Glass slide performance evaluation model training method and device and electronic equipment

Abstract

The invention provides a glass slide performance evaluation model training method and apparatus, and an electronic device. The method comprises the steps of obtaining a liquid drop contour feature corresponding to a liquid drop on each glass slide; screening the glass slides according to the contour features of the liquid drops to select a target glass slide; performing a preset performance evaluation experiment on the target liquid drops on the target glass slide, determining the quantity difference of distributions before and after the preset performance evaluation experiment on the target liquid drops, and determining the shedding rate of the distributions; classifying distributed substance groups in the target liquid drops after the preset performance evaluation experiment is carried out, and determining the distributed substance agglomeration condition of the target liquid drops; and training the glass slide performance evaluation model by using the liquid drop contour features, the distribution object falling rate and the distribution object agglomeration condition to obtain a trained glass slide performance evaluation model. By adopting the glass slide performance evaluation model training method, the glass slide performance evaluation model training device and the electronic equipment, the problem that the accuracy of the hydrophobicity detection of the glass slide is relatively low is solved.

Description

Technical Field

[0001] The present application relates to the field of instrument and equipment detection technology, and specifically, to a slide performance evaluation model training method, device and electronic equipment. Background Art

[0002] Glass slides are medical consumables widely used in pathological detection. They are of many types, large in usage, and have complicated production processes. They can be used to carry and fix samples that need to be observed, and data can be recorded on the glass slides through special coating treatments or printing techniques. They are widely used in biology, medicine, pathology, histology, and other fields. For example, glass slides can be used as carriers for fluorescence in situ hybridization (FISH) samples. The hydrophilicity and hydrophobicity of the glass slide itself affect the cell density. For glass slides with poor hydrophobicity, the cell suspension will collapse quickly and flow around, making the cells too sparse; for glass slides with too strong hydrophobicity, the droplets fail to spread and the cells are too aggregated. The density of cells will directly affect the results of pathological detection. In order to ensure better experimental and detection effects and better control cell density, it is required that the glass slide has hydrophilicity and hydrophobicity that are compatible with the experiment and stable. Therefore, how to detect the hydrophilicity and hydrophobicity (adsorption performance) of the glass slide becomes crucial.

[0003] Currently, when performing hydrophobicity testing on droplets with a small contact angle, the hydrophobicity of the slide is usually determined only based on the cell detachment rate, resulting in a low accuracy rate in the hydrophobicity testing of the slide. Summary of the invention

[0004] In view of this, the purpose of the present application is to provide a slide performance evaluation model training method, device and electronic equipment to solve the problem of low accuracy in slide hydrophobicity detection.

[0005] In a first aspect, an embodiment of the present application provides a method for training a slide performance evaluation model, comprising:

[0006] For each glass slide, a droplet profile feature of a droplet added to the glass slide is obtained, where the droplet is a droplet in a target solution system determined by a type of a preset performance evaluation experiment, and the droplet includes a distribution;

[0007] The slides are screened according to the droplet profile characteristics to select the target slides;

[0008] Performing a preset performance evaluation experiment on the droplets on the target glass slide, and determining a difference in the number of distributed objects before and after the preset performance evaluation experiment on the droplets on the target glass slide, so as to determine a shedding rate of the distributed objects according to the difference in the number of distributed objects;

[0009] Classifying the distributed object clusters in the liquid droplets on the target glass slide after the preset performance evaluation experiment to determine the distribution agglomeration of the liquid droplets on the target glass slide;

[0010] The slide performance evaluation model is trained using droplet contour features, distribution object shedding rate and distribution object agglomeration to obtain the trained slide performance evaluation model.

[0011] Optionally, the slide performance evaluation model is trained using droplet profile features, distribution material shedding rate and distribution material agglomeration, including: performing correlation verification and feature analysis on droplet profile features, distribution material shedding rate and distribution material agglomeration, so as to select important feature parameters for distinguishing different groups of dependent variables from multiple feature parameters; using the important feature parameters as independent variables and the preset evaluation results as dependent variables, so as to train the slide performance evaluation model using the values ​​of the independent variables and the dependent variables.

[0012] Optionally, determining the difference in the number of distributed objects before and after a preset performance evaluation experiment is performed on the droplets on the target glass slide, including: obtaining a first distributed object image of the droplets on the target glass slide at a preset magnification before the preset performance evaluation experiment and a second distributed object image of the droplets on the target glass slide at a preset magnification after the preset performance evaluation experiment; performing image matching processing on the first distributed object image and the second distributed object image to determine the number of distributed objects that is different between the first distributed object image and the second distributed object image.

[0013] Optionally, the distributed object clusters in the droplets on the target glass slide after the preset performance evaluation experiment are classified to determine the distribution agglomeration of the droplets on the target glass slide, including: classifying the distributed object clusters in the second distribution object image according to the preset pixel area ranges corresponding to different types of distribution clusters to determine multiple different types of distribution clusters in the second distribution object image; counting the number of each type of distribution clusters, and determining the distribution agglomeration according to the number of each type of distribution clusters.

[0014] Optionally, before screening the glass slides according to the droplet contour features, the method further includes: determining expected droplet size information based on expected experimental parameters; using the expected droplet size information to determine contour feature requirements, so as to screen the glass slides according to the determination result of whether the droplet contour features meet the contour feature requirements.

[0015] Optionally, the expected droplet size information includes an expected droplet diameter, and the expected droplet size information is determined based on expected experimental parameters, including: determining a total droplet area required to scan an expected number of distributed objects according to the expected experimental parameters; and determining an expected droplet diameter according to the total droplet area.

[0016] Optionally, the expected experimental parameters include distribution parameters and / or field of view parameters, the distribution parameters include the total amount of distribution and the density of distribution, and the field of view parameters include the field of view range and the number of fields of view of the microscope.

[0017] Optionally, the droplet profile feature includes at least one of the following items: circumference, area, roundness and diameter of the droplet.

[0018] In a second aspect, the present application also provides a method for evaluating the performance of a glass slide, comprising:

[0019] Obtaining performance characteristics of the slide to be evaluated;

[0020] The performance characteristics of the slide to be evaluated are input into the trained slide performance evaluation model to obtain the performance evaluation result of the slide to be evaluated, wherein the trained slide performance evaluation model is trained using the above-mentioned slide performance evaluation model training method.

[0021] In a third aspect, the present application also provides a slide performance evaluation model training device, the device comprising:

[0022] A profile feature acquisition module, for acquiring, for each glass slide, a droplet profile feature of a droplet added to the glass slide, wherein the droplet is a droplet in a target solution system determined by a type of a preset performance evaluation experiment, and the droplet includes a distribution;

[0023] A slide selection module is used to screen the slides according to the droplet profile characteristics to select the target slide;

[0024] A shedding rate determination module is used to perform a preset performance evaluation experiment on the droplets on the target glass slide, and determine the difference in the number of distributed objects before and after the preset performance evaluation experiment on the droplets on the target glass slide, so as to determine the shedding rate of the distributed objects according to the difference in the number of distributed objects;

[0025] A distribution agglomeration classification module is used to classify the distribution agglomerations in the droplets on the target glass slide after the preset performance evaluation experiment, so as to determine the agglomeration of the distribution agglomerations of the droplets on the target glass slide;

[0026] The model training module is used to train the slide performance evaluation model using droplet contour features, distribution material shedding rate and distribution material agglomeration conditions to obtain the trained slide performance evaluation model.

[0027] In a fourth aspect, the present application also provides a device for evaluating the performance of a glass slide, the device comprising:

[0028] A performance characteristic acquisition module, used to acquire the performance characteristics of the slide to be evaluated;

[0029] The performance evaluation module is used to input the performance characteristics of the slide to be evaluated into the trained slide performance evaluation model to obtain the performance evaluation result of the slide to be evaluated, wherein the trained slide performance evaluation model is trained using the above-mentioned slide performance evaluation model training method.

[0030] In a fifth aspect, an embodiment of the present application further provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the above-mentioned slide performance evaluation model training method or the steps of the above-mentioned slide performance evaluation method.

[0031] The embodiments of the present application bring the following beneficial effects:

[0032] The embodiments of the present application provide a method, device and electronic device for training a glass slide performance evaluation model, which can preliminarily screen the glass slides according to the droplet contour features, directly remove the glass slides with uneven expansion, and then further determine the droplet distribution material shedding rate and the distribution material agglomeration situation of the target glass slide after the preliminary screening, so as to train the glass slide performance evaluation model by using the droplet contour features, the distribution material shedding rate and the distribution material agglomeration situation, so that the glass slide performance evaluation model can evaluate the hydrophobicity of the glass slide from three dimensions of droplet contour features, distribution material shedding rate and distribution material agglomeration situation, thereby improving the accuracy and efficiency of the glass slide performance detection. Compared with the glass slide performance evaluation model training method in the prior art, the problem of low accuracy of glass slide hydrophobicity detection is solved.

[0033] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 A flow chart showing a method for training a glass slide performance evaluation model provided in an embodiment of the present application is shown;

[0036] Figure 2 A schematic diagram of a first distributed object image provided in an embodiment of the present application is shown;

[0037] Figure 3 A schematic diagram of a second distribution object image provided in an embodiment of the present application is shown;

[0038] Figure 4 A schematic diagram of a cell difference image provided in an embodiment of the present application is shown;

[0039] Figure 5 Schematic diagrams showing different types of cell clusters provided in the examples of the present application;

[0040] Figure 6 A schematic diagram showing a correlation verification result provided by an embodiment of the present application is shown;

[0041] Figure 7 A schematic diagram of the structure of a glass slide performance evaluation model training device provided in an embodiment of the present application is shown;

[0042] Figure 8 A flow chart showing a method for evaluating the performance of a glass slide provided in an embodiment of the present application is shown;

[0043] Fig. 9 A schematic diagram of the structure of a glass slide performance evaluation device provided in an embodiment of the present application is shown;

[0044] Fig.10 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0045] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application usually described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, each other embodiment obtained by those skilled in the art without making creative work belongs to the scope of protection of the present application.

[0046] It is worth noting that before the present application was proposed, slides were medical consumables widely used in pathological detection. They were of many types, used in large quantities, and had complicated production processes. They can be used to carry and fix samples that need to be observed, and can be used to record data on slides through special coating treatments or printing techniques. They are widely used in biology, medicine, pathology, histology, and other fields. For example, slides can be used as carriers for fluorescence in situ hybridization (FISH) samples. The hydrophilicity and hydrophobicity of the slide itself affect the cell density. For slides with poor hydrophobicity, the cell suspension will collapse quickly and flow around, making the cells too sparse; for slides with too strong hydrophobicity, the droplets fail to spread and the cells are too aggregated. In order to improve the hydrophilicity and hydrophobicity of the slide surface, it is usually necessary to coat the slide surface with a surface treatment material or use a surface treatment technique. Commonly used surface treatment materials include poly-polylysine, APES, acetic acid, paraffin, etc. Surface treatment technologies include dipping, spraying, coating, etc., such as siliconization, plasma treatment, and femtosecond laser glass surface modification. These methods can effectively change the surface properties of the glass slide to achieve the target hydrophilicity and hydrophobicity, thereby improving the accuracy and efficiency of the analysis.

[0047] The density of cells will directly affect the results of pathological detection. Taking FISH experiments as an example, the efficiency of probe hybridization depends on the number and density of cells. For example, when the cell density is high, the contact opportunities of probe molecules on the cell surface increase, resulting in competition between probe molecules, thereby reducing the efficiency of probe hybridization.

[0048] In addition, cell density will also affect the intensity of the fluorescent signal and cell count. For example, when the cell density is low, the probe has fewer opportunities to contact the cell, and the combination of the probe molecule and the cell is more stable, thus generating a stronger fluorescent signal. On the contrary, the intensity of the fluorescent signal will decrease. At the same time, when the cell density is low, the gap between cells is larger, and the accuracy of cell counting is higher. When the cell density is high, the gap between cells is smaller, which is prone to miscounting.

[0049] In order to ensure better experimental and detection results and better control cell density, the slide is required to have stable hydrophilic and hydrophobic properties that are compatible with the experiment. Therefore, how to detect the hydrophilic and hydrophobic properties (adsorption properties) of the slide becomes very important.

[0050] In the prior art, hydrophobicity detection is mainly carried out by measuring the contact angle between the liquid and the solid surface (slide surface) through a contact angle meter, and then evaluating the hydrophilicity of the slide through the contact angle. The contact angle meter usually drops a liquid (such as water or oil) on the surface of the material to be tested, and then uses image analysis software to measure the contact angle. The image analysis software automatically calculates the shape of the droplet to determine its contact angle, and the measurement result is usually expressed in the form of an angle or a hydrophobicity index. The larger the contact angle, the more hydrophobic the surface of the material, and vice versa. However, in some specific pathological detection scenarios, for example: FISH experiments or detection processes, the cell suspension is required to have a certain expansion performance on the slide, and the droplet is required to collapse into a liquid surface on the slide. The contact angle is small and is not within the measurement range of the contact angle meter, that is, the existing contact angle meter is not suitable for the determination of the hydrophilicity of the slide in FISH experiments or detection scenarios. When testing the hydrophobicity of droplets with a small contact angle, the hydrophobicity of the slide is usually determined only based on the cell detachment rate. If only the cell detachment rate is used to train the slide performance evaluation model, the test results of the slide performance test model will be inaccurate, resulting in a low accuracy rate for the slide hydrophobicity test.

[0051] Based on this, an embodiment of the present application provides a method for training a glass slide performance evaluation model to improve the accuracy of glass slide hydrophobicity detection.

[0052] See also Figure 1 , Figure 1 This is a flow chart of a method for training a glass slide performance evaluation model provided in an embodiment of the present application. Figure 1 As shown, the slide performance evaluation model training method provided in the embodiment of the present application includes:

[0053] Step S101, for each glass slide, obtaining a droplet profile feature of a droplet added to the glass slide;

[0054] Step S102, screening the glass slides according to the droplet profile features to select the target glass slide;

[0055] Step S103, performing a preset performance evaluation experiment on the droplets on the target glass slide, and determining the difference in the number of distributed objects before and after the preset performance evaluation experiment on the droplets on the target glass slide, so as to determine the shedding rate of the distributed objects according to the difference in the number of distributed objects;

[0056] Step S104, classifying the distributed object clusters in the droplets on the target glass slide after the preset performance evaluation experiment, so as to determine the agglomeration of the distributed objects in the droplets on the target glass slide;

[0057] Step S105 , training the glass slide performance evaluation model using the droplet profile features, the distribution object shedding rate, and the distribution object agglomeration conditions to obtain a trained glass slide performance evaluation model.

[0058] Through the above steps, the present application performs a preliminary screening of the glass slides according to the droplet contour features, directly removes the glass slides with uneven expansion, and then further determines the droplet distribution shedding rate and distribution agglomeration situation of the target glass slide after the preliminary screening, so as to train the glass slide performance evaluation model using the droplet contour features, distribution shedding rate and distribution agglomeration situation, so that the glass slide performance evaluation model can evaluate the hydrophobicity of the glass slide from three dimensions of droplet contour features, distribution shedding rate and distribution agglomeration situation, thereby improving the accuracy and efficiency of the glass slide performance detection and solving the problem of low accuracy of the glass slide hydrophobicity detection.

[0059] Below Figure 1 Each step is described in detail.

[0060] In step S101 , for each glass slide, a droplet profile feature of the droplet dropped on the glass slide is obtained.

[0061] The glass slide may refer to a carrier for carrying a solution. The glass slide is provided by a glass slide manufacturer, and the glass slide manufacturer may provide multiple batches of glass slides. The hydrophobicity of these batches of glass slides needs to be tested or evaluated.

[0062] The droplet may refer to a droplet in a target solution system determined by the type of the preset performance evaluation experiment, that is, the solution system of the droplet is determined by the type of the preset performance evaluation experiment. Exemplarily, when the preset performance evaluation experiment is a FISH test for pathological detection, the droplet may be a droplet drawn from a phosphate buffered saline solution (PBS solution), that is, a PBS droplet.

[0063] The droplets include a distribution, which may be a cell, a magnetic bead or a microsphere. For example, if the preset performance evaluation experiment is a cell experiment, the distribution is a cell, and if the preset performance evaluation experiment is a magnetic bead experiment, the distribution is a magnetic bead. The microspheres or magnetic beads may also be configured into a suspension similar to cells, and then dropped onto a glass slide, and finally adhered to the glass slide through a baking and drying step similar to cells.

[0064] The droplet profile features include at least one of the following: the circumference, area, roundness and diameter of the droplet.

[0065] In the embodiment of the present application, in order to obtain the detection data of the slide, the following equipment needs to be prepared: a microscope, a bright field light source, a low-power objective lens, a slide to be tested, a standard solution for testing, and a pipette.

[0066] The following is an example of the preset performance evaluation experiment being a FISH experiment to introduce the process of acquiring the test data. After preparing the above equipment, a low-power objective lens with a preset magnification is mounted on the microscope, and a blank glass slide is placed steadily on the stage of the microscope. Then, a fixed amount of PBS solution (standard solution) is drawn from the test tube using a pipette. The tip of the pipette is moved to 0.5cm to 1cm above the glass slide, and then the PBS solution is slowly and continuously added to the glass slide, and the tip of the pipette is prevented from touching the droplets. After the PBS solution is added, the added droplets are allowed to diffuse on the glass slide. In addition, when the added droplets collapse on the glass slide, the droplets will gradually evaporate over time, and their evaporation rate is related to the ambient temperature, ambient humidity, light temperature and ambient wind speed. If the liquid evaporates too quickly, it will affect the accuracy of subsequent tests. Therefore, the relevant laboratory environmental parameters can be recorded during the test for subsequent evaluation steps.

[0067] Among them, the volume of PBS solution dripped on each glass slide is set, for example: 20 microliters, to ensure that the extension degree of droplets on different glass slides is similar. In addition, the buffer solution under other solution systems can be various buffer solutions commonly used in biology, or can be a buffer solution containing organic solutions, such as a PBS solution containing 5% methanol or ethanol, PEG400-ethanol-PBS solution. The hydrophilicity and hydrophobicity of the solution itself will also change the distribution state of the distributed substances (distributed substances, magnetic beads or microspheres) therein. The buffer components under different solution systems are different, and their properties such as viscosity and evaporation rate are also different. Buffer solutions under other solution systems also include: tris buffer, hepes buffer, scc buffer, etc. Here, the solution system selected for use is determined by the ongoing experiment or detection scenario.

[0068] For each glass slide, after adding a droplet on the glass slide, slowly move the droplet area on the glass slide to the center of the microscope's field of view, and adjust the image parameters, for example, adjust the magnification of the microscope between one and two times, so that the droplet image is clearly visible. Then, take and save the droplet image of the glass slide, which can be saved in jpg or tiff format.

[0069] The above steps are performed for each glass slide to obtain the droplet images corresponding to each of the multiple glass slides. A number of glass slides greater than or equal to a preset number are randomly selected from each batch of glass slides for detection. The preset number may be 10 or 12. Those skilled in the art may select a specific value of the preset number according to actual conditions, and this application does not limit this.

[0070] Image analysis software is used to perform contour recognition on the droplet images corresponding to a preset number of randomly selected slides, and the values ​​of the droplet contour features such as the area, diameter, roundness, circumference, major axis length and minor axis length of the droplet in the droplet image corresponding to each slide are calculated.

[0071] Among them, the area of ​​the droplet is recorded as: area, the circumference of the droplet is recorded as: perimeter, and the diameter of the droplet is recorded as: diameter. The roundness (circularity) of the droplet can be calculated by the following formula: roundness = (4π×area) / (perimeter×perimeter).

[0072] In step S102, the glass slides are screened according to the droplet profile features to select the target glass slide.

[0073] The target glass slide may refer to a glass slide corresponding to a droplet whose droplet profile characteristics meet the profile characteristic requirements.

[0074] After determining the droplet contour features of the droplets on each slide, since the FISH experiment has requirements for cell density, these slides need to be preliminarily screened to screen out droplets that obviously do not meet the contour feature requirements, and the slides corresponding to the droplets that do not meet the contour feature requirements are eliminated, and the slides corresponding to the droplets that meet the contour feature requirements are used as target slides.

[0075] Specifically, expected droplet size information can be determined based on expected experimental parameters, and the expected droplet size information can be used to determine profile feature requirements, so that the slides can be screened based on the determination result of whether the droplet profile features meet the profile feature requirements.

[0076] In one example, the contour feature requirement includes at least one of the following items: area feature requirement, roundness feature requirement, circumference feature requirement, and diameter feature requirement. Accordingly, the expected droplet size information includes at least one of the following items: expected droplet area, expected droplet roundness, expected droplet circumference, and expected droplet diameter. The contour feature requirement corresponds to the expected droplet size information. For example, when the contour feature requirement includes a circumference feature requirement and a diameter feature requirement, the expected droplet size information includes the expected droplet circumference and the expected droplet diameter.

[0077] Here, taking the expected droplet size information including the expected droplet diameter as an example, the droplet diameter range can be determined based on the expected droplet diameter. When the actually calculated droplet diameter Diameter1 exceeds the droplet diameter range, it is determined that the droplet contour features of the droplets on the glass slide do not meet the contour feature requirements, and the glass slide is discarded; when the actually calculated droplet diameter Diameter1 is within the droplet diameter range, it is determined that the droplet contour features of the droplets on the glass slide meet the contour feature requirements, and the glass slide is used as the target glass slide.

[0078] When determining the droplet diameter range based on the expected droplet diameter, the first range coefficient and the second range coefficient can be set in advance, and the product of the expected droplet diameter and the upper limit of the first range coefficient is used as the lower limit value of the droplet diameter range, and the product of the expected droplet diameter and the second range coefficient is used as the upper limit value of the droplet diameter range, thereby determining the droplet diameter range based on the calculated lower limit value and upper limit value.

[0079] When determining the droplet size information, the total droplet area required to scan the expected number of distributed objects can be determined based on the expected experimental parameters, and then the expected droplet diameter can be determined based on the total droplet area. The expected experimental parameters are set values ​​for determining the total droplet area. The expected experimental parameters include distributed object parameters and / or field of view parameters. The distributed object parameters include the total amount of distributed objects and the density of distributed objects. The field of view parameters include the field of view range and the number of fields of view of the microscope. The total droplet area is recorded as Area, the total amount of distributed objects is recorded as Totalcell, i.e., the total number of distributed objects, the density of distributed objects is recorded as Celldensity, the expected droplet diameter is recorded as Diameter2, the field of view range is recorded as FOVarea, and the number of fields of view is recorded as FOVnum.

[0080] Here, the total droplet area Area can be calculated in two ways: one is to calculate according to the set distribution parameters, and the other is to calculate according to the field of view parameters.

[0081] When calculating the total droplet area based on the distribution parameters, it can be calculated using the following formula:

[0082] Area=Totalcell / Celldensity;

[0083] When calculating the total droplet area based on the field of view parameters, it can be calculated using the following formula:

[0084] Area = FOVarea × FOVnum;

[0085] Taking cells as an example, since the cell diameter of mononuclear cells is 12 microns, and the size of a single field of view under a microscope with a magnification of 40 is 427×370 microns, when the cells completely fill the field of view, (427×370) / (π×6×6)≈1396 cells are required. However, when the cells completely fill the field of view, the cells will aggregate abnormally and cannot be read. For this reason, it is appropriate to set the cells to occupy 20% to 22% of the entire field of view. Then, it can be calculated that 1396×20%≈280, 1396×22%≈310, that is, the number of cells in a single field of view is 280 to 310. Therefore, when capturing 50,000 cells, the number of fields of view that need to be scanned is 160 to 180. Therefore, the value of FOVarea is 427×370=157990 square microns, and the value range of FOVnum is [160,180].

[0086] In this way, the calculation formula for the expected droplet diameter can be obtained as:

[0087]

[0088] In the above formula, A represents the area coefficient, which is a set value and its value range is [1,2].

[0089] Taking the contour feature requirements including the diameter feature requirements and the roundness feature requirements as an example, it is also necessary to determine the droplet roundness range in advance according to the expected droplet roundness. If the actual calculated droplet roundness is within the droplet roundness range, it is determined that the roundness feature requirements are met. If the actual calculated droplet roundness is not within the droplet roundness range, it is determined that the roundness feature requirements are not met. Only when the actual calculated droplet diameter is within the droplet diameter range and the actual calculated droplet roundness is within the droplet roundness range can it be determined that the droplet contour feature corresponding to the droplet on the slide meets the contour feature requirements, and then the slide is used as the target slide.

[0090] In step S103, a preset performance evaluation experiment is performed on the droplets on the target glass slide, and the difference in the number of distributed objects before and after the preset performance evaluation experiment is determined, so as to determine the distribution object shedding rate according to the difference in the number of distributed objects.

[0091] The preset performance evaluation experiment may refer to a preset hydrophobicity detection experiment. For example, the preset performance evaluation experiment may be a FISH experiment.

[0092] In the embodiment of the present application, the process of FISH experiment includes the steps of sample preparation, DNA or RNA denaturation, probe hybridization, washing, fluorescence microscope observation and image analysis. Since FISH experiment belongs to the prior art, it will not be described here. There are multiple target slides, and each target slide needs to be subjected to a preset performance evaluation experiment, and the difference in the number of distributions of the droplets on the target slide before and after the preset performance evaluation experiment is determined, so as to determine the distribution shedding rate of the droplets on the target slide according to the difference in the number of distributions.

[0093] In addition, the preset performance evaluation experiment may also be other types of cell experiments, such as ISH experiments and RNA-FISH experiments.

[0094] When determining the difference in the number of distributed objects before and after the preset performance evaluation experiment, the glass slide can be fixed on a microscope before the preset performance evaluation experiment, and a first scan can be performed to obtain a first distribution image of the droplets on the target glass slide at a preset magnification before the preset performance evaluation experiment. Then, the preset performance evaluation experiment is performed on the glass slide. Taking the preset performance evaluation experiment as an example, the glass slide needs to be digested, washed, and dehydrated with ethanol gradient, and the probe mixture is dripped onto the cells, and denaturation hybridization is performed in a hybridization instrument. After the FISH experiment is completed, a staining solution is dripped onto the droplets on the target glass slide for staining. For example, a DAPI staining solution can be dripped in the FISH experiment, and a second scan is performed to obtain a second distribution image of the droplets on the target glass slide at a preset magnification after the preset performance evaluation experiment. Among them, in the FISH experiment, the first distribution object image is the first cell image, and the second distribution object image is the second cell image. The preset magnification can be 40 times or 60 times. Technical personnel in this field can choose according to actual conditions. This application is not limited here, but it is necessary to ensure that the magnification of the two images acquired before and after the preset performance evaluation experiment remains consistent.

[0095] The first distribution object image is used to record information such as the number, position, and morphology of the distribution objects in the droplets on the target glass slide before the preset performance evaluation experiment. The second distribution object image is used to record information such as the number, position, and morphology of the distribution objects in the droplets on the target glass slide after the preset performance evaluation experiment.

[0096] Refer to the following Figure 2 , Figure 3 Let's introduce the first distribution object image and the second distribution object image.

[0097] Figure 2 A schematic diagram of a first distribution object image provided in an embodiment of the present application is shown. Figure 2As shown, the first distribution image is an image of a droplet on a target glass slide collected before the FISH experiment. The first distribution image records information such as the position and morphology of each cell in the droplet before the FISH experiment. The first distribution image includes cell 201, cell 202, and cell 203. Cell 201, cell 202, and cell 203 are only used to circle out cells that have differences before and after the FISH experiment.

[0098] Figure 3 A schematic diagram of a second distribution image provided in an embodiment of the present application is shown. Figure 3 As shown in FIG. 1 , the second distribution image is an image of the droplet on the target glass slide collected after the FISH experiment. The second distribution image corresponds to the first distribution image in the above example. The second distribution image records information such as the position and morphology of each cell in the droplet after the FISH experiment. The second distribution image includes regions 301, 302, and 303. In region 301, there is no cell 201, in region 302, there is no cell 202, and in region 303, there is no cell 203. Here, dotted lines are used to indicate regions where no cells exist.

[0099] After obtaining the first distribution object image and the second distribution object image, image matching processing is performed on the first distribution object image and the second distribution object image to determine the number of distribution objects that are different between the first distribution object image and the second distribution object image. Specifically, the image matching processing includes threshold segmentation processing and image overlap processing. The first distribution object image and the second distribution object image can be firstly subjected to threshold segmentation, and then the first distribution object image and the second distribution object image are overlapped according to the position correspondence relationship. The overlapped images can be used to determine the distribution objects that are different between the first distribution object image and the second distribution object image.

[0100] Figure 4 A schematic diagram of a cell difference image provided in an embodiment of the present application is shown, such as Figure 4 As shown, since cells 201, 202, and 203 exist in the first distribution image, but cells 201, 202, and 203 do not exist in the second distribution image, each cell with a difference is marked with a special color. Here, black solid circles are used to represent cells with a difference, and the marked ones are difference cells 401, difference cells 402, and difference cells 403. Of course, you can also choose to use red or use other methods to mark cells with a difference.

[0101] The number of differentially distributed objects marked can be counted to determine the number of differentially distributed objects, and the number of differentially distributed objects is used to characterize the difference in the number of distributed objects. Then, the distribution object shedding rate is calculated based on the number of differentially distributed objects. For example, if the number of differentially distributed objects is 100 and the total number of distributed objects in the first distributed object image is 10000, the distribution object shedding rate can be determined to be 100 / 10000=1%. It should be noted that excessive shedding of distributed objects can easily cause the loss of target distributed objects and inaccurate experimental results. Therefore, the distribution object shedding rate needs to be considered when selecting a slide.

[0102] In step S104, the distributed object clusters in the liquid droplets on the target glass slide after the preset performance evaluation experiment are classified to determine the agglomeration of the distributed objects in the liquid droplets on the target glass slide.

[0103] In order to more accurately measure whether the surface of the slide is prone to causing the distribution to aggregate, it is necessary to count the aggregation degree of the distribution on the slide, so as to measure whether the surface of the slide is prone to causing the distribution to aggregate according to the aggregation degree, thereby affecting the subsequent experimental results.

[0104] In the embodiment of the present application, the distribution clusters in the second distribution object image are classified according to the preset pixel area ranges corresponding to the different types of distribution clusters, so as to determine multiple different types of distribution clusters in the second distribution object image. Here, the preset pixel area range corresponding to each type of distribution cluster can be determined first. Taking the FISH experiment as an example, the pixel area range of a single cell cluster is 250 to 750, the pixel area range of a double cell cluster is 750 to 1250, and the pixel area range of a triple cell cluster is 1250 to 1750, ..., and so on, the pixel area ranges of different types of cell clusters can be determined, wherein a double cell cluster can refer to a cell cluster composed of two cells, and a triple cell cluster can refer to a cell cluster composed of three cells. At the same time, after the second distribution object image is subjected to the threshold segmentation process, the actual pixel area of ​​each segmented distribution cluster can also be determined, and then the distribution clusters in the second distribution object image are classified according to the preset pixel area ranges corresponding to different types of distribution clusters, so as to segment multiple different types of distribution clusters from the second distribution object image.

[0105] Refer to the following Figure 5 Let's introduce the style of cell clusters.

[0106] Figure 5 Schematic diagrams of different types of cell clusters provided in the examples of the present application are shown, such as Figure 5As shown, the pixel area of ​​cell cluster 501 is 500, so cell cluster 501 is a single cell cluster, also called a single cell; the pixel area of ​​cell cluster 502 is 1000, so cell cluster 502 is a double cell cluster, also called a double cell cluster; the pixel area of ​​cell cluster 503 is 1500, so cell cluster 503 is a triple cell cluster, also called a triple cell cluster.

[0107] Then, the number of each type of distributed agglomerates is counted, and the agglomeration of the distributed objects is determined according to the number of each type of distributed agglomerates. Specifically, a distribution agglomerate distribution curve can be generated according to the number of each type of distributed agglomerates, the abscissa of the distribution agglomerate distribution curve is the type of distributed agglomerates, and different types of distributed agglomerates are arranged on the abscissa in order from small to large according to the number of distributed objects that compose the distributed agglomerates, the ordinate is the number of distributed agglomerates, and the proportion of different types of distributed agglomerates to the total number of distributed agglomerates is determined, for example: if there are 3,000 single distributed agglomerates, 1,000 double distributed agglomerates, and the total number of distributed agglomerates is 10,000, then the number of single distributed agglomerates accounts for 30%, and the number of double distributed agglomerates accounts for 10%. The agglomeration of the distributed objects can be characterized by the distribution agglomerate distribution curve and the proportion of each type of distributed agglomerates. For example: when the sum of the proportions of single-distribution clusters and double-distribution clusters is set to be greater than or equal to 80%, it is determined that the distribution is well dispersed and there is no aggregation. However, since in the above example, the sum of the proportions of single-distribution clusters and double-distribution clusters is 40%, it is determined that there is distribution aggregation, which will affect the subsequent experimental results.

[0108] In one example, in addition to considering the proportion of distributed clusters, the trend of the distribution cluster distribution curve can also be examined. If the number of different types of distributed clusters in the distribution cluster distribution curve gradually decreases, it indicates that the aggregation trend of the distributed clusters meets the requirements. Otherwise, it is determined that the aggregation trend of the distributed clusters does not meet the requirements, which will affect the subsequent experimental results.

[0109] In step S105, the slide performance evaluation model is trained using the droplet profile features, the distribution object shedding rate, and the distribution object agglomeration conditions to obtain a trained slide performance evaluation model.

[0110] In the embodiment of the present application, correlation verification and feature analysis can be performed on the droplet profile features, the distribution shedding rate and the distribution agglomeration to select important feature parameters for distinguishing different groups of dependent variables from multiple feature parameters. The droplet profile features include the droplet profile features corresponding to the target slide and the droplet profile features corresponding to other slides except the target slide.

[0111] For example, the droplet diameter, droplet area, droplet roundness, distribution shedding rate, single distribution cluster ratio, distribution clustering and other characteristic parameters in the droplet contour characteristics can be selected for correlation verification and characteristic analysis to determine the degree of correlation between the characteristic parameters.

[0112] Refer to the following Figure 6 Let's introduce the process of correlation verification.

[0113] Figure 6 A schematic diagram showing the correlation verification result provided by the embodiment of the present application is shown as follows: Figure 6 As shown in the figure, taking the FISH experiment as an example, the distribution clusters are cell clusters. The correlation coefficients between different features can be obtained through correlation verification. For example, the droplet diameter and the proportion of seven-cell clusters are correlated at -0.62, the proportion of single-cell clusters and the proportion of double-cell clusters to ten-cell clusters are negatively correlated, and the proportions of different types of multi-cell clusters (for example, double-cell clusters to nine-cell clusters) are positively correlated to a certain extent.

[0114] At the same time, variance analysis is performed on each characteristic parameter to test whether each characteristic parameter can be used to distinguish different groups of the dependent variable. Among them, different groups of the dependent variable refer to the values ​​of the dependent variable, such as excellent, qualified, and poor.

[0115] Exemplarily, the results of the feature analysis are introduced below with reference to Table 1.

[0116] Table 1 is the characteristic analysis data obtained under FISH experiment:

[0117]

[0118]

[0119]

[0120] In Table 1, the feature name field is used to list various feature parameters, such as droplet diameter, etc. The group comparison field is used to list the grouping conditions under each feature parameter, including excellent-qualified grouping, excellent-poor grouping, and qualified-poor grouping. The P value field is used to indicate the credibility probability of the hypothesis of the variance analysis result when the variance analysis is performed on the corresponding feature parameter. In this example, when the P value is less than 0.05, it means that the null hypothesis is rejected. The comparison result is used to indicate whether the current feature parameter can be used to distinguish the current different groups.

[0121] For example, the first P value corresponding to the characteristic parameter of droplet diameter is 0, which means that in general, droplet diameter is a feature that can be used to distinguish the following various groups. At the same time, the present application also performs a detailed variance analysis on different groups under droplet diameter. For example, taking the characteristic parameter of droplet diameter as an example, in the excellent-qualified group, the comparison result is True, which means that in the excellent-qualified group, droplet diameter can be used to distinguish different groups, and the different groups here refer to the excellent group and the qualified group; in the excellent-poor group, the comparison result is False, which means that in the excellent-poor group, the droplet diameter feature cannot be used to distinguish different groups, and the different groups here refer to the excellent group and the poor group.

[0122] According to the results of correlation verification and feature analysis, important feature parameters can be selected from multiple feature parameters as independent variables. For example, after correlation verification, feature parameters with autocorrelation or correlation lower than the set threshold can be eliminated and not used as independent variables; after feature analysis, feature parameters with False comparison results can be eliminated and not used as independent variables. At the same time, preset evaluation results are specified according to experimental requirements, and the preset evaluation results are used as dependent variables, wherein the preset evaluation results may refer to evaluation results determined by manual discrimination to characterize the hydrophobic properties of each glass slide, and the value of the dependent variable may be excellent hydrophobicity, qualified hydrophobicity, or poor hydrophobicity. Those skilled in the art can select the specific value of the dependent variable according to actual conditions, and this application does not limit it here.

[0123] After determining the independent variables and dependent variables, the slide performance evaluation model can be trained using the values ​​of the independent variables and dependent variables. The slide performance evaluation model can be a decision tree model or a neural network model. Those skilled in the art can choose which model to use according to actual conditions, and this application does not limit this.

[0124] In one example, the values ​​of the independent variables and the dependent variables can be used as sample data, and the sample data can be divided into a training set and a test set, where the training set accounts for 70% of the sample data and the test set accounts for 30% of the sample data. The decision tree model is trained and tested using the training set and the test set to obtain a trained slide performance evaluation model.

[0125] In a specific example, taking the FISH experiment as an example, 1000 sample data are first obtained. The sample data is a preset sample, which has been manually labeled as excellent, qualified and poor, and is used as a dependent variable for model training. The sample data is divided into a training set and a test set, wherein the training set accounts for 70% and the test set accounts for 30%. The diameter, area, roundness, shedding rate, cell cluster ratio and other features of the training set samples are extracted for correlation verification and feature analysis. According to the results of correlation verification and feature analysis, some features are selected from the extracted features as independent variables, for example: roundness, diameter, single cell cluster ratio, seven cell cluster ratio and cell shedding rate are used as independent variables, or roundness, single cell cluster ratio, three cell cluster ratio and cell shedding rate are used as independent variables. Then, the decision tree model is trained using the training set data. It can be seen from the test set that the accuracy of the decision tree model is 93%, and the model prediction effect is good.

[0126] It should be noted that there may be differences in the manual annotation results of training samples, correlation verification, feature analysis methods and model selection. Therefore, the independent variables and dependent variables used for modeling can be selected according to actual conditions.

[0127] Based on the same inventive concept, the embodiments of the present application also provide a training device for a glass slide performance evaluation model corresponding to the training method for the glass slide performance evaluation model. Since the principle of solving the problem by the device in the embodiments of the present application is similar to the training method for the glass slide performance evaluation model in the embodiments of the present application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0128] See also Figure 7 , Figure 7 This is a schematic diagram of the structure of a slide performance evaluation model training device provided in an embodiment of the present application. Figure 7 As shown in , the slide performance evaluation model training device 600 includes:

[0129] The profile feature acquisition module 601 is used to acquire the droplet profile features of the droplets added to each glass slide, where the droplets are droplets in a target solution system determined by the type of a preset performance evaluation experiment, and the droplets include a distribution;

[0130] A slide selection module 602 is used to screen the slides according to the droplet profile characteristics to select the target slide;

[0131] The shedding rate determination module 603 is used to perform a preset performance evaluation experiment on the droplets on the target glass slide, and determine the difference in the number of distributed objects before and after the preset performance evaluation experiment on the droplets on the target glass slide, so as to determine the shedding rate of the distributed objects according to the difference in the number of distributed objects;

[0132] A distribution agglomeration classification module 604 is used to classify the distribution agglomerations in the droplets on the target glass slide after the preset performance evaluation experiment, so as to determine the agglomeration of the distribution agglomerations of the droplets on the target glass slide;

[0133] The model training module 605 is used to train the slide performance evaluation model using the droplet profile features, the distribution object shedding rate and the distribution object agglomeration conditions to obtain the trained slide performance evaluation model.

[0134] See also Figure 8 , Figure 8 This is a flow chart of the slide performance evaluation method provided in the embodiment of the present application, such as Figure 8 As shown, the slide performance evaluation method provided in the embodiment of the present application includes:

[0135] Step S701, obtaining performance characteristics of the slide to be evaluated;

[0136] Step S702: input the performance characteristics of the slide to be evaluated into the trained slide performance evaluation model to obtain the performance evaluation result of the slide to be evaluated.

[0137] Among them, the trained slide performance evaluation model is obtained by training using the slide performance evaluation model training method as described above. The input of the trained slide performance evaluation model can be the droplet image of the preset performance evaluation experiment of the slide to be evaluated and the first distribution object image and the second distribution object image obtained before and after the preset performance evaluation experiment. At this time, the droplet image, the first distribution object image and the second distribution object image are the performance characteristics of the slide to be evaluated; the input of the trained slide performance evaluation model can also be a feature set determined by correlation verification and feature analysis of the droplet contour characteristics, the distribution object shedding rate and the distribution object agglomeration. At this time, the feature set is the performance characteristics of the slide to be evaluated.

[0138] In an embodiment of the present application, after obtaining a trained glass slide performance evaluation model, the trained glass slide performance evaluation model can be used to evaluate the hydrophobic performance of the glass slide to be evaluated. Here, when it is necessary to detect the glass slide to be evaluated, a droplet can be dripped on the carrier to be evaluated, and a preset performance evaluation experiment can be performed on the glass slide to be evaluated after the droplet is dripped, and the droplet image, the first distribution object image, and the second distribution object image of the glass slide to be evaluated before and after the preset performance evaluation experiment are obtained, and the droplet image, the first distribution object image, and the second distribution object image are input into the glass slide performance evaluation model to output an evaluation result for the hydrophobicity of the glass slide, for example: outputting the hydrophobicity evaluation of the glass slide to be tested as excellent, qualified, or poor.

[0139] Based on the same inventive concept, a glass slide performance evaluation device corresponding to the glass slide performance evaluation method is also provided in the embodiments of the present application. Since the principle of solving the problem by the device in the embodiments of the present application is similar to the above-mentioned glass slide performance evaluation method in the embodiments of the present application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0140] See also Fig. 9 , Fig. 9 This is a schematic diagram of the structure of a slide performance evaluation device provided in an embodiment of the present application. Fig. 9 As shown in , the slide performance evaluation device 800 includes:

[0141] The performance characteristic acquisition module 801 is used to acquire the performance characteristics of the slide to be evaluated;

[0142] The performance evaluation module 802 is used to input the performance characteristics of the slide to be evaluated into the trained slide performance evaluation model to obtain the performance evaluation result of the slide to be evaluated. The trained slide performance evaluation model is trained using the slide performance evaluation model training method as described above.

[0143] See also Fig.10 , Fig.10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Fig.10 As shown in , the electronic device 900 includes a processor 910 , a memory 920 and a bus 930 .

[0144] The memory 920 stores machine-readable instructions executable by the processor 910. When the electronic device 900 is running, the processor 910 communicates with the memory 920 via the bus 930. When the machine-readable instructions are executed by the processor 910, the above-mentioned Figure 1 The slide performance evaluation model training method in the method embodiment shown or Figure 8 The specific implementation of the steps of the slide performance evaluation method in the method embodiment shown can be found in the method embodiment, and will not be repeated here.

[0145] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0146] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0147] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0148] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0149] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application can essentially be embodied in the form of a software product, or in other words, the part that contributes to the prior art or the part of the technical solution. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0150] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for training a slide performance evaluation model, characterized in that: include: For each glass slide, a droplet profile feature of a droplet added to the glass slide is obtained, wherein the droplet is a droplet of a target solution system determined by a type of a preset performance evaluation experiment, and the droplet includes a distribution; Screening the glass slides according to the droplet profile features to select target glass slides; Performing a preset performance evaluation experiment on the droplets on the target glass slide, and determining a difference in the number of distributed objects before and after the droplets on the target glass slide are subjected to the preset performance evaluation experiment, so as to determine a shedding rate of the distributed objects according to the difference in the number of distributed objects; Classifying the distributed objects in the droplets on the target glass slide after the preset performance evaluation experiment to determine the agglomeration of the distributed objects in the droplets on the target glass slide; The glass slide performance evaluation model is trained using the droplet profile features, the distribution material shedding rate, and the distribution material agglomeration conditions to obtain a trained glass slide performance evaluation model.

2. The method according to claim 1, characterized in that The method of training the slide performance evaluation model by using the droplet profile features, the distribution object shedding rate and the distribution object agglomeration conditions includes: Performing correlation verification and feature analysis on the droplet profile features, the distribution dropout rate, and the distribution agglomeration conditions, so as to select important feature parameters for distinguishing different groups of dependent variables from a plurality of feature parameters; The important characteristic parameters are used as independent variables, and the preset evaluation results are used as dependent variables, so as to train the slide performance evaluation model using the values ​​of the independent variables and the dependent variables.

3. The method according to claim 1, characterized in that The method further comprises: determining the difference in the number of distributions of the droplets on the target glass slide before and after the preset performance evaluation experiment, including: Acquire a first distribution image of the liquid droplets on the target glass slide at a preset magnification before performing a preset performance evaluation experiment and a second distribution image of the liquid droplets on the target glass slide at the preset magnification after performing the preset performance evaluation experiment; Image matching processing is performed on the first distributed object image and the second distributed object image to determine the number of distributed objects that are different between the first distributed object image and the second distributed object image.

4. The method according to claim 3, characterized in that The classifying the distributed objects in the droplets on the target glass slide after the preset performance evaluation experiment to determine the agglomeration of the distributed objects in the droplets on the target glass slide includes: Classify the distributed object clusters in the second distributed object image according to preset pixel area ranges corresponding to different types of distributed object clusters, so as to determine a plurality of different types of distributed object clusters in the second distributed object image; The number of each type of distributed object agglomerates is counted, and the distribution agglomeration situation is determined according to the number of each type of distributed object agglomerates.

5. The method according to claim 1, characterized in that Before screening the glass slide according to the droplet profile features, the method further includes: Determining expected droplet size information based on expected experimental parameters; The expected droplet size information is used to determine the profile feature requirements, so as to screen the glass slides according to the determination result of whether the droplet profile features meet the profile feature requirements.

6. The method according to claim 5, characterized in that The expected droplet size information includes an expected droplet diameter, and the determining of the expected droplet size information based on the expected experimental parameters includes: Determining the total droplet area required to scan an expected number of distribution objects according to the expected experimental parameters; Based on the total droplet area, the expected droplet diameter is determined.

7. The method according to claim 6, characterized in that The expected experimental parameters include distribution parameters and / or field of view parameters. The distribution parameters include the total amount of distribution and the density of distribution. The field of view parameters include the field of view range and the number of fields of view of the microscope.

8. The method according to claim 1, characterized in that The droplet profile feature includes at least one of the following items: the circumference, area, roundness and diameter of the droplet.

9. A method for evaluating slide performance, characterized in that: include: Obtaining performance characteristics of the slide to be evaluated; The performance characteristics of the glass slide to be evaluated are input into the trained glass slide performance evaluation model to obtain the performance evaluation result of the glass slide to be evaluated, wherein the trained glass slide performance evaluation model is trained using the glass slide performance evaluation model training method as described in any one of claims 1 to 8.

10. A slide performance evaluation model training device, characterized in that: include: A profile feature acquisition module, for acquiring, for each glass slide, a droplet profile feature of a droplet added to the glass slide, wherein the droplet is a droplet of a target solution system determined by a type of a preset performance evaluation experiment, and the droplet includes a distribution; A glass slide selection module, used for screening the glass slides according to the droplet profile features to select a target glass slide; a shedding rate determination module, configured to perform a preset performance evaluation experiment on the droplets on the target glass slide, and determine the difference in the number of distributed objects before and after the droplets on the target glass slide are subjected to the preset performance evaluation experiment, so as to determine the shedding rate of the distributed objects according to the difference in the number of distributed objects; A distribution agglomeration classification module, used to classify the distribution agglomerations in the droplets on the target glass slide after the preset performance evaluation experiment, so as to determine the agglomeration of the distribution agglomerations of the droplets on the target glass slide; The model training module is used to train the slide performance evaluation model using the droplet profile features, the distribution fall-off rate and the distribution agglomeration conditions to obtain the trained slide performance evaluation model.

11. A slide performance evaluation device, characterized in that: include: A performance characteristic acquisition module, used to acquire the performance characteristics of the slide to be evaluated; A performance evaluation module is used to input the performance characteristics of the slide to be evaluated into a trained slide performance evaluation model to obtain the performance evaluation result of the slide to be evaluated, wherein the trained slide performance evaluation model is trained using the training method of the slide performance evaluation model as described in any one of claims 1 to 8.

12. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the steps of the slide performance evaluation model training method as described in any one of claims 1 to 8 or the slide performance evaluation method as described in claim 9.

Images