Rat ovarian growth follicles and corpus luteum grading and counting method

Abstract

A grading and counting method for follicles and corpus luteum growing in ovaries of rats relates to the technical field of image analysis, and comprises the following steps: scanning ovary staining sections; identifying needed ovarian tissues and converting the needed ovarian tissues into a first region of interest ROI1; screening out growing follicles and corpus luteum meeting requirements on ROI1; and identifying and marking the screened growing follicles meeting the requirements as primary follicles, secondary follicles, tertiary follicles and follicles which cannot be graded, identifying and marking the screened corpus luteum meeting the requirements as newborn corpus luteum, recent corpus luteum, old corpus luteum and corpus luteum which cannot be graded, and finally counting to obtain the number of the follicles and corpus luteum of each level. According to the invention, artificial intelligence of classification and counting of ovarian growth follicles and corpus luteum of rats can be realized, and rapid and accurate batch operation is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image analysis, in particular to a method for grading and counting growing follicles and corpus luteum of rat ovary. BACKGROUND

[0002] At present, the artificial grading method for growing follicles and corpus luteum of rat ovary is that a pathological diagnostician observes under a microscope, grades each growing follicle and corpus luteum according to morphological characteristics of the growing follicle and corpus luteum of rat ovary, which is of huge workload and requires high experience of the diagnostician, and is not suitable for batch operation. SUMMARY

[0003] In view of the above problems, the present application aims to provide a method for grading and counting growing follicles and corpus luteum of rat ovary, so as to achieve the purpose of grading and counting growing follicles and corpus luteum of rat ovary on HE-stained sections.

[0004] The technical scheme adopted by the present application to achieve the purpose of the present application is a method for grading and counting growing follicles and corpus luteum of rat ovary, comprising the following steps:

[0005] S1, section scanning: scanning a rat ovary HE-stained section using a digital section scanner;

[0006] S2, ovary tissue identification: identifying the required ovary tissue on the stained section and converting it into a first region of interest ROI1, and counting the area of the ovary tissue;

[0007] S3, growing follicle and corpus luteum identification and segmentation: segmenting independent growing follicles and corpus luteum on the first region of interest ROI1, screening out growing follicles and corpus luteum meeting the requirements and making corresponding marks, and counting the number of growing follicles and corpus luteum meeting the requirements;

[0008] S4, identification and counting of growing follicles and corpus luteum at each level: identifying the screened growing follicles meeting the requirements as primary follicles, secondary follicles, tertiary follicles and unclassifiable follicles, identifying the screened corpus luteum meeting the requirements as neonatal corpus luteum, recent corpus luteum, old corpus luteum and unclassifiable corpus luteum, and finally counting the number of follicles at each level and the number of corpus luteum at each level.

[0009] Further, the specific steps of the step S2, ovary tissue identification, are as follows:

[0010] S2.1, select a certain number of rat ovary HE staining sections as a training set, draw the region of interest on the staining sections of the training set, including ovarian tissue area and non-ovarian area, mark all ovarian tissue areas in the region of interest, for example, mark them as blue, mark the non-ovarian area as background, for example, mark it as green, then apply the deep learning U-Net model, i.e. Deep Learning-U-Net model, for training, and adjust it during training until the accuracy of ovarian tissue recognition reaches more than 90%;

[0011] S2.2, the staining sections that need to be formally recognized as ovarian tissue are introduced into the software that has completed the model training as described in step S2.1, the initial region of interest ROI0 is drawn, and the ovarian tissue recognition is performed in the initial region of interest ROI0, the recognized ovarian tissue area is automatically marked, and the non-ovarian area is automatically marked as background;

[0012] S2.3, further post-processing, the post-processing step includes:

[0013] S2.3.1, fill in the blank area in the ovarian tissue;

[0014] S2.3.2, replace the ovarian tissue with an area less than a preset value S1 with background;

[0015] S2.3.3, clear the background, for example, the green background, so that the staining section only retains the ovarian tissue;

[0016] S2.3.4, convert the retained ovarian tissue into the first region of interest ROI1;

[0017] S2.4, finally, the area of the ovarian tissue in the first region of interest ROI1 is calculated.

[0018] Further, the preset value S1 is equal to 10000 μm 2 .

[0019] Further, the specific steps of the step S3 of follicle and corpus luteum recognition and segmentation are:

[0020] S3.1, select a certain number of rat ovary HE staining sections as a training set, draw the region of interest on the staining sections of the training set, and mark all growing follicles and corpus luteum in the region of interest respectively, for example: growing follicles are marked in red, corpus luteum is marked in blue, non-growing follicles and non-corpus luteum region are marked as background such as green, all growing follicles and corpus luteum are added with boundaries, for example, yellow boundaries are added, so as to separate growing follicles and corpus luteum into independent individuals, after marking, a deep learning semantic segmentation model, namely Deep Learning-DeepLabv3+ model, is used for training, and the model is adjusted during training until the recognition and segmentation accuracy of growing follicles and corpus luteum reaches more than 90%;

[0021] S3.2, the first region of interest ROI1 which needs to be formally recognized and segmented for growing follicles and corpus luteum is introduced into the software in which the model training in step S3.1 is completed, and the growing follicles and corpus luteum are recognized and segmented, and the recognized growing follicles, corpus luteum and boundaries are automatically marked, for example, growing follicles are marked in red, corpus luteum is marked in blue, and boundaries are marked in yellow, and non-growing follicles and non-corpus luteum region are automatically marked as background such as green;

[0022] S3.3, then post-processing is performed, and the post-processing step includes:

[0023] S3.3.1, fill in the blank area in the corpus luteum tissue;

[0024] S3.3.2, fill in the blank area in the follicle tissue;

[0025] S3.3.3, remove the boundaries such as yellow boundaries;

[0026] S3.3.4, remove the corpus luteum with an area less than a preset value S2;

[0027] S3.3.5, remove the growing follicles with an area less than a preset value S3;

[0028] S3.3.6, remove the background such as green background, so that the staining section only retains growing follicles such as red and corpus luteum such as blue;

[0029] S3.4, finally, the number of growing follicles and the number of corpus luteum retained in the first region of interest ROI1 are counted.

[0030] Further, the preset value S2 is equal to 10000 μm 2 , and the preset value S3 is equal to 1000 μm 2 .

[0031] Further, the specific steps of step S4 for recognizing and counting growing follicles and corpus luteum at each level are:

[0032] S4.1, a certain number of rat ovary HE-stained sections are selected as a training set, the regions of interest are drawn on the stained sections of the training set, step S3 is run in the regions of interest, the growing follicles screened out in step S3 are divided and marked as primary follicles, secondary follicles, tertiary follicles, and the remaining follicles are marked as unclassifiable follicles; the corpus luteum screened out in step S3 is divided and marked as neonatal corpus luteum, recent corpus luteum, and old corpus luteum, and the remaining corpus luteum is marked as unclassifiable corpus luteum; after marking, a deep learning semantic segmentation model, i.e., a Deep Learning-DeepLabv3+ model, is used for training, and after training, reference features of the growing follicles at each level and reference features of the corpus luteum at each level are obtained, and the reference features are adjusted in the training until the recognition accuracy of the growing follicles and the corpus luteum at each level reaches more than 90%;

[0033] S4.2, the growing follicles and the corpus luteum screened out in step S3, which need to be formally recognized at each level, are introduced into the software in which the model training in step S4.1 is completed, the actual features of each growing follicle are compared with the reference features of the growing follicles at each level to calculate the feature threshold of each growing follicle corresponding to the growing follicles at each level, and the actual features of each corpus luteum are compared with the reference features of the corpus luteum at each level to calculate the feature threshold of each corpus luteum corresponding to the corpus luteum at each level;

[0034] S4.3, post-processing is further performed, and the post-processing step includes:

[0035] S4.3.1, in the corpus luteum screened out in step S3, the corpus luteum with a recent corpus luteum feature threshold greater than a preset value M1 is replaced by a recent corpus luteum;

[0036] S4.3.2, the corpus luteum with a neonatal corpus luteum feature threshold greater than a preset value M2 is replaced by a neonatal corpus luteum;

[0037] S4.3.3, the corpus luteum with an old corpus luteum feature threshold greater than a preset value M3 is replaced by an old corpus luteum;

[0038] S4.3.4, the growing follicle with a primary follicle feature threshold greater than a preset value M4 is replaced by a primary follicle;

[0039] S4.3.5, the growing follicle with a secondary follicle feature threshold greater than a preset value M5 is replaced by a secondary follicle;

[0040] S4.3.6, the growing follicle with a tertiary follicle feature threshold greater than a preset value M6 is replaced by a tertiary follicle;

[0041] S4.3.7, the remaining corpus luteum is marked as unclassifiable corpus luteum, and the remaining follicle is marked as unclassifiable follicle;

[0042] S4.4, finally statistics of the number of each level of follicle and each level of corpus luteum.

[0043] Further, the preset values are M1=0.3, M2=0.3, M3=0.3, M4=0.3, M5=0.3, and M6=0.3.

[0044] The present application has the following advantages:

[0045] The present application can realize artificial intelligence of grading and counting of growing follicles and corpus luteum of rat ovary, and realizes rapid and accurate batch operation. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 It is a total flowchart of the embodiment of the present application;

[0047] Figure 2 It is an original picture without color marking;

[0048] Figure 3 It is an effect picture after step S2 of the embodiment of the present application;

[0049] Figure 4 It is an effect picture after step S3 of the embodiment of the present application;

[0050] Figure 5 It is an effect picture after step S4 of the embodiment of the present application. DETAILED DESCRIPTION

[0051] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0052] Figure 2 One specific embodiment of the method for grading and counting growing follicles and corpus luteum of rat ovary is shown, and the steps thereof include:

[0053] S1, slice scanning: using a digital slice scanner to scan a rat ovary HE staining slice, and importing the digital slice into Visiopharm software;

[0054] S2, ovary tissue identification: identifying the required ovary tissue on the staining slice and converting it into a first region of interest ROI1, and counting the area of the ovary tissue;

[0055] S3, growing follicle and corpus luteum identification and segmentation: segmenting independent growing follicles and corpus luteum on the first region of interest ROI1, screening the required growing follicles and corpus luteum and making corresponding marks, and counting the number of the required growing follicles and corpus luteum;

[0056] S4, identification and counting of each level of growing follicle and corpus luteum: the identified qualified growing follicle is marked as primary follicle, secondary follicle, tertiary follicle, and unclassifiable follicle, and the identified qualified corpus luteum is marked as neonatal corpus luteum, recent corpus luteum, old corpus luteum, and unclassifiable corpus luteum, and finally the number of each level of follicle and corpus luteum is counted.

[0057] In this embodiment, the specific steps of the step S2 of identifying the ovarian tissue are as follows:

[0058] S2.1, a certain number of HE-stained sections of rat ovaries are selected as a training set, the region of interest is drawn on the stained sections of the training set, including the ovarian tissue region and the non-ovarian region, all ovarian tissue regions in the region of interest are marked as blue (class2), and the non-ovarian region is marked as green (class1), the identification of the ovarian and non-ovarian regions is completed by a pathologist, then a deep learning U-Net model (Deep Learning-U-Net model) is applied for training, and the model is adjusted during the training until the accuracy of ovarian tissue identification reaches more than 90%;

[0059] S2.2, the stained sections that need to be formally identified for ovarian tissue are introduced into the software in which the model training in step S2.1 is completed, the initial region of interest ROI0 is drawn, and the ovarian tissue identification is performed in the initial region of interest ROI0, the identified ovarian tissue region is automatically marked as blue (class2), and the non-ovarian region is automatically marked as green (class1);

[0060] S2.3, the post processing program is run again, and the post processing steps include:

[0061] S2.3.1, the parameter Fill Holes Lbl: class2 is set, that is, the blank area in the ovarian tissue is filled completely;

[0062] S2.3.2, the parameter change by shape Lbl: class2 with an area less than 10000 μm 2 is replaced by class1, that is, the blue ovarian tissue with an area less than 10000 μm 2 is replaced by the green background;

[0063] S2.3.3, the parameter change Lbl: class1 is replaced by clear, that is, the green background is removed, so that the stained section only retains the blue ovarian tissue;

[0064] S2.3.4, setting parameter outline as ROI Lbl: output class2 as ROI1, i.e. converting the reserved blue ovarian tissue into the first region of interest ROI1;

[0065] S2.4, finally selecting class2 in the output variable output variables, and after running, the area of the blue ovarian tissue in the first region of interest ROI1 can be obtained.

[0066] Subsequent processing is performed within this ROI1 range to narrow the range of subsequent program processing and shorten the processing time.

[0067] In this embodiment, the specific steps of the step S3 of recognizing and segmenting the growing follicle and corpus luteum are as follows:

[0068] S3.1, a certain number of rat ovarian HE staining sections are selected as a training set, the region of interest is drawn on the staining sections of the training set, all growing follicles in the region of interest are marked as red (class3), the corpus luteum is marked as blue (class2), and the non-growing follicle and non-corpus luteum region is marked as green (class1), a dilate boundary algorithm is called to add a yellow boundary (class4) to all growing follicles and corpus luteum, so as to segment the growing follicle and corpus luteum into independent individuals, after marking, a deep learning semantic segmentation model, i.e. Deep Learning-DeepLabv3+ model, is applied for training, and the model is adjusted during training until the recognition and segmentation accuracy of the growing follicle and corpus luteum reaches more than 90%;

[0069] S3.2, the first region of interest ROI1 which needs to be formally recognized and segmented for the growing follicle and corpus luteum is imported into the software in which the model training in step S3.1 is completed, and the growing follicle and corpus luteum are recognized and segmented, the recognized growing follicle is automatically marked as red (class3), the corpus luteum is automatically marked as blue (class2), the boundary is automatically marked as yellow (class4), and the non-growing follicle and non-corpus luteum region is automatically marked as green (class1);

[0070] S3.3, after running the post processing program, the post processing steps include:

[0071] S3.3.1, setting parameter Fill Holes Lbl: class2, i.e. filling the blank area in the blue corpus luteum tissue completely;

[0072] S3.3.2, setting parameter Fill Holes Lbl: class3, i.e. filling the blank area in the red follicle tissue completely;

[0073] S3.3.3 Setting parameter changeLbl: Replace class4 with clear, which will clear the yellow border;

[0074] S3.3.4, Set the parameter change by shape Lbl: This will change the area to less than 10000μm. 2 Replace class2 with clear, which means the area is less than 10000μm. 2 Blue corpus luteum removal;

[0075] S3.3.5, Set parameter change by shape Lbl: for areas smaller than 1000μm 2 Replace class3 with clear, which means the area is less than 1000μm. 2 Red growing follicles were removed;

[0076] S3.3.6. Set parameter change Lbl: Replace class1 with clear, which will remove the green background so that the stained section only retains red growing follicles and blue corpora lutea; subsequent steps will be performed on the red growing follicles and blue corpora lutea.

[0077] S3.4 Finally, select class2 and class3 in the output variables. After running, you can obtain the number of red growing follicles and blue corpus luteum retained in the first region of interest (ROI1).

[0078] In this embodiment, the specific steps for identifying and counting follicles and corpora lutea at each stage in step S4 are as follows:

[0079] S4.1 Select a certain number of HE-stained sections of rat ovaries as the training set. Draw the region of interest on the stained sections of the training set. Run step S3 within the region of interest. Classify and label the growing follicles selected in step S3 as primary follicles (class7), secondary follicles (class8), and tertiary follicles (class9). Label the remaining follicles as follicles that cannot be graded (class3). Classify and label the corpus luteum selected in step S3 as neonatal corpus luteum (class4), recent corpus luteum (class5), and old corpus luteum (class6). Label the remaining corpus luteum as corpus luteum that cannot be graded (class2). After labeling, use a deep learning semantic segmentation model, namely Deep Learning-DeepLabv3+, for training. After training, obtain the reference features of each level of growing follicles and the reference features of each level of corpus luteum. Adjust them during training until the recognition accuracy of each level of growing follicles and corpus luteum reaches more than 90%.

[0080] S4.2, the growing follicles and corpora lutea screened out in step S3 which need to be formally identified are introduced into the software completed in step S4.1, and the feature threshold of each growing follicle corresponding to each growing follicle is obtained by comparing and calculating the actual feature of each growing follicle with the reference feature of each growing follicle; the feature threshold of each corpus luteum corresponding to each corpus luteum is obtained by comparing and calculating the actual feature of each corpus luteum with the reference feature of each corpus luteum;

[0081] S4.3, the post processing program is run again, and the following post processing steps are performed in the red growing follicles and blue corpora lutea screened out in step S3:

[0082] S4.3.1, the parameter change by intensity Lbl is set: the corpus luteum with a recent corpus luteum feature threshold greater than 0.3 is replaced by a recent corpus luteum (class 5);

[0083] S4.3.2, the parameter change by intensity Lbl is set: the corpus luteum with a neonatal corpus luteum feature threshold greater than 0.3 is replaced by a neonatal corpus luteum (class 4);

[0084] S4.3.3, the parameter change by intensity Lbl is set: the corpus luteum with a senile corpus luteum feature threshold greater than 0.3 is replaced by a senile corpus luteum (class 6);

[0085] S4.3.4, the parameter change by intensity Lbl is set: the growing follicle with a primary follicle feature threshold greater than 0.3 is replaced by a primary follicle (class 7);

[0086] S4.3.5, the parameter change by intensity Lbl is set: the growing follicle with a secondary follicle feature threshold greater than 0.3 is replaced by a secondary follicle (class 8);

[0087] S4.3.6, the parameter change by intensity Lbl is set: the growing follicle with a tertiary follicle feature threshold greater than 0.3 is replaced by a tertiary follicle (class 9);

[0088] S4.3.7, the remaining corpora lutea are marked as unclassifiable corpora lutea (class 2), and the remaining follicles are marked as unclassifiable follicles (class 3);

[0089] The above steps are sequentially performed, and S4.3.1 screens out luteums with recent luteum characteristics in all luteums, S4.3.2 screens out luteums with neonatal luteum characteristics in the remaining luteums, and so on until luteums and growing follicles of all stages are screened and classified.

[0090] S4.4, finally, selects class4, class5, class6, class7, class8, class9 in output variables, and after running, the number of luteums and growing follicles of each stage can be obtained.

[0091] The embodiment of the present application is realized based on the market existing AI-based image analysis software Visiopharm software platform, and other software platforms with similar functions can also be realized.

[0092] The above embodiment of the present application is only an example for illustrating the present application, and is not a limitation on the implementation of the present application. For ordinary skilled in the art, other different forms of changes and variations can be made on the basis of the above description. Here, all the implementation ways cannot be exhausted. Any obvious changes or variations derived from the technical solutions of the present application still fall within the protection scope of the present application.

Claims

1. A method for staging and counting growing follicles and corpora lutea in rat ovaries, characterized by, Comprise the following steps: S1, slice scanning: using digital slice scanner to scan rat ovary HE staining section; S2, ovary tissue identification: identify the required ovary tissue on the staining section and convert it into the first region of interest ROI1, and count the area of the ovary tissue; S3, growing follicle and corpus luteum identification and segmentation: segmenting the independent growing follicle and corpus luteum on the first region of interest ROI1, screening out the required growing follicle and corpus luteum and making corresponding mark, and counting the number of the required growing follicle and corpus luteum; S4, identification and counting of growing follicles and corpus luteum at all levels: the screened growing follicles are identified and marked as primary follicles, secondary follicles, tertiary follicles and unclassifiable follicles, and the screened corpus luteum is identified and marked as newborn corpus luteum, recent corpus luteum, old corpus luteum and unclassifiable corpus luteum, and finally the number of follicles and corpus luteum at all levels is counted.

2. The method for grading and counting the growing follicle and corpus luteum of rat ovary according to claim 1, characterized in that, The specific steps of the step S2 ovary tissue identification are: S2.1, select a certain number of rat ovary HE staining sections as training set, draw the region of interest on the training set staining section, including ovary tissue area and non-ovary area, mark all the ovary tissue area in the region of interest, mark the non-ovary area as background, then apply deep learning U-Net model Deep Learning-U-Net model for training, and adjust it during training until the ovary tissue identification accuracy reaches more than 90%; S2.2, import the staining section that needs to be formally identified into the software that has completed the model training as described in step S2.1, draw the initial region of interest ROI0, and identify the ovary tissue in the initial region of interest ROI0, the identified ovary tissue area is automatically marked, and the non-ovary area is automatically marked as background; S2.3, further processing, the post-processing steps include: S2.3.1, fill in the blank area in the ovary tissue; S2.3.2, replace the ovary tissue with an area less than the preset value S1 with background; S2.3.3, clear the background, and keep only the ovary tissue on the staining section; S2.3.4, convert the retained ovary tissue into the first region of interest ROI1; S2.4, finally count the area of the ovary tissue in the first region of interest ROI1.

3. The method for grading and counting the growing follicle and corpus luteum of rat ovary according to claim 2, characterized in that: The preset value S1 is equal to 10000 pm 2 .

4. The method for grading and counting the growing follicle and corpus luteum of rat ovary according to claim 1, characterized in that, The specific steps of the step S3 growing follicle and corpus luteum identification and segmentation are: S3.1, select a certain number of rat ovary HE staining sections as training set, draw the region of interest on the training set staining section, mark all the growing follicle and corpus luteum in the region of interest respectively, mark the non-growing follicle and non-corpus luteum area as background, add the boundary of all growing follicle and corpus luteum, and segment the growing follicle and corpus luteum into independent individuals, then apply deep learning semantic segmentation model Deep Learning-DeepLabv3+ model for training, and adjust it during training until the growing follicle and corpus luteum identification and segmentation accuracy reaches more than 90%. S3.2, the first region of interest ROI1 which needs to be formally carried out growth follicle and corpus luteum recognition and segmentation is introduced into the software which is completed model training as described in step S3.1, growth follicle and corpus luteum recognition and segmentation are carried out, the recognized growth follicle, corpus luteum and its boundary are automatically marked, and the non-growth follicle and non-corpus luteum region are automatically marked as background; S3.3, then post-processing is carried out, the post-processing steps include: S3.3.1, the blank area in corpus luteum tissue is filled completely; S3.3.2, the blank area in follicle tissue is filled completely; S3.3.3, the boundary is removed; S3.3.4, the corpus luteum with an area less than a preset value S2 is removed; S3.3.5, the growth follicle with an area less than a preset value S3 is removed; S3.3.6, the background is removed, and the stained section only retains the growth follicle and corpus luteum; S3.4, finally, the number of growth follicles and the number of corpus luteum retained in the first region of interest ROI1 are counted.

5. The method for grading and counting the growing follicle and corpus luteum of rat ovary according to claim 4, characterized in that: Said preset value S2 is equal to 10000 pm 2 Said preset value S3 is equal to 1000 pm 2 .

6. The method for staging and counting growing follicles and corpora lutea in rat ovaries according to claim 1, characterized in that, The specific steps of step S4 for recognizing and counting the growth follicles and corpus luteum at each level are as follows: S4.1, a certain number of rat ovary HE stained sections are selected as a training set, the region of interest is drawn on the stained sections of the training set, step S3 is run in the region of interest, the growth follicles screened out in step S3 are divided and marked as primary follicles, secondary follicles and tertiary follicles, and the remaining follicles are marked as unclassifiable follicles; the corpus luteum screened out in step S3 is divided and marked as neonatal corpus luteum, recent corpus luteum and old corpus luteum, and the remaining corpus luteum is marked as unclassifiable corpus luteum; after marking, a deep learning semantic segmentation model, i.e. Deep Learning-DeepLabv3+ model, is used for training, and after training, the reference features of the growth follicles at each level and the reference features of the corpus luteum at each level are obtained, and the reference features are adjusted in the training until the recognition accuracy of the growth follicles and corpus luteum at each level reaches more than 90%; S4.2, the growth follicles and corpus luteum screened out in step S3 which need to be formally recognized at each level are introduced into the software which is completed model training as described in step S4.1, the actual features of each growth follicle are compared with the reference features of the growth follicles at each level to calculate the feature threshold of each growth follicle corresponding to each level of growth follicle; the actual features of each corpus luteum are compared with the reference features of the corpus luteum at each level to calculate the feature threshold of each corpus luteum corresponding to each level of corpus luteum; S4.3, then post-processing is carried out, the post-processing steps include: S4.3.1, among the corpus luteum screened out in step S3, the corpus luteum with a recent corpus luteum feature threshold greater than a preset value M1 is replaced by recent corpus luteum; S4.3.2, the corpus luteum with a neonatal corpus luteum feature threshold greater than a preset value M2 is replaced by neonatal corpus luteum; S4.3.3, the corpus luteum with an old corpus luteum feature threshold greater than a preset value M3 is replaced by old corpus luteum; S4.3.4, the growth follicle with a primary follicle feature threshold greater than a preset value M4 is replaced by primary follicle; S4.3.5, the growth follicle with a secondary follicle feature threshold greater than a preset value M5 is replaced by secondary follicle; S4.3.6, replacing the growing follicle with the third follicle feature threshold value greater than the preset value M6 with the third follicle; S4.3.7, marking the remaining corpus luteum as a non-classifiable corpus luteum, and marking the remaining follicle as a non-classifiable follicle; S4.4, finally counting the number of follicles and corpus luteum of each level.

7. The method for grading and counting the growing follicle and corpus luteum of rat ovary according to claim 6, characterized in that: The preset value M1 is equal to 0.3, M2 is equal to 0.3, M3 is equal to 0.3, M4 is equal to 0.3, M5 is equal to 0.3, and M6 is equal to 0.3.

Images