labeled rat vaginal exfoliated cell images are used when training the network, and a cross entropy loss function is used:

[0016] in, is the total number of pixels, is the number of cell types, including background, leukocytes, nucleated epithelial cells, and keratinized epithelial cells. is the true label of the category to which the pixel belongs, It is the prediction probability, and the training dataset is expanded using data augmentation techniques, including random rotation, flipping, scaling, and elastic deformation.

[0017] Optionally, in the cell segmentation and positioning, when positioning the segmented cells, the centroid coordinates of each cell are calculated. , the formula is:

[0018] in, is the pixel set of the cell area, are pixel coordinates and assign a unique identifier to each cell.

[0019] Optionally, in the feature extraction and analysis, shape feature extraction includes fractal dimension, which is calculated by box counting method, dividing the cell image into grids of different sizes, and counting the number of grids containing cell pixels. , fractal dimension:

[0020] It also includes extracting the aspect ratio and convex hull area ratio features of cells; On the basis of gray-level co-occurrence matrix and local binary pattern, the feature of directional gradient histogram is added to divide the cell area into multiple small cell units, the gradient direction histogram in each unit is calculated, and then the histograms are connected in series as the texture feature of the cell; A texture analysis method based on wavelet transform is used to extract wavelet coefficients at different scales as texture features to capture the subtle texture changes inside cells. The optical properties of cells under different wavelengths of light were analyzed. Multispectral imaging technology was used to collect cell images at different wavelengths of 400nm-700nm, and the spectral reflectance and absorbance optical characteristics of the cells were extracted. The optical characteristics were reduced and analyzed through principal component analysis.

[0021] The correlation-based feature selection method was used to screen out features with high correlation with cell types. The evaluation indicators of CFS are:

[0022] in, is the number of features, is the average correlation between features and classes, is the average correlation between features.

[0023] Optionally, the cell classification and identification specifically includes the following steps: constructing and training through random forests, determining the number of decision trees [K] through cross-validation experiments, [K] being an integer greater than 100, randomly selecting a feature subset of a size of [L]% of the total number of features from all features, [L] being an integer between 30 and 70, using information gain or Gini index as a splitting criterion, and stopping growth when the decision tree reaches a maximum depth [H], [H] being an integer greater than 5; SVM classifier training, linear kernel SVM uses the sequential minimum optimization algorithm, and the penalty parameter C is determined by cross-validation. The range of the polynomial kernel function is [2,5], and the range of the radial basis kernel function γ is [0.001,10]. The optimal value is determined by grid search and cross-validation. When integrating multiple SVM classifiers, the weight is determined according to the accuracy of each SVM classifier on the validation set; The classification results were optimized by setting the criteria for determining uncertain samples. The uncertain samples were manually labeled and added to the training set. The increase in the number of training set samples each time was not more than [T1]% of the original number of training set samples. [T1] was an integer between 10 and 30. Results with a confidence level lower than [U] were verified a second time. [U] was a value between 0.3 and 0.5. Feature dimensions such as cell grayscale mean, grayscale variance, and edge gradient mean were added. A deep neural network was used for reclassification. The network structure parameters were determined through experiments.

[0024] Optionally, the result verification and feedback specifically include the following steps: The accuracy, recall, F1-score, Matthews correlation coefficient, and area under the receiver operating characteristic curve of the classification results were calculated, and 20% of the samples were selected as the validation set using stratified sampling to ensure that the proportions of the three cell types in the validation set were consistent with the actual distribution of the original samples; If the classification accuracy of a certain cell type is lower than 80%, focus on analyzing the sample characteristics of that cell type, adjust the feature extraction method or feature selection strategy, and retrain the ensemble learning classifier until the classification accuracy of each cell type reaches more than 80%.

[0025] Compared with the prior art, the present invention has at least one of the following beneficial technical effects: The present invention greatly reduces the time cost of manual microscopic examination through an automatic recognition method. The microscopic examination work that traditional methods take one day to complete can be quickly processed by the present application, thereby significantly improving the detection efficiency.

[0026] Further, with the help of a variety of image processing and analysis technologies, as well as integrated learning classifiers, the subjective error of manual judgment is reduced, and the accuracy of cell type identification and estrous cycle judgment is improved. The entire identification process is based on clear algorithms and models, which is not affected by the subjective factors of the operator, ensuring the reliability and consistency of the results. As long as the established process and parameter settings are followed, different operators at different times can obtain similar results, which is convenient for research repetition and verification.

[0027] Extracting cell features from multiple dimensions, including shape, texture, and optical features, and conducting comprehensive analysis can more comprehensively describe cell characteristics and improve classification results. Through active learning strategies and confidence assessment mechanisms, the classifier performance can be continuously optimized to adapt to sample changes and ensure long-term stable detection results. Most of the parameters in the scheme are determined through experiments and can be flexibly adjusted according to actual sample conditions and research needs to adapt to different experimental conditions and data characteristics.

[0028] Through systematic and sophisticated operations, the present invention significantly improves the efficiency and accuracy of rat vaginal exfoliated cell morphology identification. From precise sample collection and image acquisition optimization to multi-dimensional feature extraction and integrated learning classification, and then to result verification and feedback, the scientific nature of the identification is fully guaranteed. It not only greatly shortens the detection time and reduces human errors, but also continuously improves performance through dynamic optimization, providing efficient and reliable technical support for the study of the rat estrous cycle. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 The figure is a flow chart of the method for determining the estrous cycle of rats based on image recognition. DETAILED DESCRIPTION

[0030] The technical solution of the present disclosure will be clearly and completely described below in conjunction with specific embodiments. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present disclosure.

[0031] like Figure 1 As shown, the method for determining the rat estrous cycle based on image recognition proposed in the present invention includes sample collection and preparation, image collection and optimization, cell segmentation and positioning, feature extraction and analysis, cell classification and identification, result verification and feedback. Each step is described in detail below.

[0032] 1. Sample collection and preparation Sample collection should be carried out strictly in accordance with the given method for collecting rat vaginal exfoliated cells to ensure the independence and accuracy of each rat sample. During the collection process, pay attention to the gentleness of the operation to avoid unnecessary damage to the cells. Use a high-precision pipette and sterile saline to ensure that the amount of liquid drawn and injected each time is accurate to ensure the consistency of the sample.

[0033] Sample processing and storage: After collection, the samples should be immediately stored in a 48-well plate at a stable ambient temperature of 25-28°C to prevent cell morphology from changing in a short period of time. Before microscopic observation, add an appropriate amount of cell fixative 4% paraformaldehyde solution, and control the ratio of fixative to sample at 1:4 to ensure that the cells maintain their original morphology during subsequent processing. The fixation time is set to 15-30 minutes.

[0034] 2. Image Acquisition and Optimization For high-resolution image acquisition, use a professional inverted microscope equipped with a high-resolution, high-sensitivity digital camera for image acquisition. The magnification of the microscope can be selected between 100x-400x according to the cell size and morphological characteristics. Adjust the light intensity and angle of the microscope to obtain a uniform and moderately contrasted lighting effect. The light intensity can be adjusted between 50%-80%, and the best lighting conditions are found through multiple experiments.

[0035] Image preprocessing, geometric correction: Due to the possible distortion of the microscope lens, the collected images are geometrically corrected. A polynomial fitting-based method is used to collect images of a known standard grid and establish a distortion model. For the quadratic polynomial distortion model, the formula is:

[0036] in are the original coordinates, are the corrected coordinates, and is the correction factor.

[0037] Color correction: In order to eliminate the color deviation caused by uneven lighting, a white balance algorithm is used for color correction. Select a white reference area in the image (such as a blank background area), calculate its average RGB value, and then normalize the RGB value of the entire image. The formula is:

[0038] in is the original RGB value, is the average RGB value of the white reference area, is the corrected RGB value.

[0039] In this embodiment, image denoising is performed: in addition to median filtering, wavelet transform denoising method can also be used. Select the appropriate wavelet basis function db4 wavelet, perform multi-layer decomposition on the image, remove high-frequency noise components through threshold processing, and then perform wavelet reconstruction. The threshold adopts a universal threshold according to the noise level. Determine, among which, is the noise standard deviation, is the total number of image pixels.

[0040] 3. Cell segmentation and localization Based on deep learning, the semantic segmentation model (such as U-Net) based on convolutional neural network (CNN) is used to segment cells. The network structure of U-Net includes a contraction path (for feature extraction) and an expansion path (for upsampling and feature fusion). When training the network, a large number of labeled rat vaginal exfoliated cell images are used for training. The cross entropy loss function is used:

[0041] in, is the total number of pixels, is the number of cell types, including background, leukocytes, nucleated epithelial cells, and keratinized epithelial cells. is the true label of the category to which the pixel belongs, is the predicted probability.

[0042] In order to improve the segmentation accuracy, data augmentation techniques such as random rotation, flipping, scaling and elastic deformation are used to expand the training dataset.

[0043] Cell positioning and labeling: accurately locate and label the segmented cells. Calculate the centroid coordinates of each cell , the formula is:

[0044] in, is the pixel set of the cell area, is the pixel coordinate, which assigns a unique identifier to each cell to facilitate subsequent feature extraction and classification.

[0045] 4. Feature Extraction and Analysis Multi-dimensional feature extraction Shape features, in addition to basic area, perimeter and circularity, also extract the shape complexity of cells (such as fractal dimension). The fractal dimension is calculated using the box counting method, dividing the cell image into grids of different sizes and counting the number of grids containing cell pixels. , fractal dimension:

[0046] The aspect ratio and convex hull area ratio features of the cells are extracted to further describe the shape characteristics of the cells.

[0047] Texture features, based on gray-level co-occurrence matrix (GLCM) and local binary pattern (LBP), add directional gradient histogram (HOG) features. The cell area is divided into multiple small cell units, the gradient direction histogram in each unit is calculated, and then these histograms are connected in series as the texture features of the cell. The texture analysis method based on wavelet transform is adopted to extract wavelet coefficients at different scales as texture features to capture the subtle texture changes inside the cell.

[0048] Optical characteristics: Analyze the optical characteristics of cells under different wavelengths of light. Use multispectral imaging technology to collect cell images at different wavelengths of 400nm-700nm, and extract the spectral reflectance and absorbance optical characteristics of cells. Use principal component analysis (PCA) and other methods to reduce the dimension and analyze the optical characteristics.

[0049] Feature selection and dimensionality reduction, using correlation-based feature selection (CFS) to screen out features with high correlation with cell types. The evaluation indicators of CFS are:

[0050] in, is the number of features, is the average correlation between features and classes, is the average correlation between features. For high-dimensional feature data, linear discriminant analysis (LDA) or principal component analysis (PCA) is used for dimensionality reduction. PCA calculates the eigenvalues ​​and eigenvectors of the covariance matrix, projects the original features into a new low-dimensional space, and retains the variance information of the data.

[0051] 5. Cell classification and identification 1. Ensemble learning classifier Random forest construction and training, determine the number of decision trees in the random forest to be [K] trees, where [K] is an integer greater than 100, and determine the optimal number through cross-validation experiments to ensure a balance between integration effect and computational efficiency. For each decision tree, randomly select a feature subset from all features, and the size of the feature subset is [L]% of the total number of features, where [L] is an integer between 30 and 70. The optimal ratio is determined through experiments to increase the diversity of decision trees. When training each decision tree, use information gain or Gini index as the splitting criterion. The information gain calculation formula is:

[0052] in, is a dataset, It is a feature, It is a feature The number of values ​​of , It is a feature The value is A subset of samples,

[0053] It is the first The proportion of class samples; the Gini index calculation formula is:

[0055] The optimal split features and split points are selected through these criteria to build a decision tree.

[0056] When the decision tree reaches the maximum depth [H], it stops growing, where [H] is an integer greater than 5. The appropriate depth is determined through experiments to prevent overfitting.

[0057] SVM classifier training and integration, for SVM with linear kernel function, the optimization algorithm used in training is the sequential minimum optimization (SMO) algorithm, which determines the model parameters by solving the following optimization problem:

[0060] in, is the sample size, is the Lagrange multiplier, is the category label of the sample, is the kernel function. For the linear kernel , is the penalty parameter, whose value is determined by cross-validation.

[0061] For the polynomial kernel function ,in is the degree of the polynomial, and its value range is [2,5]. The optimal degree is determined through experiments to adapt to different data distributions.

[0062] For the radial basis kernel function , The value range is [0.001, 10], and the optimal value is determined through grid search and cross-validation to control the complexity and generalization ability of the model.

[0063] When integrating multiple SVM classifiers, a weighted average method is used. The weight is determined according to the accuracy of each SVM classifier on the validation set. The calculation formula is:

[0064] in, is the number of SVM classifiers. The weights are adjusted dynamically in this way to optimize the performance of the integrated classifier.

[0065] 2. Optimization of classification results The active learning strategy is implemented, and the criteria for determining samples with uncertain classification are set as follows: for the random forest classifier, when the proportion of votes for the highest voted category in the voting results of all decision trees is less than [R]%, the sample is determined to be classified as uncertain, where [R] is a value between 50 and 70, determined through experiments; for the integrated SVM classifier, when the highest confidence of a sample in the prediction results of each SVM classifier is less than [S], the sample is determined to be classified as uncertain, where [S] is a value between 0.5 and 0.7, determined through experiments.

[0066] Manual labeling is performed on samples with uncertain classification. The labelers need to have professional knowledge of cell morphology. After labeling, these samples are added to the training set and the ensemble learning classifier is retrained. Each time the training is retrained, the increase in the number of training set samples does not exceed [T1]% of the original number of training set samples, where [T1] is an integer between 10 and 30, to avoid overfitting the model to the newly added samples and to ensure that the model can continuously learn new information.

[0067] The confidence evaluation mechanism is established and verified twice. For the random forest classifier, the confidence calculation formula is, where is the number of votes for the category with the most votes in all decision trees, and is the total number of votes in all decision trees. For the integrated SVM classifier, the confidence is the average confidence of each SVM classifier in the prediction result of the sample.

[0068] When the confidence is less than [U], it is determined as a classification result with low confidence, where [U] is a value between 0.3 and 0.5, which is determined through experiments.

[0069] A secondary verification was performed on the classification results with lower confidence. The feature dimensions added during the secondary verification included the grayscale mean, grayscale variance, edge gradient mean, etc. of the cells. The specific calculation formulas were: grayscale mean, where is the grayscale value of the pixels in the cell area, and is the number of pixels in the cell area; grayscale variance; edge gradient mean, where is the set of cell edge pixels, and is the gradient value of the edge pixels.

[0070] Use more complex classification algorithms for reclassification, such as deep neural networks (DNN). The network structure of DNN includes an input layer, multiple hidden layers and an output layer. The number of hidden layers is [V1] layers, where [V1] is an integer between 3 and 5, determined by experiments. The number of neurons in each layer is [W1], where [W1] is an integer between 64 and 256, determined by experiments. The activation function uses the ReLU function. The stochastic gradient descent (SGD) algorithm is used for training. The initial value of the learning rate is [X1], and the value range is [0.001, 0.1]. It is determined by experiments that the learning rate decays to [Z1] times the original value after [Y1] epochs, where [Y1] is an integer between 10 and 20, and [Z1] is a decimal between 0.5 and 0.9, determined by experiments, to improve the accuracy of the secondary verification.

[0071] 6. Result Verification and Feedback Multi-index result verification Calculate the accuracy, recall, and F1-score of the classification results, and calculate the Matthews correlation coefficient (MCC) at the same time. The formula is:

[0072] in, represents the number of true positive samples, represents the number of true negative samples, represents the number of false positive samples, Indicates the number of false negative samples. In addition, the area under the receiver operating characteristic curve (ROC) (AUC) was calculated to comprehensively evaluate the accuracy of the classification results through these indicators.

[0073] Stratified sampling was used to extract 20% of all samples as the validation set. The proportions of the three cell types, white blood cells, nucleated epithelial cells, and keratinized epithelial cells in the validation set were ensured to be consistent with the actual distribution ratios in the original samples, so as to avoid bias in the evaluation results due to data imbalance.

[0074] Feedback and Optimization If the classification accuracy of a certain cell type is found to be lower than 80% during the verification process, the sample features corresponding to the cell type will be analyzed in detail to check whether there is insufficient feature extraction or unreasonable feature selection.

[0075] According to the analysis results, the feature extraction method or feature selection strategy is adjusted in a targeted manner. If it is found that the shape feature of a certain type of cell is not highly distinguishable, a new shape descriptor is introduced; if it is found that some features are redundant, feature screening is performed again. After that, the ensemble learning classifier is retrained and verified again until the classification accuracy of each cell type reaches more than 80%.

[0076] The present invention greatly reduces the time cost of manual microscopy through an automatic recognition method. The traditional method requires one day to complete the microscopy work, which can be quickly processed by this solution, significantly improving the detection efficiency. With the help of a variety of image processing and analysis technologies, as well as integrated learning classifiers, the subjective error of manual judgment is reduced, and the accuracy of cell type recognition and estrous cycle judgment is improved. The entire recognition process is based on clear algorithms and models, is not affected by the subjective factors of the operator, and ensures the reliability and consistency of the results. As long as the established process and parameter settings are followed, similar results can be obtained at different times and by different operators, which is convenient for research repetition and verification.

[0077] It is worth noting that extracting cell features from multiple dimensions, including shape, texture, and optical features, and conducting comprehensive analysis can more comprehensively describe cell characteristics and improve classification results. Through active learning strategies and confidence assessment mechanisms, the classifier performance can be continuously optimized to adapt to sample changes and ensure long-term stable detection results. Most of the parameters in the scheme are determined through experiments and can be flexibly adjusted according to actual sample conditions and research needs to adapt to different experimental conditions and data characteristics.

[0078] The above specific embodiments are only several optional embodiments of the present invention. Based on the technical solutions of the present invention and the relevant inspirations of the above embodiments, those skilled in the art can make various alternative improvements and combinations to the above specific embodiments.

Claims

1. A method for determining the estrous cycle of rats based on image recognition, characterized in that: The following steps are involved: Sample collection and preparation, collection of rat vaginal exfoliated cell samples, processing and preservation; Image acquisition and optimization, acquisition of high-resolution images, and geometric correction, color correction and image denoising preprocessing; Cell segmentation and positioning: Use a semantic segmentation model based on a convolutional neural network to segment cells, and accurately locate and mark the segmented cells; Feature extraction and analysis: extract the shape, texture and optical features of cells, and perform feature selection and dimensionality reduction; Cell classification and identification: Use ensemble learning classifiers to classify cells and optimize the classification results; Result verification and feedback: Verify the classification results through multiple indicators, and provide feedback and optimization for each link based on the verification results.

2. The method for determining the estrous cycle of rats based on image recognition according to claim 1, characterized in that: When collecting rat vaginal exfoliated cell samples, a high-precision pipette was used, and the volume of sterile saline aspirated and injected each time was 160-260 μL. After collection, the samples were stored at an ambient temperature of 25-28°C. Before microscopic observation, 3-4% paraformaldehyde solution was added as a cell fixative. The ratio of fixative to sample was 1:4, and the fixation time was 15-30 minutes.

3. The method for determining the estrous cycle of rats based on image recognition according to claim 1, characterized in that: In the image acquisition and optimization, the geometric correction adopts a method based on polynomial fitting. For the quadratic polynomial distortion model, the formula is: ,in are the original coordinates, are the corrected coordinates, and is the correction coefficient, and the distortion model is established by collecting images of a known standard grid.

4. The method for determining the estrous cycle of rats based on image recognition according to claim 1, characterized in that: In the image acquisition and optimization, the color correction adopts the white balance algorithm, selects the white reference area in the image, calculates its average RGB value, and then normalizes the RGB value of the entire image. The formula is: ,in is the original RGB value, is the average RGB value of the white reference area, is the corrected RGB value.

5. The method for determining the estrous cycle of rats based on image recognition according to claim 1, characterized in that: In the image acquisition and optimization, the image denoising adopts the wavelet transform denoising method, selects the db4 wavelet to perform multi-layer decomposition on the image, removes the high-frequency noise component through threshold processing, and then performs wavelet reconstruction. The threshold adopts the general threshold according to the noise level. Determine, among which, is the noise standard deviation, is the total number of image pixels.

6. The method for determining the estrous cycle of rats based on image recognition according to claim 1, characterized in that: In the cell segmentation and positioning, the semantic segmentation model based on the convolutional neural network adopts U-Net, and the network structure includes a contraction path and an expansion path. A large number of labeled rat vaginal exfoliated cell images are used when training the network, and the cross entropy loss function is adopted: ,in, is the total number of pixels, is the number of cell types, including background, leukocytes, nucleated epithelial cells, and keratinized epithelial cells. is the true label of the category to which the pixel belongs, It is the prediction probability, and the training dataset is expanded using data augmentation techniques, including random rotation, flipping, scaling, and elastic deformation.

7. The method for determining the estrous cycle of rats based on image recognition according to claim 6, characterized in that: In the cell segmentation and positioning, when positioning the segmented cells, the centroid coordinates of each cell are calculated. , the formula is: ,in, is the pixel set of the cell area, are pixel coordinates and assign a unique identifier to each cell.

8. The method for determining the estrous cycle of rats based on image recognition according to claim 1, characterized in that: In the feature extraction and analysis, shape feature extraction includes fractal dimension, which is calculated by box counting method, dividing the cell image into grids of different sizes, and counting the number of grids containing cell pixels. , fractal dimension: , and also includes extracting the aspect ratio and convex hull area ratio features of cells; On the basis of gray-level co-occurrence matrix and local binary pattern, the feature of directional gradient histogram is added to divide the cell area into multiple small cell units, the gradient direction histogram in each unit is calculated, and then the histograms are connected in series as the texture feature of the cell; A texture analysis method based on wavelet transform is used to extract wavelet coefficients at different scales as texture features to capture the subtle texture changes inside cells. Analyze the optical characteristics of cells under different wavelengths of light. Use multispectral imaging technology to collect cell images at different wavelengths of 400nm-700nm, extract the spectral reflectance and absorption optical characteristics of cells, and reduce and analyze the optical characteristics through principal component analysis. The correlation-based feature selection method was used to screen out features with high correlation with cell types. The evaluation indicators of CFS are: ,in, is the number of features, is the average correlation between features and classes, is the average correlation between features.

9. The method for determining the estrous cycle of rats based on image recognition according to claim 1, characterized in that: The cell classification and identification specifically includes the following steps: constructing and training a random forest, determining the number of decision trees [K] through a cross-validation experiment, [K] being an integer greater than 100, randomly selecting a feature subset of a size of [L]% of the total number of features from all features, [L] being an integer between 30 and 70, using information gain or Gini index as a splitting criterion, and stopping growth when the decision tree reaches a maximum depth [H], [H] being an integer greater than 5; SVM classifier training, linear kernel SVM uses the sequential minimum optimization algorithm, and the penalty parameter C is determined by cross-validation. The range of the polynomial kernel function is [2,5], and the range of the radial basis kernel function γ is [0.001,10]. The optimal value is determined by grid search and cross-validation. When integrating multiple SVM classifiers, the weight is determined according to the accuracy of each SVM classifier on the validation set; The classification results were optimized by setting the criteria for determining uncertain samples. The uncertain samples were manually labeled and added to the training set. The increase in the number of training set samples each time during retraining did not exceed [T1]% of the original number of training set samples. [T1] was an integer between 10 and 30. Results with a confidence level lower than [U] were verified twice. [U] was a value between 0.3 and 0.

5. The feature dimensions of cell grayscale mean, grayscale variance, and edge gradient mean were increased. A deep neural network was used for reclassification. The network structure parameters were determined through experiments.

10. The method for determining the estrous cycle of rats based on image recognition according to claim 1, characterized in that: The result verification and feedback specifically include the following steps: The accuracy, recall, F1-score, Matthews correlation coefficient, and area under the receiver operating characteristic curve of the classification results were calculated, and 20% of the samples were selected as the validation set using stratified sampling to ensure that the proportions of the three cell types in the validation set were consistent with the actual distribution of the original samples; If the classification accuracy of a certain cell type is lower than 80%, focus on analyzing the sample characteristics of that cell type, adjust the feature extraction method or feature selection strategy, and retrain the ensemble learning classifier until the classification accuracy of each cell type reaches above 80%.

Images