Method and system for analyzing melanin content based on migration control

Abstract

The application discloses a kind of based on the analysis method and system of melanin content of migration control, and the present application relates to the technical field of quantitative analysis method, and along the illumination parameter is photographed to zebra embryo, and multiple images of different angles of zebra embryo are collected;According to multiple images, corresponding stereoscopic data are determined, and time series three-dimensional model is generated according to the alignment of the stereoscopic data and time point data, to ensure the accuracy of time series three-dimensional model.Therefore, melanin area is determined based on the identification of time series three-dimensional model, and melanin three-dimensional volume is determined according to the area position of melanin area and the voxel statistics of melanin area;The migration path of melanin area is collected, and the quantitative analysis of melanin content is triggered according to the migration path of melanin area, the migration speed of melanin area and melanin three-dimensional volume, to ensure the accuracy of quantitative analysis of melanin content.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of quantitative analysis methods, and in particular to a melanin content analysis method and system based on migration control. BACKGROUND

[0002] With the development of technology, zebrafish embryos have become a common object in modern melanin research due to their transparent body and easy observation. The melanin refers to a biological pigment present in the skin, eyes and other tissues of zebrafish. In the prior art, researchers collect and process images of zebrafish to identify melanin regions and quantify melanin content by calculating parameters such as area and gray value of the melanin region. However, the prior art only stays at the two-dimensional image level and does not involve the generation and migration path of melanin, which limits the existing melanin content quantitative analysis method in deeper melanin research. SUMMARY

[0003] The present application aims to overcome the shortcomings of the prior art and provides a melanin content analysis method and system based on migration control.

[0004] The present application provides a melanin content analysis method based on migration control, which comprises:

[0005] When the zebrafish is in an online detection state, the corresponding illumination parameters are determined according to the thickness of the zebrafish embryo and the environmental conditions in which the zebrafish is located;

[0006] The zebrafish embryo is photographed along the illumination parameters, and multiple images of different angles of the zebrafish embryo are collected;

[0007] The corresponding stereoscopic data are determined according to the multiple images, and a time-series three-dimensional model is generated according to the alignment of the stereoscopic data and the time point data;

[0008] The melanin region is determined based on the recognition of the time-series three-dimensional model, and the melanin three-dimensional volume is determined according to the region position of the melanin region and the voxel statistics of the melanin region;

[0009] The migration path of the melanin region is collected, and the quantitative analysis of the melanin content is triggered according to the migration path of the melanin region, the migration speed of the melanin region and the melanin three-dimensional volume.

[0010] The present application provides a melanin content analysis system based on migration control, which is applied to the above-mentioned melanin content analysis method based on migration control. The melanin content analysis system based on migration control comprises:

[0011] The lighting parameter module is used for determining corresponding lighting parameters according to the thickness of the zebrafish embryo and the environmental conditions where the zebrafish is located when the zebrafish is in an online detection state.

[0012] The image module is used for photographing the zebrafish embryo along the lighting parameters and collecting multiple images of the zebrafish embryo at different angles.

[0013] The time-series three-dimensional model module is used for determining corresponding stereoscopic data according to the multiple images and generating a time-series three-dimensional model according to the alignment of the stereoscopic data and the time point data.

[0014] The melanin three-dimensional volume module is used for determining a melanin region based on the recognition of the time-series three-dimensional model and determining a melanin three-dimensional volume according to the area position of the melanin region and the voxel statistics of the melanin region.

[0015] The quantitative analysis module is used for collecting a migration path of the melanin region, triggering quantitative analysis of melanin content according to the migration path of the melanin region, the migration speed of the melanin region and the melanin three-dimensional volume.

[0016] Compared with the prior art, the present application has the following advantages:

[0017] In the embodiment of the present application, when the zebrafish is in an online detection state, the corresponding lighting parameters are determined according to the thickness of the zebrafish embryo and the environmental conditions where the zebrafish is located; the zebrafish embryo is photographed along the lighting parameters, and multiple images of the zebrafish embryo at different angles are collected; the corresponding stereoscopic data are determined according to the multiple images, and a time-series three-dimensional model is generated according to the alignment of the stereoscopic data and the time point data, thereby ensuring the accuracy of the time-series three-dimensional model.

[0018] Therefore, the melanin region is determined based on the recognition of the time-series three-dimensional model, and the melanin three-dimensional volume is determined according to the area position of the melanin region and the voxel statistics of the melanin region; the migration path of the melanin region is collected, and quantitative analysis of melanin content is triggered according to the migration path of the melanin region, the migration speed of the melanin region and the melanin three-dimensional volume, thereby ensuring the accuracy of the quantitative analysis of melanin content. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 is a flowchart of the melanin content analysis method based on migration control in the embodiment of the present application;

[0020] Figure 2 is a flowchart of step S11 in the melanin content analysis method based on migration control in the embodiment of the present application;

[0021] Figure 3 is a flowchart of step S12 in the melanin content analysis method based on migration control in the embodiments of the present application;

[0022] Figure 4 is a flowchart of step S13 in the melanin content analysis method based on migration control in the embodiments of the present application;

[0023] Figure 5 is a flowchart of step S14 in the melanin content analysis method based on migration control in the embodiments of the present application;

[0024] Figure 6 is a flowchart of step S15 in the melanin content analysis method based on migration control in the embodiments of the present application;

[0025] Figure 7 is a structural composition diagram of the melanin content analysis system based on migration control in the embodiments of the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0027] Please refer to Figures 1 to 7 A melanin content analysis method based on migration control is applied to the quantitative analysis of the melanin content of zebrafish; the melanin content analysis method based on migration control includes:

[0028] Step S11: When the zebrafish is in an online detection state, determine the corresponding illumination parameters according to the thickness of the zebrafish embryo and the environmental conditions in which the zebrafish is located;

[0029] Step S12: Capture the zebrafish embryo along the illumination parameters and collect multiple images of different angles of the zebrafish embryo;

[0030] Step S13: Determine the corresponding stereoscopic data according to the multiple images, and generate a time-series three-dimensional model according to the alignment of the stereoscopic data and the time point data;

[0031] Step S14: Determine the melanin region based on the recognition of the time-series three-dimensional model, and determine the melanin three-dimensional volume according to the region position of the melanin region and the voxel statistics of the melanin region;

[0032] Step S15: Collect the migration path of the melanin region, and trigger the quantitative analysis of the melanin content according to the migration path of the melanin region, the migration speed of the melanin region, and the melanin three-dimensional volume;

[0033] Reference Figure 2 In step S11, when the zebrafish is in an online detection state, corresponding illumination parameters are determined according to the thickness of the zebrafish embryo and the environmental conditions in which the zebrafish is located;

[0034] In the implementation of the present application, the specific steps are as follows:

[0035] S111: The location of the zebrafish is collected, the corresponding online detection mode is determined according to the location of the zebrafish and the species of the zebrafish, and the online detection of the zebrafish is triggered along the online detection mode;

[0036] S112: The location of the zebrafish embryo is determined based on the online detection of the zebrafish, the thickness of the zebrafish embryo is determined according to the identification of the location of the zebrafish embryo, the environmental conditions in which the zebrafish is located are collected, and the corresponding illumination parameters are determined according to the thickness of the zebrafish embryo, the environmental conditions in which the zebrafish is located, and the illumination mapping relationship.

[0037] In the embodiment of the present application, the location of the zebrafish is collected, the corresponding online detection mode is determined according to the location of the zebrafish and the species of the zebrafish, and the online detection of the zebrafish is triggered along the online detection mode, which introduces the online detection of the zebrafish.

[0038] At this time, the image of the experimental area is captured using a camera or an image sensor, and the image of the experimental area is recognized to facilitate the collection of the location of the zebrafish, and the location of the zebrafish and the species of the zebrafish are introduced.

[0039] The most suitable online detection mode is selected according to the location of the zebrafish and the species of the zebrafish, at this time, a plurality of online detection modes are preset in the system, each online detection mode is for a specific zebrafish species and the location of the zebrafish; according to the position information obtained in step one, it is determined whether the zebrafish is in the correct position of the detection area; according to the species of the zebrafish (for example, wild type or mutant, juvenile or adult), the corresponding online detection mode is selected; optionally, the corresponding online detection mode includes a multi-species zebrafish online detection mode.

[0040] Once the online detection mode is determined, the system automatically triggers the detection process, including adjusting the camera focal length, starting the light sheet microscopic imaging system, starting data collection and other steps, the system will collect the data of the zebrafish in real time or at regular intervals for subsequent analysis and processing.

[0041] Specifically, assume that a study on zebrafish melanin distribution is being conducted, and there are two types of zebrafish in the experiment: wild-type zebrafish and Albino mutant zebrafish, which have significant differences in melanin content and distribution; at the beginning of the experiment, high-definition cameras capture images of the entire experimental device, and through image recognition algorithms, the positions of each zebrafish are identified and marked on the image. It is found that two wild-type zebrafish are located at the center of the detection area, while one Albino mutant zebrafish is located at the edge.

[0042] According to the position information of the zebrafish, it is confirmed that all zebrafish are in the correct position of the detection area; since it involves two different types of zebrafish, the "multi-species zebrafish online detection mode" is selected; the "multi-species zebrafish online detection mode" can handle data of wild-type and mutant zebrafish simultaneously and distinguish them according to their characteristics.

[0043] The system automatically triggers the online detection process, and the light sheet microscopy system starts working, performing high-resolution imaging on each zebrafish. The data acquisition system collects image data in real time and stores it in the designated location of the server. It can accurately determine the position and species of the zebrafish and select the most suitable online detection mode, providing a solid foundation for subsequent data analysis and processing.

[0044] Therefore, based on the online detection of zebrafish, the area of zebrafish embryos is determined, the thickness of zebrafish embryos is determined based on the identification of the area of zebrafish embryos, the environmental conditions where zebrafish are located are collected, and the corresponding lighting parameters are determined based on the thickness of zebrafish embryos, the environmental conditions where zebrafish are located, and the lighting mapping relationship, ensuring the accuracy of the lighting parameters.

[0045] At this time, based on online detection, the specific location area of zebrafish embryos in the experimental device is accurately identified. The online detection system has provided the position information of the zebrafish, and further determines the area of the zebrafish embryos. Optionally, considering the slight position changes of the embryos due to movement or growth, the area of the zebrafish embryos is updated in real time.

[0046] By identifying the area of zebrafish embryos, the thickness is indirectly or directly measured, which is one of the key parameters for adjusting the lighting parameters. At this time, image processing techniques such as depth estimation or volume measurement are used to extract thickness information from the identified area of zebrafish embryos; if the embryo is in a specific posture (such as lying on one side), the maximum cross-sectional diameter is directly measured as an approximate value of the thickness.

[0047] Obtain the environmental conditions in which the zebrafish embryos are located, such as water quality, temperature, light, etc., which affect the imaging quality and melanin distribution; at this time, use sensors to monitor the environmental conditions inside the experimental device, such as temperature sensors, water quality monitors, etc., manually record or automatically collect these data, and associate them with the image data of the embryos.

[0048] Based on the embryo thickness, environmental conditions, and pre-set illumination mapping relationship, determine the optimal illumination parameters to reduce phototoxicity and improve imaging quality; at this time, the illumination mapping relationship is a pre-set lookup table that provides recommended illumination parameters based on embryo thickness and environmental conditions; based on the information obtained in steps two and three, query the illumination mapping relationship to determine appropriate illumination intensity, wavelength, exposure time, etc.; consider the specific needs of the experiment (such as high-contrast imaging, long-term monitoring, etc.), and fine-tune the recommended parameters.

[0049] Specifically, suppose we are studying the changes in melanin distribution of zebrafish embryos at different temperatures; in the experiment, zebrafish embryos are placed in culture dishes at different temperatures and monitored using a light sheet microscopy system; the online detection system identifies that the zebrafish embryo is roughly located in the center of the culture dish, and uses more refined image recognition algorithms, such as contour detection, to determine the precise position and shape of the embryo.

[0050] From the identified embryo area, use image processing techniques to measure its maximum cross-sectional diameter as an approximation of thickness; considering the transparency and internal structure of the embryo, select the appropriate measurement method, and obtain an estimated thickness of 0.5 millimeters.

[0051] The temperature inside the culture dish is monitored using a temperature sensor and recorded as 28 degrees Celsius; at the same time, the water quality conditions are also checked to ensure that there are no contaminants affecting the imaging quality; at this time, based on the embryo thickness (0.5 millimeters) and environmental temperature (28 degrees Celsius), the pre-set illumination mapping relationship is queried, which recommends using lower intensity blue light illumination for embryos with a thickness of 0.5 millimeters at 28 degrees Celsius to reduce phototoxicity and improve imaging contrast; based on these recommended parameters, the illumination settings of the light sheet microscopy system are adjusted, and the experimental monitoring begins; through this process, the position, thickness, and environmental conditions of the zebrafish embryo can be accurately determined, and the best illumination parameters can be selected based on this information, providing a reliable foundation for subsequent imaging and analysis.

[0052] In an embodiment of the present application, a pre-set embryo thickness matching table is provided to associate embryo thickness, environmental conditions, and illumination parameters. The embryo thickness matching table is shown in Table One:

[0053] Table One Embryo Thickness Matching Table

[0054]

[0055] Assuming the detected embryo thickness is 0.4mm and the ambient temperature is 27℃, according to the embryo thickness matching table, the illumination parameters of light intensity 1200lx, wavelength 480nm and exposure time 50ms are selected.

[0056] Reference Figure 3 In step S12, the zebrafish embryo is photographed along the illumination parameters, and multiple images of the zebrafish embryo at different angles are collected.

[0057] In the specific implementation of the present application, the specific steps are as follows:

[0058] S121: The illumination system of the zebrafish embryo is controlled based on the illumination parameters, and the zebrafish embryo is dynamically photographed based on the photographing system.

[0059] S122: In the dynamic photographing of the zebrafish embryo, multiple angle positions of the zebrafish embryo are collected, and multiple images of the zebrafish embryo at different angles are collected according to the multiple angle positions, the photographing system and the illumination system.

[0060] In the embodiment of the present application, the illumination system of the zebrafish embryo is controlled based on the illumination parameters, and the zebrafish embryo is dynamically photographed based on the photographing system, which realizes the dynamic photographing of the zebrafish embryo.

[0061] At this time, the illumination parameters are introduced, which usually include light intensity, wavelength (or color), spot size, light direction and exposure time, etc., which are selected according to the characteristics of the zebrafish embryo, the experimental requirements and the previous detection results (such as the embryo thickness and environmental conditions determined in step S112); at the same time, the illumination system of the zebrafish embryo is controlled based on the illumination parameters, and the illumination parameters are optionally adjusted in real time to respond to the movement or morphological changes of the embryo during the photographing process, so as to ensure the continuous optimal illumination conditions.

[0062] Specifically, assuming that a study on the melanin distribution of zebrafish embryo is being conducted, and the illumination parameters have been determined according to the thickness of the embryo (0.5mm) and the ambient temperature (28℃): the light intensity is 1500lx, the wavelength is 488nm (blue light, suitable for exciting melanin fluorescence), the spot size is adjusted to cover the entire embryo, and the light direction is perpendicular to the embryo plane.

[0063] A programmable LED light source system was used, and these parameters were input through its control software; the light source automatically adjusted to the specified intensity and selected the 488nm wavelength through the built-in filter; at the same time, the position and angle of the light source were ensured, so that the light spot could uniformly cover the embryo; if the system supports real-time adjustment, a feedback mechanism is also set up to fine-tune the light intensity according to the image quality (such as contrast, brightness, etc.) captured, to ensure that the best lighting conditions are maintained throughout the shooting process.

[0064] At the same time, the shooting system includes high-resolution cameras, microscopes (such as light sheet microscopes, confocal microscopes, etc.), image sensors, and related image acquisition software; dynamic shooting means continuously capturing a series of images within a period of time to record the morphological changes, movements, or other dynamic processes of the embryo, which usually involves setting appropriate frame rates (number of images captured per second) and exposure times to ensure clear images and not miss important information; at the same time, in some cases, the shooting system needs to be synchronized with the lighting system to ensure that the images are captured under the correct lighting conditions, which is achieved through hardware trigger signals or software synchronization mechanisms.

[0065] Specifically, a high-resolution CCD camera was used, connected to a light sheet microscope specially designed for high-resolution imaging of transparent samples (such as zebrafish embryos); the frame rate of the shooting system was set to 10 frames per second, and the exposure time was matched with the light source of the lighting system (assuming 50ms) to ensure that clear images were obtained at each exposure.

[0066] To ensure synchronization between shooting and lighting, a software synchronization mechanism was used; before shooting started, a synchronization program was launched, which sent trigger signals to the lighting system and the shooting system to ensure that they started working at the same time; during the entire shooting process, a series of images were continuously captured, recording the morphological changes of the zebrafish embryo at different time points, which were used for subsequent analysis, such as quantification of melanin distribution, measurement of morphological parameters, etc.

[0067] Therefore, in the dynamic shooting of zebrafish embryos, multiple angle positions of the zebrafish embryo are collected, and multiple images of the zebrafish embryo at different angles are collected according to the multiple angle positions, the shooting system, and the lighting system.

[0068] At this time, in order to fully capture the morphological features of the zebrafish embryo, multiple different angles need to be selected for shooting, which should be able to cover the main feature surfaces of the embryo, such as the front, side, top, and bottom, etc.; optionally, angle adjustment is achieved by rotating the culture dish, moving the shooting system (such as the camera or microscope), or changing the placement of the embryo; in some cases, a mechanical arm or rotating device is needed to accurately control the angle.

[0069] During the dynamic shooting process, if the embryo moves or changes in shape, the angle needs to be adjusted in real time to ensure that the required features can always be captured; at the same time, at each selected angle position, the image of the zebrafish embryo is collected using the shooting system and the lighting system; ensure that clear and high-contrast images are obtained at each angle, and the images contain sufficient detailed information.

[0070] According to the change of angle, the parameters of the lighting system need to be adjusted to ensure that the best lighting effect is obtained at different angles, which involves adjusting the light intensity, wavelength or spot size, etc.; the shooting system and the lighting system should work synchronously to ensure that the images are captured under the correct lighting conditions; at the same time, the shooting angle and related information of each image should be recorded for subsequent analysis and comparison.

[0071] Specifically, when conducting morphological research on zebrafish embryos, it is decided to shoot from four main angles (front, left side, right side and top); a rotating device is used, which can accurately rotate the culture dish to the specified angle; first, place the embryo in the culture dish and adjust the rotating device so that the front is facing the camera; then, start the shooting system and set the appropriate frame rate and exposure time; during the dynamic shooting process, a series of front angle images are captured; next, the culture dish is rotated to the left side, right side and top in turn, and the shooting process is repeated, thereby obtaining multiple groups of images at these four angles.

[0072] When collecting images at each angle, the shooting system and the lighting system work synchronously; a programmable LED light source system is used, which can automatically adjust the light intensity and wavelength according to the shooting angle; for example, when collecting images at the front angle, the light intensity is set to 1500lx and the wavelength is set to 488nm (blue light) to excite the melanin fluorescence in the embryo; when collecting images at the side and top angles, the light intensity and wavelength need to be adjusted according to the transparency and morphology of the embryo to ensure image quality.

[0073] At each angle, a series of images are continuously captured, and the shooting angle, timestamp and lighting parameters of each image are recorded, which are crucial for subsequent three-dimensional reconstruction, morphological analysis and melanin distribution research.

[0074] In an embodiment of the present application, the zebrafish embryo multi-angle shooting matching table is shown in Table Two:

[0075] Table Two: Zebrafish embryo multi-angle shooting matching table

[0076]

[0077] Suppose a zebrafish embryo is being photographed from multiple angles; according to the zebrafish embryo multi-angle photographing matching table, first adjust the photographing system to the front photographing mode (focal length 4X, exposure time 50ms), and set the illumination system to light intensity 1500lx and wavelength 488nm; then, capture a series of front angle images; next, rotate the photographing system to the left side angle and adjust the focal length to 2X, and increase the exposure time to 70ms to adapt to the change in transparency; the illumination system is also adjusted to light intensity 1200lx and wavelength 520nm accordingly; under this setting, the left side angle image is captured; similarly, repeat the above process to capture the right side and top angle images respectively, and adjust the parameters of the photographing system and the illumination system according to the matching table before each photographing.

[0078] Reference Figure 4 In step S13, the corresponding stereoscopic data is determined according to the multiple images, and the time sequence three-dimensional model is generated according to the alignment of the stereoscopic data and the time point data;

[0079] In the specific implementation of the present application, the specific steps are as follows:

[0080] S131: corresponding stereoscopic features are determined based on image segmentation of the multiple images, and the corresponding stereoscopic data is generated according to the combination of the multiple stereoscopic features;

[0081] S132: the multiple stereoscopic data are labeled with corresponding time sequence nodes, and the time point data is determined according to the matching of the multiple time sequence nodes; at this time, the time point data is aligned in the time dimension, and the multiple stereoscopic features are gradually constructed in the alignment process to generate the time sequence three-dimensional model.

[0082] In the embodiments of the present application, corresponding stereoscopic features are determined based on image segmentation of the multiple images, and the corresponding stereoscopic data is generated according to the combination of the multiple stereoscopic features, which ensures the synthesis accuracy of the stereoscopic data.

[0083] At this time, the zebrafish embryo images photographed from multiple angles are preprocessed, including denoising, contrast enhancement, distortion correction, etc., to improve the image quality; the feature regions in the images are identified using a pre-set edge detection model, which are usually related to the morphology of the zebrafish embryo; the image segmentation algorithm (such as graph cut algorithm) is applied to separate the feature regions from the background to form the segmented images.

[0084] Specifically, assuming there are three zebrafish embryo images taken from different angles: front view, left side view, and right side view; first, pre-process these images, including noise removal and contrast enhancement; then, use edge detection algorithms to identify edge features in the images, which correspond to the contours of the zebrafish embryo; finally, apply region growing algorithms to grow complete embryo regions from the identified edge features, forming segmented images that will be used in subsequent steps to extract stereo features.

[0085] Meanwhile, in the segmented images, key feature points are identified using a feature point detection model; the detected feature points are described to generate feature descriptors, which can uniquely represent the local shape information of the feature points, and determine the corresponding feature points in different angle images, thereby establishing the correspondence between feature points; according to the correspondence of the feature points, the depth information of each feature point is estimated using the principle of triangulation.

[0086] Optionally, in the segmented front, left side, and right side images, SIFT algorithm is used to detect feature points and generate feature descriptors for each feature point; then, using stereo matching algorithm, corresponding feature points are found between the front view image and the left side view image, establishing their correspondence; similarly, the correspondence of feature points between the front view image and the right side view image is also established; finally, using the principle of triangulation, the depth information of each feature point is estimated according to the correspondence of the feature points, which will be used in subsequent steps to generate stereo data.

[0087] The extracted feature points and corresponding depth information are combined to generate point cloud data; each point in the point cloud data contains three-dimensional coordinate information (X, Y, Z) to convert the point cloud data into continuous stereo data; the generated stereo data is optimized to improve the accuracy and visualization of the model.

[0088] Specifically, the extracted feature points and corresponding depth information are combined to generate a point cloud data containing the three-dimensional morphology of the zebrafish embryo; then, Poisson surface reconstruction algorithm is used to convert the point cloud data into continuous stereo data, which shows the three-dimensional morphology and surface details of the zebrafish embryo; finally, the stereo data is optimized, including steps such as smoothing the surface and filling the holes, to improve the accuracy and visualization of the model.

[0089] Therefore, the corresponding time nodes are labeled for multiple stereo data, and the time point data is determined according to the matching of multiple time nodes; at this time, the time point data is aligned in the time dimension, and the stereo construction of multiple stereo features is gradually performed in the alignment process to generate a time series three-dimensional model, ensuring the accuracy of the time series three-dimensional model.

[0090] At this time, it is ensured that a series of time-ordered stereoscopic data has been acquired; the stereoscopic data comes from zebrafish embryo shots at different time points; a unique time node marker, usually a timestamp or serial number, is assigned to each stereoscopic data to identify the data's position in the time sequence; in addition to the stereoscopic data itself, metadata related to each data point, such as shooting time, environmental conditions, embryo development stage, etc., also need to be recorded.

[0091] Corresponding feature points or regions are found between stereoscopic data at different time points using feature matching algorithms such as ICP, NDT, etc.; based on the results of feature matching, the correspondence between stereoscopic data at different time points is established, which usually involves matching the three-dimensional coordinates and / or descriptors of feature points; according to the established correspondence, time point data is generated, which describes the morphological changes of the embryo between different time points.

[0092] At the same time, the timestamps of the data need to be adjusted or time interpolation needs to be performed to fill in missing time points, ensuring that all time point data is synchronized in time and sorted in the order of the time sequence to ensure they can be processed in time order.

[0093] On the aligned time point data, stereoscopic features at each time point are gradually constructed, which connects the stereoscopic features at different time points to form a continuous time sequence three-dimensional model, which usually involves interpolation processing to smooth the morphological changes between adjacent time points; the generated time sequence three-dimensional model is optimized to improve the accuracy and visualization effect of the model.

[0094] Specifically, assume there is a zebrafish embryo stereoscopic data set containing 10 time points; each time point has a corresponding stereoscopic data representing the three-dimensional morphology of the embryo at that moment; each stereoscopic data is assigned a unique serial number as a time node marker, from T1 to T10; at the same time, the shooting time (such as hours: minutes) and embryo development stage (such as somite stage, pharyngeal pouch stage, etc.) of each time point are recorded.

[0095] ICP algorithm is used to find corresponding feature points between stereoscopic data at T1 and T2 time points, which are mainly located in the head, tail and body midline of the embryo, etc. significant locations; by comparing the three-dimensional coordinate changes of these feature points, the correspondence between T1 and T2 time points is established; similarly, feature matching and correspondence establishment are also performed for other adjacent time points; finally, a series of time point data is generated, describing the morphological changes of the embryo between T1 and T10 time points.

[0096] The timestamps of all time point data were checked and found to be in order of shooting time; therefore, no time adjustment or interpolation was needed; the time point data was confirmed to be in order of T1 to T10, ready for subsequent alignment and stereoscopic construction steps.

[0097] The stereoscopic data of each time point was gradually constructed using three-dimensional modeling software; at T1 time point, the initial morphology of the embryo was constructed, including basic structures such as head, body and tail; over time, more details were gradually added at subsequent time points, such as segmentation of somites, formation of fins, etc.; at the same time, interpolation was used to smooth the morphological changes between adjacent time points; finally, a continuous time sequence three-dimensional model was generated, showing the complete morphological change process of the embryo from T1 to T10 time points; the model was optimized to improve its accuracy and visualization effect.

[0098] Reference Figure 5 In step S14, the melanin region is determined based on the identification of the time sequence three-dimensional model, and the melanin three-dimensional volume is determined according to the region position of the melanin region and the voxel statistics of the melanin region;

[0099] In the specific implementation process of the present application, the specific steps are:

[0100] S141: divide the time sequence three-dimensional model, and generate a plurality of to-be-identified regions according to the division of the time sequence three-dimensional model, the plurality of to-be-identified regions are divided based on each time sequence node in the time sequence three-dimensional model, and the plurality of to-be-identified regions correspond to different parts of the time sequence three-dimensional model respectively;

[0101] S142: determine a plurality of melanin features according to the identification of the plurality of to-be-identified regions, determine a melanin region according to the location of the plurality of melanin features and the convergence range of the melanin features, mark the region position of the melanin region, determine the voxel statistics of the melanin region based on real-time monitoring of the melanin region, and determine the melanin three-dimensional volume based on the synthesis of the region position of the melanin region and the voxel statistics of the melanin region.

[0102] In the embodiments of the present application, the time sequence three-dimensional model is divided, and a plurality of to-be-identified regions are generated according to the division of the time sequence three-dimensional model, the plurality of to-be-identified regions are divided based on each time sequence node in the time sequence three-dimensional model, and the plurality of to-be-identified regions correspond to different parts of the time sequence three-dimensional model respectively, ensuring the accuracy of the plurality of to-be-identified regions.

[0103] At this time, the time series three-dimensional model is a collection of multiple time point three-dimensional data, each time point corresponds to a three-dimensional model, which collectively describes the dynamic changes of an object or phenomenon over time; when processing the time series three-dimensional model, it is necessary to first understand its data structure, including the number of time points, the resolution of each time point three-dimensional model, the features contained in the model, etc.

[0104] According to the research purpose and specific needs, determine the division rule of the time series three-dimensional model; the division rule is based on the interval of time points, the morphological characteristics of three-dimensional models, the spatial position relationship and other factors; the purpose of division is to divide the model into multiple regions to be identified with specific meaning or characteristics, so as to facilitate subsequent analysis and processing.

[0105] According to the determined division rule, the time series three-dimensional model is divided to extract regions with specific characteristics or located in specific positions; the result of division is a set of regions to be identified, each region corresponds to a specific part or feature in the time series three-dimensional model. At the same time, each region to be identified is marked so that they can be easily identified and processed in subsequent analysis; the marking includes assigning a unique identifier to each region, recording the position information (such as coordinate range) of the region, describing the characteristics of the region, etc.

[0106] Therefore, according to the identification of multiple regions to be identified, multiple melanin features are determined, and according to the location of the multiple melanin features and the convergence range of the melanin features, a melanin region is determined, the region position of the melanin region is marked, the voxel statistics of the melanin region are determined based on the real-time monitoring of the melanin region, and the melanin three-dimensional volume is determined based on the synthesis of the region position of the melanin region and the voxel statistics of the melanin region, which ensures the accuracy of the melanin three-dimensional volume.

[0107] At this time, in each region to be identified, image processing or machine learning algorithm is used to identify melanin features; melanin features are manifested as specific colors (such as dark brown or black), textures (such as spots or stripes), shapes (such as circles or ellipses), etc.; the identification algorithm includes color threshold segmentation, morphological operation, machine learning classifier, etc.

[0108] According to the identified melanin features, their specific positions in three-dimensional space are determined; the convergence of melanin features is analyzed, that is, their distribution and aggregation degree in space; the convergence range is evaluated by calculating the spatial density, connectivity and other indicators of melanin features.

[0109] Based on the location and convergence range of melanin features, a melanin region is determined; the melanin region is a continuous or nearly continuous region composed of multiple melanin features; each melanin region is marked to record its position information (such as coordinate range, center point coordinates, etc.).

[0110] Real-time monitoring of melanin regions, recording their changes at different time points; calculate the voxel statistics of melanin regions, including the number of voxels (i.e. the number of voxels contained in the melanin region), volume (i.e. the size of the space occupied by the melanin region) and so on; the calculation of voxel statistics involves three-dimensional reconstruction, volume measurement and other image processing techniques. At the same time, the melanin regions at different time points are registered and fused to form a continuous three-dimensional volume representation; based on the registered melanin regions and voxel statistics, the three-dimensional volume of melanin is calculated.

[0111] Specifically, suppose there is a time series three-dimensional model of zebrafish melanin patch changes, which records the three-dimensional morphology of zebrafish at different time points; the goal is to identify melanin patches and calculate their three-dimensional volume.

[0112] In each region to be identified (i.e. different parts of the zebrafish), color threshold segmentation algorithm is used to identify dark brown or black pixel points as melanin features; then, morphological operations (such as dilation and erosion) are used to smooth the boundaries of melanin features and remove noise; the spatial coordinates of each melanin feature are calculated, and their distribution and aggregation on the surface of the zebrafish are analyzed; by calculating the spatial density and connectivity of melanin features, the convergence range of melanin patches is determined.

[0113] Based on the convergence range of melanin features, the specific location of melanin patches is determined and marked as a continuous region; the coordinate range, center point coordinates and other location information of each melanin patch are recorded; the zebrafish is continuously monitored and the changes of melanin patches at different time points are recorded; using three-dimensional reconstruction technology, the voxel number and volume of each melanin patch are calculated.

[0114] The melanin patches at different time points are registered and fused to form a continuous three-dimensional volume representation; based on the registered melanin patches and voxel statistics, the three-dimensional volume of melanin is calculated, and a three-dimensional visualization model is generated to show the distribution and volume changes of melanin patches on the surface of the zebrafish.

[0115] Reference Figure 6 In step S15, the migration path of the melanin region is collected, and the quantitative analysis of melanin content is triggered according to the migration path of the melanin region, the migration speed of the melanin region and the three-dimensional volume of melanin.

[0116] In the specific implementation process of the present application, the specific steps are:

[0117] S151: Label the melanin region and determine the migration state of the melanin region according to the movement of the label of the melanin region, wherein the migration path of the melanin region is determined according to the movement route of the label of the melanin region and the area change amount of the melanin region;

[0118] S152: Determine a plurality of migration nodes according to the division of the migration path of the melanin region, and determine the migration speed of the corresponding melanin region according to the matching of the plurality of migration nodes; determine a plurality of parameter combinations based on the migration path of the melanin region, the migration speed of the melanin region, and the three-dimensional volume of the melanin, and determine a plurality of analysis features according to the plurality of parameter combinations, construct a quantitative analysis logic of the melanin content according to the plurality of analysis features, and trigger the quantitative analysis of the melanin content.

[0119] In the embodiments of the present application, the melanin region is labeled, and the migration state of the melanin region is determined according to the movement of the label of the melanin region, wherein the migration path of the melanin region is determined according to the movement route of the label of the melanin region and the area change amount of the melanin region, which ensures the accuracy of the migration path of the melanin region.

[0120] At this time, the melanin region is labeled, its migration state is determined, and the migration path is further determined. At each time point or image frame, image processing or machine learning algorithm is used to identify and label the melanin region; the label is a unique identifier (such as ID number) for distinguishing different melanin regions; at the same time, the position information of each melanin region is recorded, such as coordinate range, center point coordinate or boundary contour, etc.

[0121] Compare the position information of the melanin region at consecutive time points or image frames; if the position of a certain melanin region changes significantly (such as the moving distance of the center point coordinate exceeds a certain threshold), it is considered that the region is in a migration state; the migration state is further subdivided into starting migration, continuous migration and stopping migration stages. Optionally, compare the position information of "region 1" in the two consecutive images; if it is found that the center point coordinate of "region 1" moves from (x1, y1) to (x2, y2), and the moving distance exceeds the preset threshold (such as 5 pixel units), it is judged that "region 1" is in a migration state.

[0122] After determining that the melanin region is in a migration state, the position information of the region at different time points or image frames is tracked and recorded; the position information is connected to form a continuous path, i.e., the migration path of the melanin region; the migration path is represented as a series of points (such as center point coordinates) or line segments (such as line segments connecting adjacent time points), and optionally, the position information of the "region 1" on the subsequent image frames is tracked and recorded; the position information is connected to form a continuous path; for example, the center point coordinates of the "region 1" on three consecutive images are represented as (x1, y1), (x2, y2) and (x3, y3) respectively, and the migration path is represented as a line segment from (x1, y1) to (x2, y2) and then to (x3, y3).

[0123] Specifically, assuming there is a series of zebrafish images recording the changes of a certain melanin region over a period of time; using image processing software to process these images, the melanin region is identified and marked; by comparing the position information of the melanin region on consecutive images, it is found that the region starts to move from the initial position (x1, y1), undergoes a series of position changes, and finally reaches the position (xn, yn); connecting these position information forms a clear migration path, which not only shows the moving track of the melanin region, but also provides important information for subsequent analysis of migration speed, migration direction, etc.

[0124] Further, a plurality of migration nodes are determined according to the division of the migration path of the melanin region, and the migration speed of the corresponding melanin region is determined according to the matching of the plurality of migration nodes; based on the migration path of the melanin region, the migration speed of the melanin region and the three-dimensional volume of the melanin, a plurality of parameter combinations are determined, and according to the plurality of parameter combinations, corresponding analysis features are determined, and according to the plurality of analysis features, a quantitative analysis logic of melanin content is constructed, and the quantitative analysis of melanin content is triggered, which takes into account the migration path of the melanin region, the migration speed of the melanin region and the three-dimensional volume of the melanin, ensuring the accuracy of the quantitative analysis of melanin content.

[0125] At this time, according to the migration path of the melanin region, the migration nodes and the migration speed are determined, how to determine the analysis features based on these parameters and the three-dimensional volume of the melanin, and finally construct the quantitative analysis logic of melanin content; at this time, according to the morphological changes or specific events (such as speed changes, direction changes, etc.) on the migration path of the melanin region, a plurality of migration nodes are divided; the migration nodes are key points on the path, such as starting point, ending point, speed change point or direction turning point, etc.; each migration node records the corresponding time point and position information.

[0126] Optionally, assuming there is a migration path of melanin region, which starts from point A(x1, y1, t1), goes through a series of changes, and finally ends at point B(xn, yn, tn); on this path, it is observed that the speed has obvious acceleration or deceleration, or there is a significant turning at a certain position; according to these changes, a number of migration nodes are divided on the path, such as node 1(x2, y2, t2), node 2(x3, y3, t3), etc.

[0127] At the same time, for each migration node, the time interval and distance change between it and its adjacent nodes are calculated, so as to determine the migration speed of the section; the migration speed is the average speed, also the instantaneous speed, which depends on the accuracy requirement of analysis; the migration speed of each section is recorded and smoothed to reduce the influence of noise; optionally, for the section between node 1 and node 2, the time interval Δt = t2 - t1 and the distance change Δd (using Euclidean distance or other appropriate distance measure) are calculated; then, the migration speed v = Δd / Δt of the section is calculated; repeat this process for all sections on the path to get a series of migration speed values.

[0128] Based on the migration path of melanin region, migration speed and melanin three-dimensional volume, a plurality of analysis features are determined; the analysis features include the total length of the migration path, the average migration speed, the maximum migration speed, the minimum migration speed, the stability of the migration direction, the change amount of the melanin volume, etc., which provide a comprehensive description of the migration behavior of the melanin region; optionally, the total length of the migration path, i.e. the sum of all section distances, is calculated; the average migration speed, i.e. the average value of all section migration speeds, is calculated; the maximum and minimum migration speeds are found out, as well as their corresponding positions and time points; the stability of the migration direction is analyzed by calculating the angle of direction change or using other direction stability indicators; the value of melanin volume at each time point is recorded, and the total change amount of volume is calculated.

[0129] Further, according to a plurality of analysis features, a quantitative analysis logic of melanin content is constructed, which involves comparing the analysis features with known melanin content standards or reference ranges to determine the normal, abnormal or change trend of melanin content; the quantitative analysis logic is a simple threshold judgment, also a complex machine learning model.

[0130] Optionally, a series of thresholds are set, such as thresholds for average migration speed, thresholds for volume change amount, etc.; the calculated analysis features are compared with these thresholds to determine the state of melanin content; if a certain feature exceeds the corresponding threshold, an alarm is triggered or the area is marked as abnormal; in addition, a machine learning algorithm (such as support vector machine, random forest, etc.) is also used to train a classification model, which can automatically determine the state of melanin content according to the analysis features.

[0131] The constructed quantitative analysis logic is applied to quantitatively analyze the melanin content, and the quantitative analysis result is output, including numerical representation, trend chart, and abnormal alarm of the melanin content; at this time, the analysis features are input into the constructed quantitative analysis logic to obtain the quantitative analysis result of the melanin content; if the result shows abnormality, an alarm is triggered, and relevant information is sent to technical personnel.

[0132] In an embodiment of the present application, the present application proposes a high-efficiency and accurate zebrafish embryo melanin migration monitoring and analysis system, which integrates advanced light sheet fluorescence imaging technology (LS-FIS), artificial intelligence algorithm and dynamic correction module, aiming to realize comprehensive and dynamic monitoring and quantitative analysis of the melanin migration process in zebrafish. The technical process, advantages and key points of the present application will be described in detail below.

[0133] In the image acquisition stage, the present application adopts adaptive light sheet adjustment technology, which can dynamically adjust the illumination parameters such as light sheet thickness, intensity and position according to the thickness of zebrafish embryo, thereby minimizing the influence of phototoxicity on living samples while ensuring the imaging quality; in addition, by integrating a high-speed camera (frame rate ≥100 fps), the system can capture the dynamic migration process of melanin cells, providing high-quality time-series image data for subsequent quantitative analysis.

[0134] Based on the improved U-Net network, the system can process multi-angle light sheet images and output high-precision three-dimensional data. In this step, a graph optimization algorithm (such as SLAM technology) is introduced, which can automatically align data collected at multiple time points to generate time-consistent three-dimensional models. This innovation not only improves the accuracy of three-dimensional reconstruction, but also ensures the continuity of time-series data, providing a solid foundation for subsequent dynamic analysis.

[0135] At this time, three-dimensional reconstruction is a crucial step in the zebrafish embryo melanin migration monitoring and analysis system; in order to realize high-precision and time-consistent three-dimensional model construction, an innovative method based on improved U-Net network and graph optimization algorithm (such as SLAM technology) is adopted; the process will be described in detail below, and guidance for implementation is provided for those skilled in the art.

[0136] The improved U-Net network is optimized based on the original U-Net network to adapt to the characteristics of the light sheet images. The network uses deeper convolutional layers, larger receptive fields, and strategies such as skip connections to improve the accuracy and efficiency of image segmentation. At the same time, the network is designed with multiple input channels to handle image data from different angles simultaneously. During the training phase, the improved U-Net network is trained using a large dataset of labeled zebrafish light sheet images, which cover different developmental stages, different angles, and different lighting conditions of zebrafish embryos. Through data augmentation techniques such as rotation, scaling, and flipping, the training samples are further enriched, and the generalization ability of the model is improved. In the verification phase, the model is evaluated using an independent test dataset to ensure its segmentation accuracy and robustness. The trained improved U-Net network can accurately segment the input multi-angle light sheet images. The network can distinguish between melanin regions and background noise, thereby extracting melanocyte cells within the zebrafish embryo. The output of this step is a multi-angle two-dimensional segmentation image.

[0137] For dynamic alignment of multi-time point data, the SLAM algorithm first constructs an initial three-dimensional point cloud model using the multi-angle two-dimensional segmentation images as input. Then, the algorithm continuously adjusts the positions of the points in the point cloud model through an iterative optimization process to minimize the reprojection error between different time points and different angle images. This process achieves automatic alignment of multi-time point data and generates a temporally consistent three-dimensional model. To improve the accuracy and efficiency of alignment, various optimization strategies are used, such as using sparse representation to reduce computational load, introducing prior knowledge (such as the anatomical structure of zebrafish) to constrain the optimization process, and using parallel computing techniques to speed up the iteration process.

[0138] For those skilled in the art, implementing light sheet image three-dimensional reconstruction based on improved U-Net network and graph optimization algorithm requires the following steps: collecting and labeling a large amount of zebrafish light sheet image data, including images from different angles, different time points, and different lighting conditions; designing an improved U-Net network structure and training it using the prepared dataset; adjusting network parameters during training to optimize segmentation accuracy; segmenting input multi-angle light sheet images using the trained network to extract melanin regions; using SLAM algorithm for automatic alignment of multi-time point data using the segmented two-dimensional images as input; adjusting optimization strategies during alignment to improve accuracy and efficiency according to actual needs; constructing a three-dimensional point cloud model based on the aligned data and performing post-processing (such as smoothing and denoising) to improve model quality.

[0139] Further, to achieve accurate segmentation and quantitative analysis of melanin regions, a multi-task deep learning model (based on an improved version of Mask R-CNN) is constructed; this model can simultaneously achieve semantic segmentation (distinguish melanin regions from background noise), volume calculation (quantify melanin three-dimensional volume based on voxel statistics), and dynamic tracking (analyze melanin migration path and rate through a time series model); to improve the generalization ability of the model, a transfer learning strategy is adopted, which first pre-trains on public biological image datasets (such as ImageNet-Zebrafish), and then fine-tunes on a small number of labeled samples.

[0140] To reduce artifacts caused by motion of live samples during long-term imaging, a motion compensation algorithm based on optical flow is designed; this algorithm can correct the displacement of the sample in real time, ensuring the accuracy of the imaging data; in addition, reinforcement learning algorithms are combined to optimize imaging parameters (such as exposure time, light sheet intensity) to further improve the signal-to-noise ratio and ensure imaging quality.

[0141] Specifically, collect imaging data of live samples such as zebrafish, including image sequences of consecutive frames and corresponding imaging parameters; select an appropriate optical flow algorithm and implement and optimize it; at the same time, design displacement correction algorithms and post-processing steps; define state space, action space and reward function, design policy network structure, and use collected data for training; during training, continuously adjust model parameters and optimize strategies; integrate the optical flow motion compensation algorithm and reinforcement learning model into the imaging system and test and verify; evaluate the effectiveness and performance of the algorithm through comparative experiments.

[0142] At this time, optical flow is a method to describe the motion velocity vector of pixels in an image; in live imaging, optical flow algorithms are used to estimate the displacement of samples between consecutive frames; optical flow algorithms calculate the motion vector of each pixel by analyzing the brightness changes of pixels in image sequences, thus achieving real-time tracking of sample motion; the motion compensation algorithm based on optical flow is implemented through the following steps: pre-processing of consecutive frame imaging data, including denoising, contrast enhancement, etc., to improve the accuracy of optical flow calculation; using optical flow algorithms to calculate the pixel motion vector between consecutive frames; commonly used optical flow algorithms include Lucas-Kanade algorithm, Farneback algorithm, etc.; according to the calculated motion vector, each frame of image is corrected for displacement, thus eliminating artifacts caused by sample motion; post-processing of corrected images, such as smoothing filtering, denoising, etc., to further improve image quality.

[0143] A reinforcement learning model is designed, including state space, action space, reward function and policy network; state space: including image features of the current frame, historical imaging parameters and other information; action space: the value range of the imaging parameters, such as the adjustment step of the exposure time, the increase and decrease of the light slice intensity, etc.; reward function: the reward value is defined according to the imaging quality (such as signal-to-noise ratio); the higher the signal-to-noise ratio, the greater the reward value; policy network: a deep learning model (such as a convolutional neural network) is used to approximate the optimal policy, that is, to select the optimal action according to the current state; through interaction with the environment, the reinforcement learning model continuously learns and optimizes the imaging parameters; during the training process, the policy gradient algorithm is used to update the parameters of the policy network to maximize the cumulative reward value; at the same time, in order to speed up the convergence speed and improve the optimization effect, common techniques in reinforcement learning such as experience replay, target network, etc. are used.

[0144] Please refer to Figure 7 , Figure 7 is a structural composition schematic diagram of the melanin content analysis system based on migration control in the embodiment of the application; the melanin content analysis system based on migration control comprises:

[0145] The illumination parameter module 21 is configured to determine corresponding illumination parameters according to the thickness of the zebrafish embryo and the environmental conditions in which the zebrafish is located when the zebrafish is in an online detection state.

[0146] The image module 22 is configured to shoot the zebrafish embryo along the illumination parameters and collect multiple images of the zebrafish embryo at different angles.

[0147] The time sequence three-dimensional model module 23 is configured to determine corresponding stereoscopic data according to the multiple images and generate a time sequence three-dimensional model according to the alignment of the stereoscopic data and time point data.

[0148] The melanin three-dimensional volume module 24 is configured to determine a melanin region based on the recognition of the time sequence three-dimensional model and determine a melanin three-dimensional volume according to the region position of the melanin region and the voxel statistical amount of the melanin region.

[0149] The quantitative analysis module 25 is configured to collect a migration path of the melanin region, trigger quantitative analysis of the melanin content according to the migration path of the melanin region, the migration speed of the melanin region and the melanin three-dimensional volume.

[0150] Any combination of the technical features of the above embodiments is possible, and in order to make the description concise, not all combinations of the technical features in the above embodiments are described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present application.

Claims

1. An analysis method of melanin content based on migration control, characterized by, The application relates to an online detection method for zebrafish embryos, which comprises the following steps: When the zebrafish is in an online detection state, corresponding illumination parameters are determined according to the thickness of the zebrafish embryo and the environmental conditions in which the zebrafish is located; The zebrafish embryo is photographed along the illumination parameters, and a plurality of images of the zebrafish embryo at different angles are collected; Corresponding stereoscopic data are determined according to the plurality of images, and a time-series three-dimensional model is generated according to the alignment of the stereoscopic data and time point data, which comprises the following steps: corresponding stereoscopic features are determined based on image segmentation of the plurality of images, and the corresponding stereoscopic data are generated according to the combination of the plurality of stereoscopic features; in the segmented images, key feature points are identified by using a feature point detection model; feature descriptors are generated by describing the detected feature points, the feature descriptors can uniquely represent the local morphological information of the feature points, and corresponding feature points are found in images at different angles; the extracted feature points and corresponding depth information are combined to generate point cloud data; each point in the point cloud data contains three-dimensional coordinate information, so as to convert the point cloud data into continuous stereoscopic data; the plurality of stereoscopic data are marked with corresponding time-series nodes, and time point data are determined according to the matching of the plurality of time-series nodes; at this time, the time point data are aligned in the time dimension, and the stereoscopic features are gradually constructed in the alignment process to generate the time-series three-dimensional model; the stereoscopic data at each time point are gradually constructed; at a T1 time point, the preliminary morphology of the embryo is constructed, including a head, a body and a tail; with the passage of time, more details are gradually added in subsequent time points, including differentiation of body segments and formation of fins; meanwhile, the morphological changes between adjacent time points are smoothed by using interpolation processing; finally, a continuous time-series three-dimensional model is generated; Melanin regions are determined based on the recognition of the time-series three-dimensional model, and a melanin three-dimensional volume is determined according to the region position of the melanin region and the voxel statistical amount of the melanin region; A migration path of the melanin region is collected, and quantitative analysis of melanin content is triggered according to the migration path of the melanin region, the migration speed of the melanin region and the melanin three-dimensional volume.

2. The method of claim 1, wherein the melanin content is analyzed based on the migration control. When the zebrafish is in an online detection state, corresponding illumination parameters are determined according to the thickness of the zebrafish embryo and the environmental conditions in which the zebrafish is located, which comprises the following steps: The position of the zebrafish is collected, a corresponding online detection mode is determined according to the position of the zebrafish and the species of the zebrafish, and online detection of the zebrafish is triggered along the online detection mode; at this time, according to the position information of the zebrafish, it is confirmed that all the zebrafish are in the correct position of the detection area; since two different species of zebrafish are involved, the "multi-species zebrafish online detection mode" is selected; Based on the online detection of zebrafish, the location of the zebrafish embryo is determined, the thickness of the zebrafish embryo is determined according to the identification of the location of the zebrafish embryo, the environmental conditions where the zebrafish is located are collected, and the corresponding lighting parameters are determined according to the thickness of the zebrafish embryo, the environmental conditions where the zebrafish is located, and the lighting mapping relationship; the lighting mapping relationship is a pre-set lookup table that provides recommended lighting parameters based on embryo thickness and environmental conditions; query the lighting mapping relationship to determine the appropriate lighting intensity, wavelength, and exposure time.

3. The method of claim 1, wherein the melanin content is analyzed based on the migration control. The zebrafish embryo is photographed along the lighting parameters, and multiple images of the zebrafish embryo at different angles are collected, including: The lighting system of the zebrafish embryo is controlled based on the lighting parameters, and the zebrafish embryo is dynamically photographed based on the shooting system; the lighting parameters are adjusted in real time to respond to the movement or morphological changes of the embryo during the shooting process, ensuring the best lighting conditions continuously; During the dynamic shooting of the zebrafish embryo, multiple angle positions of the zebrafish embryo are collected, and multiple images of the zebrafish embryo at different angles are collected based on the multiple angle positions, the shooting system, and the lighting system; the shooting system and the lighting system should work synchronously to ensure that the images are captured under the correct lighting conditions.

4. The method of claim 1, wherein the melanin content is analyzed based on the migration control. Based on the identification of the time series three-dimensional model, the melanin region is determined, and the melanin three-dimensional volume is determined according to the region position of the melanin region and the voxel statistics of the melanin region, including: The time series three-dimensional model is divided, and multiple to-be-identified regions are generated based on the division of the time series three-dimensional model, the multiple to-be-identified regions are divided based on each time series node in the time series three-dimensional model, and the multiple to-be-identified regions correspond to different parts of the time series three-dimensional model respectively; the time series three-dimensional model is a collection of multiple time point three-dimensional data, each time point corresponds to a three-dimensional model, and these three-dimensional models collectively describe the dynamic changes of an object or phenomenon over time; each to-be-identified region is labeled, including assigning a unique identifier to each region, recording the position information of the region, and describing the characteristics of the region.

5. The method of claim 4, wherein the melanin content is analyzed based on the migration control. 5 Based on the identification of the time series three-dimensional model, the melanin region is determined, and the melanin three-dimensional volume is determined according to the region position of the melanin region and the voxel statistics of the melanin region, further including: Based on the identification of the multiple to-be-identified regions, multiple melanin features are determined, and the melanin region is determined according to the location of the multiple melanin features and the convergence range of the melanin features, the region position of the melanin region is marked, the voxel statistics of the melanin region are determined based on real-time monitoring of the melanin region, and the melanin three-dimensional volume is determined based on the synthesis of the region position of the melanin region and the voxel statistics of the melanin region; the melanin region is a continuous or nearly continuous region composed of multiple melanin features; each melanin region is labeled, and its position information is recorded; at the same time, the melanin regions at different time points are registered and fused to form a continuous three-dimensional volume representation; based on the registered melanin region and the voxel statistics, the three-dimensional volume of the melanin is calculated.

6. The method of claim 1, wherein the melanin content is analyzed based on the migration control. The migration path of the melanin region is collected according to the migration path of the melanin region, the migration speed of the melanin region, and the three-dimensional volume of the melanin to trigger quantitative analysis of the melanin content, including: The melanin region is marked, and the migration state of the melanin region is determined according to the movement of the mark of the melanin region. In the migration state of the melanin region, the migration path of the melanin region is determined according to the movement route of the mark of the melanin region and the area change amount of the melanin region. After determining that the melanin region is in the migration state, the position information of the region at different time points or image frames is tracked and recorded. The position information is connected to form a continuous path, that is, the migration path of the melanin region.

7. The method according to claim 6, wherein the melanin content is analyzed based on the migration control. The migration path of the melanin region is collected according to the migration path of the melanin region, the migration speed of the melanin region, and the three-dimensional volume of the melanin to trigger quantitative analysis of the melanin content, including: According to the division of the migration path of the melanin region, a plurality of migration nodes are determined, and the migration speed of the corresponding melanin region is determined according to the matching of the plurality of migration nodes. Based on the migration path of the melanin region, the migration speed of the melanin region, and the three-dimensional volume of the melanin, a plurality of parameter combinations are determined, and corresponding analysis features are determined according to the plurality of parameter combinations. The quantitative analysis logic of the melanin content is constructed according to the plurality of analysis features, and the quantitative analysis of the melanin content is triggered. The migration node is a key point on the path, including the starting point, the ending point, the speed change point or the direction turning point. Each migration node records the corresponding time point and position information. The quantitative analysis logic is constructed to quantitatively analyze the melanin content, and the quantitative analysis result is output. The quantitative analysis result includes numerical representation, trend chart and abnormal alarm of the melanin content.

8. An analysis system for melanin content based on migration control, characterized by, The melanin content analysis system based on migration control is applied to the melanin content analysis method based on migration control as claimed in any one of claims 1-7, and the melanin content analysis system based on migration control includes: An illumination parameter module is configured to determine corresponding illumination parameters according to the thickness of the zebrafish embryo and the environmental conditions in which the zebrafish is located when the zebrafish is in an online detection state; An image module is configured to capture multiple images of the zebrafish embryo at different angles along the illumination parameters; A time-series three-dimensional model module is configured to determine corresponding stereoscopic data according to the multiple images, and generate a time-series three-dimensional model according to the alignment of the stereoscopic data and time point data; A melanin three-dimensional volume module is configured to determine a melanin region based on the recognition of the time-series three-dimensional model, and determine a melanin three-dimensional volume according to the area position of the melanin region and the voxel statistical amount of the melanin region; A quantitative analysis module is configured to collect the migration path of the melanin region, and trigger quantitative analysis of the melanin content according to the migration path of the melanin region, the migration speed of the melanin region, and the three-dimensional volume of the melanin.

Images