A method for detecting appearance change of a monoclonal antibody injection solution after being mixed with various drugs

Abstract

The present application relates to the technical field of image analysis, and particularly relates to a method for detecting appearance change of monoclonal antibody injection solution after being mixed with multiple drugs, which comprises the following steps: obtaining dark field images of a target solution after monoclonal antibody injection solution is mixed with a preset number of drugs within a current time period, and bright field images of the target solution at a current time; identifying target micro-object regions from the bright field images, and screening reference micro-object region sequences corresponding to the target micro-object regions; determining bright field bubble possibility indexes and dark field bubble possibility indexes corresponding to each target micro-object region; determining target bubble possibility indexes corresponding to each target micro-object region; and judging whether the target micro-object regions represent bubbles according to the target bubble possibility indexes. The present application realizes bubble detection by analyzing bright field images and dark field images of a solution after monoclonal antibody injection solution is mixed with multiple drugs, and improves the bubble detection accuracy of monoclonal antibody injection solution after being mixed with multiple drugs to a certain extent.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image analysis, and in particular to a method for detecting appearance changes of a monoclonal antibody injection solution after being mixed with multiple drugs. BACKGROUND

[0002] The monoclonal antibody injection solution is a complex protein preparation. During the mixing process of the monoclonal antibody injection solution with excipients or other therapeutic drugs, air is often introduced, forming sub-visible microbubbles, usually referring to gaseous particles of 5-100 microns. Sub-visible bubbles mainly exist in a physical manner, and their hazards are often different from other particles. For example, the degree of harm of sub-visible bubbles is often less than that of protein aggregates. Therefore, the bubbles in the mixed solution can be detected to assist the doctor in judging the overall effect of the mixed solution. At present, the bubbles in the solution are often detected by manual observation under a microscope.

[0003] However, when detecting the bubbles in the solution after mixing the monoclonal antibody injection solution with multiple drugs by manual observation under a microscope, the bubble detection is usually realized by microscope observation of the staff, mainly relying on manual operation, which often leads to poor detection efficiency, and the subjectivity of manual operation is often strong, which may lead to poor accuracy of bubble detection after mixing the monoclonal antibody injection solution with multiple drugs, thereby leading to poor accuracy of appearance change detection after mixing the monoclonal antibody injection solution with multiple drugs. SUMMARY

[0004] In order to solve the technical problem of poor accuracy of bubble detection after mixing the monoclonal antibody injection solution with multiple drugs, the present application provides a method for detecting appearance changes of a monoclonal antibody injection solution after being mixed with multiple drugs.

[0005] In the first aspect, the present application provides a method for detecting appearance changes of a monoclonal antibody injection solution after being mixed with multiple drugs, which comprises:

[0006] obtaining dark field images of a target solution after mixing a monoclonal antibody injection solution with a preset number of drugs in a current time period, and bright field images of the target solution at a current time, wherein the current time is the end time of the current time period;

[0007] identifying a target micro-object region from the bright field image, and screening a region matching the target micro-object region from all dark field images in the current time period as a reference micro-object region sequence corresponding to the target micro-object region;

[0008] determining a bright field bubble possible indicator corresponding to each target micro-object region according to the internal and external gray scale differences and shape features of each target micro-object region;

[0009] According to the motion trajectory analysis processing of the reference micro-object area sequence corresponding to each target micro-object area, a dark-field bubble possibility index corresponding to each target micro-object area is obtained.

[0010] According to the area, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object area, and the total length of the motion trajectory corresponding to the reference micro-object area sequence corresponding to each target micro-object area, a target bubble possibility index corresponding to each target micro-object area is determined.

[0011] According to the target bubble possibility index corresponding to each target micro-object area, it is judged whether each target micro-object area represents a bubble.

[0012] In combination with the first aspect, in a possible implementation manner, the target micro-object area is identified from the bright-field image, and the target micro-object area includes:

[0013] The bright-field image is subjected to spot detection through a LOG model, and a target micro-object area is obtained.

[0014] In combination with the first aspect, in a possible implementation manner, the region matched with the target micro-object area is filtered out from all the dark-field images in the current time period as the reference micro-object area sequence corresponding to the target micro-object area, and the reference micro-object area sequence includes:

[0015] Any one target micro-object area is determined as a marker micro-object area, and a temporary micro-object area is filtered out from the last dark-field image in the current time period through a SIFT feature matching algorithm, and the temporary micro-object area is matched with the marker micro-object area.

[0016] The region matched with the temporary micro-object area is filtered out from all the dark-field images in the current time period through an optical flow method, and the reference micro-object area sequence corresponding to the marker micro-object area is constituted, and the marker micro-object area and the reference micro-object area in the reference micro-object area sequence corresponding to the marker micro-object area represent the same object.

[0017] In combination with the first aspect, in a possible implementation manner, the bright-field bubble possibility index corresponding to each target micro-object area is determined according to the internal and external gray-scale difference and the shape feature of each target micro-object area, and the bright-field bubble possibility index includes:

[0018] According to the centroid of each target micro-object area and different preset directions, a shape feature index corresponding to each target micro-object area is determined.

[0019] According to the preset neighborhood corresponding to different edge pixel points on each target micro-object area, an internal and external gray-scale difference corresponding to each target micro-object area is determined.

[0020] The product between the shape feature index corresponding to each target micro-object region and the internal-external gray difference is normalized to obtain the bright-field bubble possibility index corresponding to each target micro-object region.

[0021] In a possible implementation manner of the first aspect, the shape feature index corresponding to each target micro-object region is determined according to the centroid of each target micro-object region and different preset directions, and the shape feature index corresponding to each target micro-object region is determined according to the following steps.

[0022] An arbitrary target micro-object region is determined as a marker micro-object region, and a target straight line of the marker micro-object region in each preset direction is drawn from the centroid of the marker micro-object region along each preset direction.

[0023] An intersection between the marker micro-object region and the target straight line of the marker micro-object region in each preset direction is determined as a target intersection line segment of the marker micro-object region in each preset direction.

[0024] The length of the target intersection line segment of the marker micro-object region in each preset direction is determined as a target length of the marker micro-object region in each preset direction.

[0025] The shape feature index corresponding to the marker micro-object region is determined according to the maximum value and the minimum value of the target lengths of the marker micro-object region in all preset directions.

[0026] In a possible implementation manner of the first aspect, the internal-external gray difference corresponding to each target micro-object region is determined according to the preset neighborhood corresponding to each edge pixel point of each target micro-object region, and the internal-external gray difference corresponding to each target micro-object region is determined according to the following steps.

[0027] An arbitrary target micro-object region is determined as a marker micro-object region, and a union set of the preset neighborhoods corresponding to all edge pixel points of the marker micro-object region is determined as a target union region corresponding to the marker micro-object region.

[0028] All pixel points in the target union region corresponding to the marker micro-object region that do not belong to the marker micro-object region form a target peripheral region corresponding to the marker micro-object region.

[0029] The internal-external gray difference corresponding to the marker micro-object region is determined according to the gray difference between the marker micro-object region and the target peripheral region corresponding to the marker micro-object region.

[0030] In a possible implementation manner of the first aspect, the dark-field bubble possibility index corresponding to each target micro-object region is obtained by performing motion trajectory analysis processing on the reference micro-object region sequence corresponding to each target micro-object region, and the dark-field bubble possibility index corresponding to each target micro-object region is obtained according to the following steps.

[0031] determining any one target micro-object region as a marked micro-object region, and aliquoting a sequence of reference micro-object regions corresponding to the marked micro-object region to obtain a reference micro-object region sub-segment;

[0032] constructing a sequence of reference micro-object region sub-segments by using all the reference micro-object region sub-segments;

[0033] determining a mean square displacement corresponding to each reference micro-object region sub-segment according to a position change of a reference micro-object region in each reference micro-object region sub-segment in the sequence of reference micro-object region sub-segments to obtain a sequence of mean square displacements corresponding to the marked micro-object region;

[0034] determining a target change characteristic value corresponding to the marked micro-object region according to the sequence of mean square displacements corresponding to the marked micro-object region;

[0035] determining a directionality coefficient corresponding to the marked micro-object region according to a motion trajectory condition corresponding to the sequence of reference micro-object regions corresponding to the marked micro-object region;

[0036] normalizing a product between the target change characteristic value corresponding to the marked micro-object region and the directionality coefficient to obtain a dark field bubble possibility index corresponding to the marked micro-object region.

[0037] In a possible implementation manner of the first aspect, the determining of the target change characteristic value corresponding to the marked micro-object region according to the sequence of mean square displacements corresponding to the marked micro-object region comprises:

[0038] performing linear fitting on the sequence of mean square displacements corresponding to the marked micro-object region to obtain a linear fitting goodness corresponding to the marked micro-object region;

[0039] determining the target change characteristic value corresponding to the marked micro-object region according to the linear fitting goodness corresponding to the marked micro-object region, wherein the linear fitting goodness and the target change characteristic value are in a negative correlation relationship.

[0040] In a possible implementation manner of the first aspect, the determining of the directionality coefficient corresponding to the marked micro-object region according to the motion trajectory condition corresponding to the sequence of reference micro-object regions corresponding to the marked micro-object region comprises:

[0041] determining a target distance between each two reference micro-object regions in the sequence of reference micro-object regions as a distance between the centroids of the each two reference micro-object regions;

[0042] According to a target distance between a first reference micro-object region and a last reference micro-object region in a reference micro-object region sequence corresponding to the marked micro-object region, and an accumulated value of target distances between all adjacent reference micro-object regions in the reference micro-object region sequence corresponding to the marked micro-object region, a directionality coefficient corresponding to the marked micro-object region is determined.

[0043] With the first aspect above, in a possible implementation, the target bubble possibility indicator corresponding to each target micro-object region is determined according to an area corresponding to each target micro-object region, a bright-field bubble possibility indicator and a dark-field bubble possibility indicator corresponding to each target micro-object region, and a total length of a motion trajectory corresponding to a reference micro-object region sequence corresponding to each target micro-object region, and the determination comprises:

[0044] The area corresponding to each target micro-object region is normalized to obtain a bright-field reliability factor corresponding to each target micro-object region;

[0045] The total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region is normalized to obtain a dark-field reliability factor corresponding to each target micro-object region;

[0046] The bright-field weight and the dark-field weight corresponding to each target micro-object region are determined according to the bright-field reliability factor and the dark-field reliability factor corresponding to each target micro-object region;

[0047] The target bubble possibility indicator corresponding to each target micro-object region is determined according to a product between the bright-field weight corresponding to each target micro-object region and the bright-field bubble possibility indicator, and the dark-field weight corresponding to each target micro-object region and the dark-field bubble possibility indicator.

[0048] In a second aspect, the present application provides a system for detecting appearance changes of a monoclonal antibody injection solution after being mixed with multiple drugs, and the system comprises:

[0049] An image acquisition module is configured to acquire dark-field images of a target solution obtained by mixing a monoclonal antibody injection solution with a preset number of drugs within a current time period, and bright-field images of the target solution at a current time point, wherein the current time point is an ending time point of the current time period.

[0050] An identification and screening module is configured to identify target micro-object regions from the bright-field images, and screen regions matching the target micro-object regions from all dark-field images within the current time period as a reference micro-object region sequence corresponding to the target micro-object regions.

[0051] A bright-field bubble possibility indicator determination module is configured to determine a bright-field bubble possibility indicator corresponding to each target micro-object region according to internal and external gray scale differences and shape features of each target micro-object region.

[0052] The motion trajectory analysis processing module is configured to perform motion trajectory analysis processing on each target micro-object region according to the corresponding reference micro-object region sequence of the target micro-object region, and obtain a dark-field bubble possibility index corresponding to each target micro-object region;

[0053] The target bubble possibility index determination module is configured to determine a target bubble possibility index corresponding to each target micro-object region according to the area, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object region, and a total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region.

[0054] The region judgment module is configured to judge whether each target micro-object region represents a bubble according to the target bubble possibility index corresponding to each target micro-object region.

[0055] In a third aspect, a server is provided, including a memory and a processor. The memory is configured to store executable program code, and the processor is configured to call and run the executable program code from the memory, so that the device executes the method in the first aspect or any possible implementation manner of the first aspect.

[0056] In a fourth aspect, a computer program product is provided, which includes computer program code. When the computer program code runs on a computer, the computer program code causes the computer to execute the method in the first aspect or any possible implementation manner of the first aspect.

[0057] In a fifth aspect, a computer readable storage medium is provided, which stores computer program code. When the computer program code runs on a computer, the computer program code causes the computer to execute the method in the first aspect or any possible implementation manner of the first aspect.

[0058] The present application has the following beneficial effects:

[0059] The appearance change detection method of the monoclonal antibody injection solution after being mixed with multiple drugs realizes bubble detection by analyzing the bright field image and the dark field image of the solution after the monoclonal antibody injection solution is mixed with multiple drugs, solves the technical problem that the bubble detection accuracy of the monoclonal antibody injection solution after being mixed with multiple drugs is poor, and improves the bubble detection accuracy of the monoclonal antibody injection solution after being mixed with multiple drugs to a certain extent. Specifically, the application quantifies a plurality of bubble feature-related indexes, such as the bright field bubble possible index, the dark field bubble possible index and the target bubble possible index, relatively objectively by analyzing the dark field image of the target solution in the current time period and the bright field image of the target solution at the current time, and judges whether the target micro-object region represents a bubble based on the target bubble possible index corresponding to the target micro-object region, relatively objectively realizes bubble detection, reduces the influence of subjective factors to a certain extent, thereby improving the bubble detection accuracy of the monoclonal antibody injection solution after being mixed with multiple drugs, and improving the appearance change detection accuracy of the monoclonal antibody injection solution after being mixed with multiple drugs. BRIEF DESCRIPTION OF DRAWINGS

[0060] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0061] Figure 1 The flow chart of the appearance change detection method of the monoclonal antibody injection solution after being mixed with multiple drugs according to the present application;

[0062] Figure 2 The composition structure schematic diagram of the appearance change detection system of the monoclonal antibody injection solution after being mixed with multiple drugs according to the present application;

[0063] Figure 3 The structure schematic diagram of the computer device according to the present application. DETAILED DESCRIPTION

[0064] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific implementation, structure, features and effects of the technical solutions according to the present application are described in detail as follows by combining with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0065] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0066] Reference Figure 1 , shows the flow of some embodiments of the appearance change detection method of the monoclonal antibody injection solution after being combined with multiple drugs. The appearance change detection method of the monoclonal antibody injection solution after being combined with multiple drugs includes the following steps:

[0067] Step S1, obtain the dark field image of the target solution after the monoclonal antibody injection solution is combined with a preset number of drugs within a current time period, and the bright field image at the current time.

[0068] Wherein, the monoclonal antibody injection solution, also known as monoclonal antibody injection solution, is a complex protein preparation. The preset number can be a pre-set number, which can be 2. The drug can be but not limited to: antibiotics and electrolyte solution. The target solution can be a sample of the solution formed after the monoclonal antibody injection solution is combined with a preset number of drugs. The target solution can be placed in a quartz chamber culture dish. The duration of the current time period can be 5 minutes. The current time can be the end time of the current time period. The bright field image is often a transmission beam imaging. The dark field image is often a diffraction beam imaging. The bright field image and the dark field image can be alternately collected by the bright / dark field inverted microscope, that is, in each alternately collection process, a bright field image is collected first and then a dark field image is collected. In actual situations, the bright field image and the dark field image can be collected in millisecond or microsecond time scale, that is, the duration of each alternately collection process is often short, therefore, the bright field image and the dark field image collected in each alternately collection process can be considered as images collected at the same time.

[0069] As an example, through the bright / dark field inverted microscope facing the side of the target quartz chamber culture dish, alternately collect every 1 second within the current time period, in each alternately collection process, collect a bright field image first and then collect a dark field image, and the alternately collection process is as short as possible, at this time, all the dark field images collected constitute the dark field image within the current time period, the last frame of bright field image can be recorded as the bright field image at the current time, it should be noted that the bright field image involved in subsequent processing refers to the bright field image at the current time unless otherwise specified. Wherein, the target quartz chamber culture dish can be a cylindrical transparent quartz chamber culture dish for placing the target solution.

[0070] It should be noted that during the acquisition process of the dark field image and the bright field image, a switchable bright / dark field inverted microscope system can also be deployed, such as an Olympus IX83 equipped with an electric DIC (Differential Interference Contrast, differential interference contrast) / dark field condenser, a high-sensitivity SCMOS (Scientific Complementary Metal-Oxide-Semiconductor, scientific complementary metal oxide semiconductor) camera (frame rate ≥ 100 fps, pixel size ≤ 6.5 μm) and a 60× oil immersion objective lens (NA ≥ 1.4). The system integrates a temperature-controlled stage (accuracy ±0.5℃) to ensure that the sample is observed in a constant temperature environment of 25±1℃, avoiding the influence of temperature fluctuations on Brownian motion. The light source uses an LED (Light-Emitting Diode, light-emitting diode) cold light source (bright field: transmitted white light; dark field: ring oblique illumination) to reduce phototoxicity. Take 1 mL of single antibody-drug mixture and inject it into a special quartz chamber culture dish (depth 0.2 mm), which can be left for 1 minute to eliminate large air bubbles.

[0071] Step S2, identify the target micro-object region from the bright field image, and screen out the regions matched with the target micro-object region from all dark field images in the current time period as the reference micro-object region sequence corresponding to the target micro-object region.

[0072] As an example, this step can include the following steps:

[0073] First, by the LOG model, the above bright field image is subjected to spot detection to obtain the target micro-object region.

[0074] Wherein, the target micro-object region can represent 5-100 μm microparticles, which can be sub-visible bubbles and other 5-100 μm microparticles, for example, other 5-100 μm microparticles can be protein aggregates. In actual situations, when the single antibody injection is mixed with excipients or other therapeutic drugs (such as antibiotics, electrolyte solutions), changes in pH, ion strength, chemical incompatibility or mechanical stress (such as mixing, shaking) may cause changes in protein molecular conformation or abnormal interactions, forming protein aggregates through hydrophobic interaction, hydrogen bonding, disulfide bond mismatching, etc.

[0075] It should be noted that the LOG (Laplacian of Gaussian, spot detection) model is often used for spot detection, which can detect 5-100 μm microparticle regions.

[0076] Secondly, any one target micro-object region is determined as a marked micro-object region, and a temporary micro-object region is filtered out from the last dark-field image in the current time period by a SIFT (Scale-Invariant Feature Transform) feature matching algorithm, which is matched with the marked micro-object region.

[0077] The last dark-field image in the current time period is the latest dark-field image collected in the current time period. The object represented by the temporary micro-object region can be the same as the object represented by the marked micro-object region.

[0078] Thirdly, a reference micro-object region sequence corresponding to the marked micro-object region is constructed by filtering out regions matched with the temporary micro-object region from all dark-field images in the current time period by an optical flow method.

[0079] The marked micro-object region and the reference micro-object region in the reference micro-object region sequence corresponding to the marked micro-object region can represent the same object.

[0080] Step S3, determining a bright-field bubble possible indicator corresponding to each target micro-object region according to the internal and external gray difference and shape feature of each target micro-object region.

[0081] It should be noted that under bright field, the bubble is often the interface between gas and liquid. Due to the strong refraction of the bubble edge, light often converges to form a bright ring, and there is less light passing through the center area, which often leads to the most prominent feature of the bubble being a bright edge and a relatively dark center. This "bright ring-dark center" morphology is often a hallmark feature of bubbles. Moreover, bubbles are usually perfect spheres or close to spheres.

[0082] As an example, the present step can include the following steps:

[0083] Firstly, determining a shape feature indicator corresponding to each target micro-object region according to the centroid of each target micro-object region and different preset directions.

[0084] The preset direction can be a pre-set direction. The number of preset directions can be pre-set, which can be 9. For example, the preset directions can be, but are not limited to, a horizontal direction, a 10° direction, a 20° direction, a 30° direction, a 40° direction, a 50° direction, a 60° direction, a 70° direction, an 80° direction, and a vertical direction.

[0085] For example, determining a shape feature indicator corresponding to each target micro-object region can include the following sub-steps:

[0086] The first sub-step is to determine any one target micro-object region as a marked micro-object region, and to draw a target straight line of the marked micro-object region in each preset direction through the centroid of the marked micro-object region.

[0087] For example, the target straight line of the marked micro-object region in the horizontal direction can be a straight line drawn through the centroid of the marked micro-object region in the horizontal direction.

[0088] The second sub-step is to determine the intersection between the marked micro-object region and the target straight line of the marked micro-object region in each preset direction as a target intersection line segment of the marked micro-object region in each preset direction.

[0089] The third sub-step is to determine the length of the target intersection line segment of the marked micro-object region in each preset direction as a target length of the marked micro-object region in each preset direction.

[0090] The fourth sub-step is to determine the shape feature index corresponding to the marked micro-object region according to the maximum and minimum values of the target lengths of the marked micro-object region in all preset directions.

[0091] For example, the formula for determining the shape feature index corresponding to the marked micro-object region can be: ;

[0092] wherein, is the shape feature index corresponding to the marked micro-object region. is the minimum value of the target lengths of the marked micro-object region in all preset directions. is the maximum value of the target lengths of the marked micro-object region in all preset directions.

[0093] It should be noted that in actual situations, the finer the preset directions are divided, the more the shape feature index corresponding to the target micro-object region can represent the ratio between the length of the shortest connecting line and the length of the longest connecting line of the target micro-object region. When A is larger, the lengths of the longest connecting line and the shortest connecting line of the marked micro-object region are closer, and the marked micro-object region is more likely to be circular.

[0094] The second step is to determine the internal and external gray scale differences corresponding to each target micro-object region according to the preset neighborhood corresponding to different edge pixel points on each target micro-object region.

[0095] The preset neighborhood can be a preset neighborhood, which can be an N×N neighborhood. N can be equal to half of the maximum value of the target lengths of the target micro-object region in all preset directions.

[0096] For example, determining the internal and external gray scale differences corresponding to each target micro-object region can include the following sub-steps:

[0097] The first sub-step is to determine any one target micro-object region as a marked micro-object region, and determine the union of the preset neighborhoods corresponding to all edge pixel points in the marked micro-object region as a target union region corresponding to the marked micro-object region.

[0098] The second sub-step is to form a target peripheral region corresponding to the marked micro-object region from all pixel points in the target union region corresponding to the marked micro-object region but not belonging to the marked micro-object region.

[0099] The third sub-step is to determine an internal-external gray difference corresponding to the marked micro-object region according to the gray difference between the marked micro-object region and the target peripheral region corresponding thereto.

[0100] For example, the formula for determining the internal-external gray difference corresponding to the marked micro-object region can be: ;

[0101] wherein, is the internal-external gray difference corresponding to the marked micro-object region. is a normalization function. G is the mean value of the gray values corresponding to all pixel points in the target peripheral region corresponding to the marked micro-object region. is the mean value of the gray values corresponding to all pixel points in the marked micro-object region.

[0102] It should be noted that when is larger, it often indicates that the external gray of the marked micro-object region is more likely to be higher than the internal gray, it often indicates that the marked micro-object region is more likely to meet the "bright ring-dark center" feature, and it often indicates that the marked micro-object region is more likely to represent a bubble.

[0103] The third step is to normalize the product between the shape feature index corresponding to each target micro-object region and the internal-external gray difference to obtain a bright-field bubble possibility index corresponding to each target micro-object region.

[0104] It should be noted that when the bright-field bubble possibility index corresponding to the target micro-object region is larger, it often indicates that the target micro-object region exhibits the characteristics of a bubble under bright field.

[0105] Step S4, according to the reference micro-object region sequence corresponding to each target micro-object region, motion trajectory analysis processing is performed to obtain a dark-field bubble possibility index corresponding to each target micro-object region.

[0106] It should be noted that the protein aggregate is often a solid or semi-solid particle formed by the abnormal connection of protein molecules (proteins in monoclonal antibodies or other compatible drugs) through hydrophobic interaction, hydrogen bond, disulfide bond, etc.; Its density is usually close to or slightly greater than the surrounding liquid medium. In solution, Brownian force is often the most critical force driving aggregates to move, and aggregates often exhibit rapid, chaotic, and directionless tremors or shaking near the original field of view under a microscope, and the movement trajectory is often relatively tortuous and often has no obvious overall movement direction. While sub-visible bubbles are often introduced during drug compounding, they are hollow spheres, and because the density of the gas is often much less than the density of the liquid, the buoyancy is often the most critical force. Therefore, bubbles often exhibit clear, stable, and continuous movement on the screen under a microscope, and the movement trajectory is relatively straight.

[0107] As an example, the present step can include the following steps:

[0108] First, any one target micro-object region is determined as a marker micro-object region, and the reference micro-object region sequence corresponding to the marker micro-object region is equally divided to obtain a reference micro-object region sub-section.

[0109] The time length corresponding to the reference micro-object region sub-section can be 30 seconds.

[0110] Second, all reference micro-object region sub-sections are constructed into a reference micro-object region sub-section sequence.

[0111] The reference micro-object region sub-section sequence can be a time sequence.

[0112] Third, according to the position change of the reference micro-object region in each reference micro-object region sub-section in the reference micro-object region sub-section sequence, the mean square displacement corresponding to each reference micro-object region sub-section is determined to obtain the mean square displacement sequence corresponding to the marker micro-object region.

[0113] The mean square displacement sequence corresponding to the marker micro-object region can be a time sequence. The mean square displacement corresponding to the reference micro-object region sub-section can represent the movement of the object represented by the reference micro-object region sub-section in the time period corresponding to the reference micro-object region sub-section.

[0114] For example, the method for obtaining the mean square displacement corresponding to the reference micro-object region sub-section can be: the square of the displacement between each adjacent region in the reference micro-object region sub-section is denoted as a local displacement square index, a local displacement square index sequence corresponding to the reference micro-object region sub-section is obtained, and the mean value of all local displacement square indices in the local displacement square index sequence corresponding to the reference micro-object region sub-section is denoted as the mean square displacement corresponding to the reference micro-object region sub-section.

[0115] The fourth step of determining the target change characteristic value corresponding to the marked micro-object region according to the mean square displacement sequence corresponding to the marked micro-object region can include the following sub-steps:

[0116] The first sub-step is to perform linear fitting on the mean square displacement sequence corresponding to the marked micro-object region with time as the horizontal axis and mean square displacement as the vertical axis to obtain the linear fitting goodness of the marked micro-object region.

[0117] The linear fitting goodness can be in the range of [0, 1]. The greater the linear fitting goodness, the more likely the mean square displacement sequence is linear.

[0118] The second sub-step is to determine the target change characteristic value corresponding to the marked micro-object region according to the linear fitting goodness of the marked micro-object region.

[0119] The linear fitting goodness can be negatively correlated with the target change characteristic value.

[0120] For example, the formula for determining the target change characteristic value corresponding to the marked micro-object region can be:

[0121] ;

[0122] Wherein, L is the target change characteristic value corresponding to the marked micro-object region. R is the linear fitting goodness corresponding to the marked micro-object region.

[0123] It should be noted that sub-visible bubbles are often introduced during drug compounding, which are hollow spheres. Since the density of the gas is often much smaller than the density of the liquid, the buoyancy is often the most critical force. Therefore, the mean square displacement of sub-visible bubbles at different times is often different. In the solution, the Brownian force is often the most critical force driving the motion of the aggregate, so the difference between the mean square displacements of the aggregate at different times is often small. When L The greater the linear fitting goodness, the greater the difference between the mean square displacements of the marked micro-object region at different times, and the more likely the marked micro-object region represents a bubble.

[0124] The fifth step of determining the directionality coefficient corresponding to the marked micro-object region according to the motion trajectory of the reference micro-object region sequence corresponding to the marked micro-object region can include the following sub-steps:

[0125] The first sub-step is to determine the distance between the centroids of each two reference micro-object regions in the reference micro-object region sequence as the target distance between each two reference micro-object regions.

[0126] The reference micro-object region is a region in a sequence of reference micro-object regions. The method for obtaining the target distance between two reference micro-object regions can be as follows: the two reference micro-object regions are respectively denoted as a first reference micro-object region and a second reference micro-object region, a region with the same position as the first reference micro-object region is screened out from the dark field image to which the second reference micro-object region belongs, and the region is denoted as a third reference micro-object region, and the Euclidean distance between the center of mass of the second reference micro-object region and the center of mass of the third reference micro-object region is denoted as the target distance between the first reference micro-object region and the second reference micro-object region.

[0127] In the second sub-step, the directionality coefficient corresponding to the marked micro-object region is determined according to the target distance between the first reference micro-object region and the last reference micro-object region in the sequence of reference micro-object regions corresponding to the marked micro-object region and the cumulative value of the target distances between all adjacent reference micro-object regions in the sequence of reference micro-object regions corresponding to the marked micro-object region.

[0128] For example, the formula for determining the directionality coefficient corresponding to the marked micro-object region can be as follows:

[0129] ;

[0130] wherein, Q is the directionality coefficient corresponding to the marked micro-object region. d is the target distance between the first reference micro-object region and the last reference micro-object region in the sequence of reference micro-object regions corresponding to the marked micro-object region. D is the cumulative value of the target distances between all adjacent reference micro-object regions in the sequence of reference micro-object regions corresponding to the marked micro-object region.

[0131] It should be noted that, in actual situations, bubbles often move directionally due to buoyancy, while aggregates often move randomly, so the motion displacement of bubbles is often consistent with the total path length, while aggregates randomly walk due to Brownian motion, and the motion displacement is often much smaller than the total path length. Therefore, when Q is larger, it often indicates that the motion displacement represented by the marked micro-object region is more likely to be consistent with the total path length, and it often indicates that the marked micro-object region is more likely to be a bubble.

[0132] In the sixth step, the product of the target change characteristic value corresponding to the marked micro-object region and the directionality coefficient is normalized to obtain the dark field bubble possibility index corresponding to the marked micro-object region.

[0133] It should be noted that, when the dark field bubble possibility index corresponding to the marked micro-object region is larger, it often indicates that the marked micro-object region exhibits the characteristics of a bubble under dark field.

[0134] Step S5, determining a target bubble possibility index corresponding to each target micro-object region according to an area corresponding to each target micro-object region, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object region, and a total length of a motion trajectory corresponding to a reference micro-object region sequence corresponding to each target micro-object region.

[0135] The total length of the motion trajectory corresponding to the reference micro-object region sequence can be equal to an accumulated value of target distances between all adjacent reference micro-object regions in the reference micro-object region sequence.

[0136] As an example, the present step can include the following steps:

[0137] Firstly, normalizing the area corresponding to each target micro-object region to obtain a bright-field reliability factor corresponding to each target micro-object region.

[0138] It should be noted that the bright-field data is often closely related to the size of the particle area in the image. The larger the particle area, the more reliable the bright-field data.

[0139] Secondly, normalizing the total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region to obtain a dark-field reliability factor corresponding to each target micro-object region.

[0140] It should be noted that the longer the motion trajectory of the particle in the dark field, the more obvious the motion, and the higher the dark-field reliability.

[0141] Thirdly, determining a bright-field weight and a dark-field weight corresponding to each target micro-object region according to the bright-field reliability factor and the dark-field reliability factor corresponding to each target micro-object region.

[0142] For example, the formula for determining the bright-field weight and the dark-field weight corresponding to each target micro-object region can be as follows:

[0143] ;

[0144] ;

[0145] wherein, is the bright-field weight corresponding to the i-th target micro-object region. i is the dark-field weight corresponding to the i-th target micro-object region. is the serial number of the target micro-object region. i is the bright-field reliability factor corresponding to the i-th target micro-object region. i is the dark-field reliability factor corresponding to the i-th target micro-object region. i i

[0146] ​​​​In the fourth step, the target bubble possibility index corresponding to each target micro-object region is determined according to the product of the bright-field weight and the bright-field bubble possibility index corresponding to each target micro-object region and the dark-field weight and the dark-field bubble possibility index corresponding to each target micro-object region.

[0147] For example, the formula corresponding to the determination of the target bubble possibility index corresponding to the target micro-object region can be:

[0148] ;

[0149] wherein, is the target bubble possibility index corresponding to the i th target micro-object region. i is the serial number of the target micro-object region. i is the bright-field weight corresponding to the i th target micro-object region. is the dark-field weight corresponding to the i th target micro-object region. i is the bright-field bubble possibility index corresponding to the i th target micro-object region. is the dark-field bubble possibility index corresponding to the i th target micro-object region. i i i It should be noted that,

[0150] can be used as the weight of the bright-field weight. can be used as the weight of the dark-field weight. When the i th target micro-object region is more likely to represent a bubble, the value of i is larger. i

[0151] In the step S6, whether each target micro-object region represents a bubble is determined according to the target bubble possibility index corresponding to each target micro-object region.

[0152] For example, if the target bubble possibility index corresponding to the target micro-object region is greater than a preset bubble threshold, it is determined that the target micro-object region represents a bubble. The preset bubble threshold can be a threshold set in advance, which can be 0.7.

[0153] With reference to the above method embodiments, based on the same inventive concept, the present application provides a system for detecting appearance changes of a monoclonal antibody injection liquid after being mixed with multiple drugs, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the above computer program is executed by the processor to implement the steps of a method for detecting appearance changes of a monoclonal antibody injection liquid after being mixed with multiple drugs, which can specifically include: Figure 2

[0154] ​​​​​​​​​​The image acquisition module 201 is configured to acquire dark field images of the target solution after the monoclonal antibody injection solution is mixed with a preset number of drugs within a current time period, and a bright field image of the target solution at a current time point, wherein the current time point is an ending time point of the current time period.

[0155] The identification and screening module 202 is configured to identify a target micro-object region from the bright field image, and screen a region matched with the target micro-object region from all dark field images within the current time period as a reference micro-object region sequence corresponding to the target micro-object region.

[0156] The bright field bubble possibility indicator determination module 203 is configured to determine a bright field bubble possibility indicator corresponding to each target micro-object region according to the internal and external gray scale difference and shape feature of each target micro-object region.

[0157] The motion trajectory analysis processing module 204 is configured to perform motion trajectory analysis processing on the reference micro-object region sequence corresponding to each target micro-object region to obtain a dark field bubble possibility indicator corresponding to each target micro-object region.

[0158] The target bubble possibility indicator determination module 205 is configured to determine a target bubble possibility indicator corresponding to each target micro-object region according to the area, the bright field bubble possibility indicator and the dark field bubble possibility indicator corresponding to each target micro-object region, and a total length of a motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region.

[0159] The region judgment module 206 is configured to judge whether each target micro-object region represents a bubble according to the target bubble possibility indicator corresponding to each target micro-object region.

[0160] Figure 3 is a structural schematic diagram of a computer device provided by an embodiment of the present application. As shown in the example, Figure 3 the computer device 300 includes a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302, wherein the processor 302 executes the computer program 303, so that the computer device can execute any one of the above-mentioned monoclonal antibody injection solution and multiple drug mixing appearance change detection methods.

[0161] Based on the same inventive concept as the above method embodiment, the present application provides a server including a memory and a processor. The memory is configured to store executable program code, and the processor is configured to call and run the executable program code from the memory, so that the device executes any one of the above-mentioned monoclonal antibody injection solution and multiple drug mixing appearance change detection methods.

[0162] Based on the same inventive concept as the above method embodiments, the present application provides a computer program product comprising computer program code which, when run on a computer, causes the computer to perform any one of the above appearance change detection methods after the monoclonal antibody injection solution is mixed with multiple drugs.

[0163] Based on the same inventive concept as the above method embodiments, the present application provides a computer readable storage medium storing computer program code which, when run on a computer, causes the computer to perform any one of the above appearance change detection methods after the monoclonal antibody injection solution is mixed with multiple drugs.

[0164] In summary, by analyzing the dark field image of the target solution after mixing in the current time period and the bright field image at the current time, the present application relatively objectively quantifies a plurality of bubble feature related indexes, such as the bright field bubble possible index, the dark field bubble possible index and the target bubble possible index, and judges whether the target micro-object region represents a bubble based on the target bubble possible index corresponding to the target micro-object region, relatively objectively realizes bubble detection, to a certain extent, reduces the influence of subjective factors, thereby improving the bubble detection accuracy after the monoclonal antibody injection solution is mixed with multiple drugs, and improving the appearance change detection accuracy after the monoclonal antibody injection solution is mixed with multiple drugs.

[0165] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit it; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for detecting appearance changes after a monoclonal antibody injection is combined with multiple drugs, characterized in that, The method comprises the following steps: obtaining a dark field image of a target solution prepared by mixing a monoclonal antibody injection with a preset number of drugs in a current time period, and a bright field image of the target solution at a current time point, wherein the current time point is an ending time point of the current time period; identifying a target micro-object region from the bright field image, and screening a region matching the target micro-object region from all dark field images in the current time period as a reference micro-object region sequence corresponding to the target micro-object region; determining a bright field bubble possibility index corresponding to each target micro-object region according to an internal and external gray scale difference and a shape feature of each target micro-object region; performing motion trajectory analysis processing on the reference micro-object region sequence corresponding to each target micro-object region to obtain a dark field bubble possibility index corresponding to each target micro-object region; determining a target bubble possibility index corresponding to each target micro-object region according to an area, the bright field bubble possibility index and the dark field bubble possibility index corresponding to each target micro-object region, and a total length of a motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region; judging whether each target micro-object region represents a bubble according to the target bubble possibility index corresponding to each target micro-object region.

2. The method according to claim 1, wherein the change in appearance of the single antibody injection solution after being mixed with the plurality of drugs is detected. The identification of the target micro-object region from the bright field image comprises: performing spot detection on the bright field image by a LOG model to obtain the target micro-object region.

3. The method according to claim 1, wherein the change in appearance of the single antibody injection solution after being mixed with the plurality of drugs is detected. The screening of the region matching the target micro-object region from all dark field images in the current time period as the reference micro-object region sequence corresponding to the target micro-object region comprises: determining any target micro-object region as a marker micro-object region, and screening a region matching the marker micro-object region from a last dark field image in the current time period as a temporary micro-object region by a SIFT feature matching algorithm; screening regions matching the temporary micro-object region from all dark field images in the current time period by an optical flow method to constitute a reference micro-object region sequence corresponding to the marker micro-object region, wherein the marker micro-object region and the reference micro-object regions in the reference micro-object region sequence corresponding to the marker micro-object region represent the same object.

4. The method for detecting appearance changes after a monoclonal antibody injection is combined with multiple drugs according to claim 1, characterized in that, The determination of the bright field bubble possibility index corresponding to each target micro-object region according to the internal and external gray scale difference and the shape feature of each target micro-object region comprises: determining a shape feature index corresponding to each target micro-object region according to a centroid of each target micro-object region and different preset directions; determining an internal and external gray scale difference corresponding to each target micro-object region according to a preset neighborhood corresponding to different edge pixel points on each target micro-object region; normalizing a product between the shape feature index and the internal and external gray scale difference corresponding to each target micro-object region to obtain the bright field bubble possibility index corresponding to each target micro-object region.

5. The method according to claim 4, wherein the change in appearance of the single antibody injection solution after being mixed with the plurality of drugs is detected. The determination of the shape feature index corresponding to each target micro-object region according to the centroid of each target micro-object region and the different preset directions comprises: determining a marker micro-object region, and drawing a target straight line of the marker micro-object region in each preset direction through the centroid of the marker micro-object region with each preset direction as an extension direction. An intersection between the marked micro-object region and a target straight line of the marked micro-object region in each preset direction is determined as a target intersection line segment of the marked micro-object region in each preset direction; A length of the target intersection line segment of the marked micro-object region in each preset direction is determined as a target length of the marked micro-object region in each preset direction; A shape feature index corresponding to the marked micro-object region is determined according to a maximum value and a minimum value of the target lengths of the marked micro-object region in all preset directions.

6. The method according to claim 4, wherein the change in appearance of the single antibody injection solution after being mixed with the plurality of drugs is detected. The determining of the internal-external gray difference corresponding to each target micro-object region according to the preset neighborhood corresponding to each different edge pixel point on each target micro-object region comprises: Any one target micro-object region is determined as a marked micro-object region, and a union set of the preset neighborhoods corresponding to all edge pixel points on the marked micro-object region is determined as a target union region corresponding to the marked micro-object region; All pixel points in the target union region corresponding to the marked micro-object region that do not belong to the marked micro-object region constitute a target peripheral region corresponding to the marked micro-object region; The internal-external gray difference corresponding to the marked micro-object region is determined according to a gray difference between the marked micro-object region and the target peripheral region corresponding to the marked micro-object region.

7. The method according to claim 1, wherein the change in appearance of the single antibody injection solution after being mixed with the plurality of drugs is detected. The motion trajectory analysis processing of the reference micro-object region sequence corresponding to each target micro-object region to obtain a dark-field bubble possibility index corresponding to each target micro-object region comprises: Any one target micro-object region is determined as a marked micro-object region, and the reference micro-object region sequence corresponding to the marked micro-object region is equally divided to obtain a reference micro-object region subsegment; All reference micro-object region subsegments constitute a reference micro-object region subsegment sequence; A mean square displacement corresponding to each reference micro-object region subsegment is determined according to a position change of a reference micro-object region in each reference micro-object region subsegment in the reference micro-object region subsegment sequence, to obtain a mean square displacement sequence corresponding to the marked micro-object region; A target change feature value corresponding to the marked micro-object region is determined according to the mean square displacement sequence corresponding to the marked micro-object region; A directionality coefficient corresponding to the marked micro-object region is determined according to a motion trajectory condition corresponding to the reference micro-object region sequence corresponding to the marked micro-object region; A product between the target change feature value corresponding to the marked micro-object region and the directionality coefficient is normalized to obtain the dark-field bubble possibility index corresponding to the marked micro-object region.

8. The method according to claim 7, wherein the change in appearance of the single antibody injection solution after being mixed with the plurality of drugs is detected. The determining of the target change feature value corresponding to the marked micro-object region according to the mean square displacement sequence corresponding to the marked micro-object region comprises: A straight line fitting goodness of the marked micro-object region is obtained by performing straight line fitting on the mean square displacement sequence corresponding to the marked micro-object region; The target change feature value corresponding to the marked micro-object region is determined according to the straight line fitting goodness of the marked micro-object region, and the straight line fitting goodness and the target change feature value are in a negative correlation relationship.

9. The method according to claim 7, wherein the change in appearance of the single antibody injection solution after being mixed with the plurality of drugs is detected. The determining of the directionality coefficient corresponding to the marked micro-object region according to the motion trajectory condition corresponding to the reference micro-object region sequence corresponding to the marked micro-object region comprises: The distance between the centers of each two reference micro-object regions in the reference micro-object region sequence is determined as the target distance between each two reference micro-object regions; The target distance between the first reference micro-object region and the last reference micro-object region in the reference micro-object region sequence corresponding to the marked micro-object region and the cumulative value of the target distances between all adjacent reference micro-object regions in the reference micro-object region sequence corresponding to the marked micro-object region are used to determine the directionality coefficient corresponding to the marked micro-object region.

10. The method according to claim 1, wherein the method is characterized by the fact that the appearance change of the single antibody injection solution after being mixed with the plurality of drugs is detected. The target bubble possibility index corresponding to each target micro-object region is determined according to the area corresponding to each target micro-object region, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object region, and the total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region, and the target bubble possibility index corresponding to each target micro-object region is determined according to the area corresponding to each target micro-object region, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object region, and the total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region, and the target bubble possibility index corresponding to each target micro-object region is determined according to the area corresponding to each target micro-object region, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object region, and the total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region, and the target bubble possibility index corresponding to each target micro-object region is determined according to the area corresponding to each target micro-object region, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object region, and the total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region, and the target bubble possibility index corresponding to each target micro-object region is determined according to the area corresponding to each target micro-object region, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object region, and the total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region, and the target bubble possibility index corresponding to each target micro-object region is determined according to the area corresponding to each target micro-object region, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object region, and the total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region, and the target bubble possibility index corresponding to each target micro-object region is determined according to the area corresponding to each target micro-object region, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object region, and the total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region, and the target bubble possibility index corresponding to each target micro-object region is determined according to the area corresponding to each target micro-object region, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object region, and the total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region, and the target bubble possibility index corresponding to each target micro-object region is determined according to the area corresponding to each target micro-object region, the bright-field bubble possibility index and the dark-field bubble possibility index corresponding to each target micro-object region, and the total length of the motion trajectory corresponding to the reference micro-object region sequence corresponding to each target micro-object region, and the target bubble possibility index corresponding to each target micro-object region is determined according to the area corresponding to each ​ ​ ​ ​

Images