CN115641364A - Embryo division cycle intelligent prediction system and method based on embryo dynamics parameters - Google Patents
Embryo division cycle intelligent prediction system and method based on embryo dynamics parameters Download PDFInfo
- Publication number
- CN115641364A CN115641364A CN202211654884.3A CN202211654884A CN115641364A CN 115641364 A CN115641364 A CN 115641364A CN 202211654884 A CN202211654884 A CN 202211654884A CN 115641364 A CN115641364 A CN 115641364A
- Authority
- CN
- China
- Prior art keywords
- embryo
- roi
- image
- images
- optical flow
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Image Analysis (AREA)
- Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)
Abstract
The invention provides an embryo splitting period intelligent prediction system and method based on embryo dynamics parameters. According to the invention, YOLO3 is matched and fused with the template, so that the impurity interference outside the embryo area in the image is eliminated, and Lucas-Kanade optical flow is used for quantifying the change of the embryo for the first time according to the optical flow characteristics, thereby providing a more convenient and more intuitive data support for an embryologist to monitor the division period of the embryo in the embryo development process.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to an embryo division period intelligent prediction system and method based on embryo dynamics parameters.
Background
The division time in the embryo development process is the main embodiment of the development potential of the embryo, and the embryo with higher transplantation development potential can improve the pregnancy rate of patients. The time difference incubator can provide a stable in-vitro culture environment for the embryo, and has the function of periodically and continuously acquiring the whole process image of the embryo in-vitro development. Embryologists compare the movement change of embryos in the time difference shooting images, locate the division cycle, combine the common consensus of embryo scoring at home and abroad, and select the embryos with the best quality for patients to transplant. In order to solve the problem of timeliness of positioning the embryo division cycle by an embryologist, the activity degree in the embryo development process can be monitored by means of an image processing technology, and the embryologist is further helped to finish the positioning work of the embryo division cycle rapidly and accurately.
However, the image shot by the time difference incubator includes an embryo region and a background region, the embryo kinetic parameters in the embryo development process are important indexes for judging the embryo activity degree, the embryo activity degree reflects the quality of the embryo, and because the background region in the image shot by the time difference incubator has interference factors such as impurities, light sources and the like, a computer cannot accurately and intelligently obtain the embryo kinetic parameters through calculation, so that the purpose of monitoring the change in the embryo development process is achieved.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an embryo division period intelligent prediction system and method based on embryo kinetic parameters, aiming at describing the intensity of an embryo in the change process by using an image processing technology and further monitoring the change of the embryo in development.
In order to achieve the above object, the present invention provides an intelligent embryo division cycle prediction system based on embryo dynamics parameters, which is characterized in that the system comprises:
an image preprocessing module: the image preprocessing module is used for preprocessing the image acquired by the time difference incubator;
and (3) network model: for storing a YOLO3 network model for outputting images with and without embryo ROI regions;
image database: the time difference incubator is used for storing images acquired by the marked time difference incubator, and comprises a training set, a verification set and a test set;
a change detection analysis module: the system is used for analyzing an image with an embryo ROI (region of interest) by using a YOLO3 network model, positioning the embryo ROI in a subsequent image through template matching, performing optical flow analysis on the embryo ROI, and outputting division characteristics and corresponding time points in the embryo development process;
an output module: and the system is used for outputting a prediction result, wherein no embryo is output for the image without the embryo ROI area, the change of the optical flow value is analyzed for the image with the embryo ROI area, whether the embryo is split during the development process is output, and if the embryo is split, the splitting characteristic and the corresponding time point during the development process are output.
Further, the change detection analysis module comprises an embryo positioning unit, an ROI extraction unit, an optical flow value calculation unit and a division time point analysis unit;
the embryo positioning unit: inputting images in the embryo development sequence into a trained YOLO3 model according to a time sequence, outputting whether embryos exist in the images or not, if the embryos exist, outputting the images to an ROI extraction unit and simultaneously outputting the positions of ROI areas of the embryos in the images, otherwise, outputting the images without the ROI areas;
the ROI extraction unit: intercepting an ROI (region of interest) of an embryo in an image according to the position of the embryo obtained by the embryo positioning unit, calculating an edge image of the ROI of the embryo and an edge image of the embryo at the next moment by adopting a Canny edge operator, calculating the position corresponding to the embryo at the next moment by taking the edge image of the ROI of the embryo as a template, further obtaining the ROI of the embryo at the next moment, taking the ROI of the embryo as a new template, repeating the steps, extracting the ROI of the embryo in all the images, and outputting the ROI to an optical flow value calculating unit;
the optical flow value calculating unit: arranging the extracted embryo ROI areas in sequence according to embryo development time, and calculating the optical flow value of two adjacent frames of images by adopting a Lucas-Kanade method;
the splitting time point analyzing unit: and drawing an optical flow curve according to the embryo development time, when the optical flow value is changed to be larger than a preset value, keeping the embryo in a splitting state, recording a corresponding frame image as a splitting characteristic frame, and recording a corresponding splitting time point.
Further, the division time point analysis unit arranges the image light flow values in sequence according to the embryo development time and records the values asv 1 ,v 2 ,…, v l-1 ，lRepresenting the number of the collected images, and taking m as the size of a sliding window to calculatev 1 ,v 2 ,…,v m Is recorded as the mean optical flow value of the first windowA 1 Calculatingv 2 ,v 3 ,…,v m+1 Is recorded as the average optical flow value of the second windowA 2 Repeat until calculatingv l m- , v l m-+1 ,…, v l-1 Is taken as the average optical flow value of the last windowA l m- ；
The average light flow values exceeding the set value are taken and arranged according to the time sequence, the sliding window corresponding to the first average light flow value contains the embryo splitting time, and the time corresponding to the last light flow value contained in the window is taken as the first timejThe point in time of the secondary splitting is,j=1,2,3。
further, the formula of the optical flow value calculation unit calculating the optical flow value is as follows:
whereinnThe number of all pixel points in the ROI of the embryo,x i ,y i , i=1,2,…,nrespectively, in the ROI region of the embryo of the previous frameiThe abscissa and the ordinate of the point are,I xi , I yi , i=1,2,…,nrespectively represent the ROI area of the embryo in the previous frame at the point (x i ,y i ) In the gray scale gradient, Δ, in the x and y directionsI i , i=1,2,…,nShowing that the ROI area of two adjacent frames is in the secondiThe difference in the gray level of the dots,v x , v y respectively representing the light flow values of the ROI areas of the two adjacent frames of images in the x direction and the y direction, and V representing the light flow values of the ROI areas of the two adjacent frames of images.
Furthermore, after the embryo positioning unit outputs the first image with the embryo, the detection is stopped, all subsequent images are directly output to the ROI extracting unit, and the ROI extracting unit extracts the ROI of the embryo in all the images.
Furthermore, the image preprocessing module carries out preprocessing in such a way that a Gaussian filter operator is adopted to eliminate noise in the embryo ROI area, the edge enhancement images in the x direction and the y direction are respectively calculated through a Sobel operator, and the edge enhancement in the images is completed in a weighted average mode.
The invention also provides a method of the embryo division cycle intelligent prediction system based on the embryo dynamics parameters, which is characterized by comprising the following steps:
s1, collecting images of a time difference incubator, and inputting the images into an image preprocessing module for image preprocessing;
s2, inputting the preprocessed image into the change detection and analysis module, wherein an embryo positioning unit in the change detection and analysis module divides the image into an image with an embryo ROI (region of interest) and an image without the embryo ROI, the former goes to the step S3, and the latter goes to the step S6;
s3, extracting ROI (region of interest) of embryos in all images by the change detection analysis module;
s4, the change detection and analysis module arranges the extracted embryo ROI areas in sequence according to embryo development time, and calculates the optical flow values of two adjacent frames of images by adopting a Lucas-Kanade method;
s5, the change detection and analysis module draws an optical flow curve according to the embryo development time, when the optical flow value is changed to be larger than a preset value, the embryo is in a splitting state, a corresponding frame image is recorded as a splitting characteristic frame, and a corresponding splitting time point is recorded;
s6, the output module outputs the splitting state of each frame of image, outputs the non-embryo state of the image without the embryo ROI area, analyzes the change of the optical flow value of the image with the embryo ROI area, outputs whether the embryo is split in the development process, and outputs the second step if the embryo is splitjThe secondary split and the corresponding split time point,j=1,2,3。
preferably, the images of the time difference incubator collected in step S1 are all images collected in D1-D3 cycles for the same embryo.
Preferably, the method for extracting the embryo ROI region in step S3 includes capturing the ROI region of the embryo in the first frame image, calculating an edge image of the embryo ROI region and an edge image of the embryo at the next time by using a Canny edge operator, calculating a position corresponding to the embryo at the next time by using the edge image of the embryo ROI region as a template, further obtaining the ROI region of the embryo at the next time, performing the next round of calculation by using the embryo ROI region as a new template, and repeating the steps to extract the ROI regions of the embryos in all the images.
The invention further proposes a computer-readable storage medium, in which a computer program is stored, which computer program, when being executed by a processor, carries out the above-mentioned method.
In the development process of the embryo in the time difference incubator, a light source needs to be provided for acquiring an image, however, due to the interference of the movement of a machine and the like, the light source can generate certain shaking, and the calculation of the light flow value of the embryo can be influenced. The invention adopts a method of matching and combining the YOLO3 network and the template, accurately positions the embryo region in the embryo development sequence, well solves the influence of light source shaking on the calculation of the embryo optical flow value, and more accurately reflects the activity of the embryo.
The invention has the beneficial effects that:
1. the system provided by the invention is used for carrying out optical flow analysis on the preprocessed embryo ROI area and accurately outputting the division characteristics and the corresponding time points in the embryo development process.
2. The method can eliminate the interference factors such as impurities, light sources and the like in the background area in the images shot by the incubator due to the time difference, and accurately and intelligently obtain the embryo dynamics parameters by a computer calculation method.
3. The optical flow algorithm adopted by the invention is used for depicting the intensity of the embryo in the change process by calculating the velocity vector size of the corresponding pixel point in the embryo area in the adjacent image, and further is used for monitoring the change in the embryo development and assisting an embryologist to complete the rapid positioning of the embryo division time.
Drawings
FIG. 1 is a block diagram of the structure of an embryo division cycle intelligent prediction system based on embryo dynamics parameters;
FIG. 2 is a block diagram of the variation detection and analysis module of FIG. 1;
FIG. 3 is a flow chart of a method in an embodiment;
FIG. 4 is a first image frame with an ROI area of an embryo according to an embodiment;
FIG. 5 is a diagram showing the extraction result of ROI of embryo in the example;
FIG. 6 is a characteristic curve of cleavage during the development of an embryo in the example, wherein t1, t2 and t3 are marked to indicate the time points of the first, second and third cleavage of the embryo, respectively.
Detailed Description
The invention is described in further detail below with reference to the figures and the specific embodiments.
As shown in FIG. 1, the invention relates to an embryo splitting cycle intelligent prediction system based on embryo dynamics parameters, which comprises an image preprocessing module, a network model, an image database, a change detection analysis module and an output module, wherein:
an image preprocessing module: the image preprocessing module is used for preprocessing the image acquired by the time difference incubator; the time difference incubator shoots embryos at certain time intervals in the embryo development process, and in the actual shooting process, the time difference incubator has external interference to influence the brightness, the definition and the like of images; the image preprocessing method comprises the steps of eliminating noise in an embryo ROI area by adopting a Gaussian filter operator, respectively calculating edge enhancement images in the x direction and the y direction by using a Sobel operator, and completing edge enhancement in the images by adopting a weighted average mode;
and (3) network model: the system is used for storing a YOLO3 network model, and the YOLO3 network model is trained by images of a training set and passes verification and testing and is used for outputting images with and without an embryo ROI area;
image database: the time difference incubator is used for storing images acquired by the marked time difference incubator, and comprises a training set, a verification set and a test set; the invention collects images developed to the third day and organizes a plurality of embryologists to label embryos. Dividing the marked images into a training set, a verification set and a test set, and carrying out preprocessing operations such as turning, rotating, translating and the like on the images in the training set to expand a data set;
a change detection analysis module: the system is used for analyzing the image with the embryo ROI area by using a YOLO3 network model, positioning the embryo ROI area in a subsequent image through template matching, performing optical flow analysis on the embryo ROI area, and outputting the division characteristics and the corresponding time points in the embryo development process;
an output module: and the system is used for outputting a prediction result, wherein no embryo is output for the image without the embryo ROI area, the change of the optical flow value is analyzed for the image with the embryo ROI area, whether the embryo is split during the development process is output, and if the embryo is split, the splitting characteristic and the corresponding time point during the development process are output.
Specifically, as shown in fig. 2, the change detection analysis module includes an embryo localization unit, an ROI extraction unit, an optical flow value calculation unit, and a division time point analysis unit. Wherein:
an embryo positioning unit: inputting images in the same culture dish embryo development sequence into a trained YOLO3 model, outputting whether an embryo exists in the images, if so, outputting the images to an ROI extraction unit and simultaneously outputting the position of an embryo ROI area in the images, otherwise, outputting the images without the embryo ROI area; when the YOLO3 model outputs the first image, the embryo is formed, the calculation is stopped, and all subsequent images are directly output to the ROI extracting unit.
An ROI extraction unit: and intercepting an ROI (region of interest) of the embryo in the image according to the position of the embryo obtained by the embryo positioning unit, calculating an edge image of the ROI of the embryo and an edge image of the embryo at the next moment by adopting a Canny edge operator, calculating the position of the embryo corresponding to the embryo at the next moment by taking the edge image of the ROI of the embryo as a template, further obtaining the ROI of the embryo at the next moment, taking the ROI of the embryo as a new template, repeating the steps, extracting the ROI of the embryo in all the images, and outputting the ROI to the optical flow value calculating unit.
An optical flow value calculation unit: arranging the extracted embryo ROI areas in sequence according to embryo development time, and calculating the optical flow values of two adjacent frames of images by adopting a Lucas-Kanade method; the formula of the optical flow value calculation unit for calculating the optical flow value is as follows:
whereinnThe number of all pixel points in the ROI of the embryo,x i ,y i , i=1,2,…,nrespectively, in the ROI region of the embryo of the previous frameiThe abscissa and the ordinate of the point are,I xi , I yi , i=1,2,…,nrespectively represent the ROI area of the embryo in the previous frame at the point (x i ,y i ) In the x-and y-directions of the gray scale gradient, ΔI i , i=1,2,…,nShowing that the ROI areas of two adjacent frames of embryos are iniThe difference in the gray levels of the dots,v x , v y respectively indicate the optical flow values of the ROI area of the two adjacent frames of embryos in the x direction and the y direction, and V indicates the optical flow value of the ROI area of the two adjacent frames of embryos.
Splitting time point analysis unit: the optical flow values of the images are arranged in sequence according to the time of embryo development and are recorded asv 1 ,v 2 ,…, v l-1 ，lRepresenting the number of images acquired, m is the size of the sliding window (m is 15 in this example), and the calculation is performedv 1 ,v 2 ,…,v m Is taken as the average optical flow value of the first windowA 1 Calculatingv 2 ,v 3 ,…,v m+1 Is taken as the average optical flow value of the second windowA 2 Repeat until calculatingv l m- , v l m-+1 ,…, v l-1 Is taken as the average optical flow value of the last windowA l m- 。
The continuous variation time of the embryo is gradually increased along with the increase of the division times of the embryo, and the variation time of the optical flow value of the embryo is gradually increased and recordedk j Is as followsj（jNumber of images (in this example), in which optical flow value changes at time of 1,2,3) divisionk 1 ,k 2 ,k 3 The values are respectively 8,12 and 15). If it isA i ,A i+1 ,A i+2 ,…,A i+kj (i=1,2,…,l-m-kj) All exceed the set value sigma (according to expert experience knowledge, sigma takes 17.8 in the example), the values are determinedA i Calculating the time when the window first contains the embryo division time, and the time corresponding to the last optical flow value contained in the window is the second timejTime point of secondary division, notet j (j=1,2,3)。
The change in the embryo development process is reflected as the change of pixel points in the embryo region in the two adjacent images. The optical flow algorithm delineates the intensity of the embryo in the changing process by calculating the velocity vector of the corresponding pixel point in the embryo area in the adjacent image, and is further used for monitoring the change in the embryo development and assisting an embryologist in completing the rapid positioning of the embryo division time.
Considering the interference of impurities such as granular cells and the like around the embryo in the image on the calculation of the Lucas-Kanade optical flow method, the invention adopts a method of matching YOLO3 with the template, and only takes the ROI area of the embryo in the image as the analysis target of the Lucas-Kanade optical flow method, thereby more accurately depicting the change of the embryo.
Based on the system, as shown in fig. 3, the embryo division cycle intelligent prediction method based on the embryo dynamics parameters, provided by the invention, comprises the following steps:
s1, collecting images of a time difference incubator, and inputting the images into an image preprocessing module for image preprocessing; the images are all images collected in a D1-D3 period for the same embryo;
s2, inputting the preprocessed first frame image into an embryo positioning unit, dividing the image into an image with an embryo ROI (region of interest) and an image without the embryo ROI, turning to the step S3 to output the specific position of the embryo in the image for the next calculation, turning to the step S6 to perform no subsequent calculation, and showing a first frame image with the embryo ROI in the graph 4;
s3, extracting ROI areas of embryos in all the images by a change detection and analysis module, wherein the extraction result is shown in FIG. 5;
the change detection analysis module obtains embryo positions from an embryo ROI (region of interest) in an image output by the network model, intercepts the ROI of the embryo in the image, calculates an edge image of the ROI and an edge image of the embryo at the next moment by adopting a Canny edge operator, calculates the embryo position corresponding to the embryo at the next moment by taking the edge image of the ROI as a template, further obtains the ROI of the embryo at the next moment, performs the next round of calculation by taking the ROI of the embryo as a new template, and repeatedly extracts the ROI of the embryo in all the images;
s4, the change detection and analysis module arranges the extracted embryo ROI areas in sequence according to embryo development time, and a Lucas-Kanade method is adopted to calculate the optical flow value of two adjacent frames of images;
s5, drawing an optical flow curve by the change detection and analysis module according to the embryo development time, when the optical flow value is changed to be larger than a preset value, the embryo is in a split state, recording a corresponding frame image as a split characteristic frame, and recording a corresponding split time point;
according to the statistical analysis of the light flow value and the embryo change, when the embryo is in the split state, the light flow value is increased sharply to reach the peak value, when the split state is completed, the light flow value is reduced gradually, and the whole splitting process presents the shape of the peak. Setting a threshold value of a peak value according to the change characteristic of the optical flow during embryo division, when the optical flow is higher than the threshold value, enabling the embryo to be in the division process, and outputting the time corresponding to the maximum value of the optical flow in the process as the division time point of the embryo;
s6, an output module outputs a division characteristic curve of the embryo in the embryo development process in the D1-D3 period, the division state is drawn according to the time sequence, the image without the embryo ROI area is output, the image with the embryo ROI area is analyzed for the change of an optical flow value, whether the embryo is divided in the development process is output, and if the embryo is divided, the output module outputs the second division characteristic curvejThe secondary split and the corresponding split time point,j=1,2,3, and the output result is shown in fig. 6.
Before the intelligent embryo division cycle prediction system is used for prediction, a network model needs to be trained, verified and tested. The method comprises the steps of collecting images collected by a time difference incubator, and inviting an embryologist to label an embryo area and an empty dish image; and inputting the labeled image into the built YOLO3 network model, training parameters in the model, storing the optimal parameter model, and obtaining the network model capable of being predicted through verification and test.
The images shot by the time difference incubator not only comprise embryo areas, but also have background areas interfered by impurities, light sources and the like, and the change detection of the embryos is only related to the embryo areas. In order to more accurately detect the change of the embryo in the development process, the invention combines two different algorithms of YOLO3 and template matching. Detecting a first embryo region in the time difference image through YOLO3, in order to guarantee the influence of the image size on the result in the subsequent calculation, detecting the embryo region in a second image by using the region as a template through a template matching method, matching the embryo region in a third image by using the second embryo region, and so on, and extracting the embryo region in the whole development process.
Optical flow is the movement of an image object in two consecutive frames of images due to movement of the target object or camera. In order to quantify the change of the embryo in the image, the Lucas-Kanade optical flow endows a velocity vector to a pixel point in the ROI area of the embryo, a motion vector field is further formed in the image, and the embryo can be dynamically analyzed according to the size and the direction of the velocity vector of the pixel point. If the embryo changes little in the embryo development process, the change of the optical flow vector in the ROI area of the embryo is also little, on the contrary, when the change exists in the ROI area of the embryo, the changed pixel point moves relative to the background, the formed speed inevitably changes greatly, and therefore the change in the embryo development process can be monitored.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Details not described in this specification are within the skill of the art that are well known to those skilled in the art.
Claims (10)
1. An embryo division cycle intelligent prediction system based on embryo dynamics parameters is characterized in that: the system comprises:
an image preprocessing module: the image preprocessing module is used for preprocessing the image acquired by the time difference incubator;
and (3) network model: for storing a YOLO3 network model for outputting images with and without embryo ROI regions;
image database: the time difference incubator is used for storing images acquired by the marked time difference incubator, and comprises a training set, a verification set and a test set;
a change detection analysis module: the method comprises the steps of analyzing an image with an embryo ROI area by using a YOLO3 network model, positioning the embryo ROI area in a subsequent image through template matching, performing optical flow analysis on the image with the embryo ROI area, and outputting splitting characteristics and corresponding time points in the embryo development process;
an output module: and the system is used for outputting a prediction result, wherein no embryo is output for the image without the embryo ROI area, the change of the optical flow value is analyzed for the image with the embryo ROI area, whether the embryo is split during the development process is output, and if the embryo is split, the splitting characteristic and the corresponding time point during the development process are output.
2. The intelligent embryo division cycle prediction system based on embryo dynamics parameters according to claim 1, characterized in that: the change detection analysis module comprises an embryo positioning unit, an ROI extraction unit, an optical flow value calculation unit and a splitting time point analysis unit;
the embryo positioning unit: inputting images in the embryo development sequence into a trained YOLO3 model according to a time sequence, outputting whether embryos exist in the images or not, if the embryos exist, outputting the images to an ROI extraction unit and simultaneously outputting the positions of the ROI areas of the embryos in the images, otherwise, outputting the images without the ROI areas of the embryos;
the ROI extraction unit: intercepting an ROI (region of interest) of an embryo in an image according to the position of the embryo obtained by the embryo positioning unit, calculating an edge image of the ROI of the embryo and an edge image of the embryo at the next moment by adopting a Canny edge operator, calculating the position corresponding to the embryo at the next moment by taking the edge image of the ROI of the embryo as a template, further obtaining the ROI of the embryo at the next moment, outputting the ROI of the embryo to an optical flow value calculating unit by taking the ROI of the embryo as a new template, repeating the steps, extracting the ROI of the embryo in all the images, and outputting the ROI to the optical flow value calculating unit;
the optical flow value calculating unit: arranging the extracted embryo ROI areas in sequence according to embryo development time, and calculating the optical flow value of two adjacent frames of images by adopting a Lucas-Kanade method;
the splitting time point analyzing unit: and drawing an optical flow curve according to the embryo development time, when the optical flow value is changed to be larger than a preset value, the embryo is in a split state, recording a corresponding frame image as a split characteristic frame, and recording a corresponding split time point.
3. The intelligent embryo division cycle prediction system based on embryo dynamics parameters according to claim 2, characterized in that: the splitting time point analysis unit arranges the image light stream values in sequence according to the embryo development time and records the values asv 1 ,v 2 ,…, v l-1 ，lRepresenting the number of the collected images, and taking m as the size of a sliding window to calculatev 1 ,v 2 ,…,v m Is recorded as the mean optical flow value of the first windowA 1 Calculatingv 2 ,v 3 ,…,v m+1 Is taken as the average optical flow value of the second windowA 2 Repeat until calculatingv l m- , v l m-+1 ,…, v l-1 Is recorded as the average optical flow value of the last windowA l m- ；
The average light flow values exceeding the set value are taken and arranged according to the time sequence, the sliding window corresponding to the first average light flow value contains the embryo splitting time, and the time corresponding to the last light flow value contained in the window is taken as the first timejThe point in time of the secondary splitting,j=1,2,3。
4. the embryo division cycle intelligent prediction system based on embryo dynamics parameters as claimed in claim 2, characterized in that: the formula of the optical flow value calculation unit for calculating the optical flow value is as follows:
whereinnThe number of all pixel points in the ROI area of the embryo,x i ,y i , i=1,2,…,nrespectively, in the ROI region of the embryo of the previous frameiThe abscissa and the ordinate of the point are,I xi , I yi , i=1,2,…,nrespectively represent the ROI area of the embryo in the previous frame at the point (x i ,y i ) In the x-and y-directions of the gray scale gradient, ΔI i , i=1,2,…,nShowing that the ROI area of two adjacent frames is in the secondiThe difference in the gray level of the dots,v x , v y respectively representing the light flow values of ROI areas of two adjacent frames of images in the x direction and the y direction, and V represents the light flow values of ROI areas of two adjacent frames of embryos.
5. The embryo division cycle intelligent prediction system based on embryo dynamics parameters as claimed in claim 2, characterized in that: and after the embryo positioning unit outputs the first image with the embryo, stopping detection, directly outputting all subsequent images to the ROI extraction unit, and extracting ROI areas of the embryos in all the images by the ROI extraction unit.
6. The intelligent embryo division cycle prediction system based on embryo dynamics parameters according to claim 1, characterized in that: the image preprocessing module adopts a Gaussian filter operator to eliminate noise in an embryo ROI (region of interest), calculates edge enhancement images in the x direction and the y direction respectively through a Sobel operator, and completes edge enhancement in the images in a weighted average mode.
7. A method of the embryo division cycle intelligent prediction system based on the embryo dynamics parameters, which is characterized in that: the method comprises the following steps:
s1, collecting images of a time difference incubator, and inputting the images into an image preprocessing module for image preprocessing;
s2, inputting the preprocessed image into the change detection and analysis module, wherein an embryo positioning unit in the change detection and analysis module divides the image into an image with an embryo ROI (region of interest) and an image without the embryo ROI, the former goes to the step S3, and the latter goes to the step S6;
s3, extracting ROI areas of embryos in all images by the change detection analysis module;
s4, the change detection and analysis module arranges the extracted embryo ROI areas in sequence according to embryo development time and calculates the optical flow value of two adjacent frames of images by adopting a Lucas-Kanade method;
s5, the change detection and analysis module draws an optical flow curve according to the embryo development time, when the optical flow value is changed to be larger than a preset value, the embryo is in a splitting state, a corresponding frame image is recorded as a splitting characteristic frame, and a corresponding splitting time point is recorded;
s6, the output module outputs the splitting state of each frame of image, outputs the non-embryo state of the image without the embryo ROI, analyzes the change of the optical flow value of the image with the embryo ROI, outputs whether the image is split in the embryo development process, and outputs the second frame of image if the image is splitjThe sub-divisions and the corresponding division time points,j=1,2,3。
8. the method of intelligent embryo division cycle prediction system based on embryo dynamics parameters as claimed in claim 7, wherein: the images collected in the time difference incubator in the step S1 are all images collected in the D1-D3 period for the same embryo.
9. The method of intelligent embryo division cycle prediction system based on embryo dynamics parameters as claimed in claim 7, wherein: the method for extracting the embryo ROI area in the step S3 comprises the steps of intercepting the ROI area of the embryo in the first frame image, calculating an edge image of the embryo ROI area and an edge image of the embryo at the next moment by adopting a Canny edge operator, calculating the position corresponding to the embryo at the next moment by taking the edge image of the ROI area as a template, further obtaining the ROI area of the embryo at the next moment, performing the next round of calculation by taking the embryo ROI area as a new template, and repeating the steps to extract the ROI areas of the embryos in all the images.
10. A computer-readable storage medium, storing a computer program, characterized in that the computer program, when being executed by a processor, carries out the method of any one of claims 7 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211654884.3A CN115641364B (en) | 2022-12-22 | 2022-12-22 | Embryo division period intelligent prediction system and method based on embryo dynamics parameters |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211654884.3A CN115641364B (en) | 2022-12-22 | 2022-12-22 | Embryo division period intelligent prediction system and method based on embryo dynamics parameters |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115641364A true CN115641364A (en) | 2023-01-24 |
| CN115641364B CN115641364B (en) | 2023-03-21 |
Family
ID=84947855
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211654884.3A Active CN115641364B (en) | 2022-12-22 | 2022-12-22 | Embryo division period intelligent prediction system and method based on embryo dynamics parameters |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115641364B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116051560A (en) * | 2023-03-31 | 2023-05-02 | 武汉互创联合科技有限公司 | Embryo dynamics intelligent prediction system based on embryo multidimensional information fusion |
| CN116739949A (en) * | 2023-08-15 | 2023-09-12 | 武汉互创联合科技有限公司 | Blastomere edge enhancement processing method of embryo image |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012116185A1 (en) * | 2011-02-23 | 2012-08-30 | The Board Of Trustees Of The Leland Stanford Junior University | Methods of detecting aneuploidy in human embryos |
| CN110363218A (en) * | 2019-06-06 | 2019-10-22 | 张孝东 | A kind of embryo's noninvasively estimating method and device |
| CN111539308A (en) * | 2020-04-20 | 2020-08-14 | 浙江大学 | Embryo quality comprehensive evaluation device based on deep learning |
| CN111681209A (en) * | 2020-05-15 | 2020-09-18 | 武汉互创联合科技有限公司 | Intelligent detection method and system for cleavage state of blastomere |
| CN114067164A (en) * | 2021-11-30 | 2022-02-18 | 北京航空航天大学合肥创新研究院(北京航空航天大学合肥研究生院) | Method and device for detecting embryo development stage, electronic equipment and storage medium |
-
2022
- 2022-12-22 CN CN202211654884.3A patent/CN115641364B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012116185A1 (en) * | 2011-02-23 | 2012-08-30 | The Board Of Trustees Of The Leland Stanford Junior University | Methods of detecting aneuploidy in human embryos |
| CN110363218A (en) * | 2019-06-06 | 2019-10-22 | 张孝东 | A kind of embryo's noninvasively estimating method and device |
| CN111539308A (en) * | 2020-04-20 | 2020-08-14 | 浙江大学 | Embryo quality comprehensive evaluation device based on deep learning |
| CN111681209A (en) * | 2020-05-15 | 2020-09-18 | 武汉互创联合科技有限公司 | Intelligent detection method and system for cleavage state of blastomere |
| CN114067164A (en) * | 2021-11-30 | 2022-02-18 | 北京航空航天大学合肥创新研究院(北京航空航天大学合肥研究生院) | Method and device for detecting embryo development stage, electronic equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 赵宝连 等: "《肺结节生长预测的研究现状及展望》" * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116051560A (en) * | 2023-03-31 | 2023-05-02 | 武汉互创联合科技有限公司 | Embryo dynamics intelligent prediction system based on embryo multidimensional information fusion |
| CN116739949A (en) * | 2023-08-15 | 2023-09-12 | 武汉互创联合科技有限公司 | Blastomere edge enhancement processing method of embryo image |
| CN116739949B (en) * | 2023-08-15 | 2023-11-03 | 武汉互创联合科技有限公司 | Blastomere edge enhancement processing method of embryo image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115641364B (en) | 2023-03-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115641364B (en) | Embryo division period intelligent prediction system and method based on embryo dynamics parameters | |
| CN113537106B (en) | Fish ingestion behavior identification method based on YOLOv5 | |
| CN111814741B (en) | Method for detecting embryo-sheltered pronucleus and blastomere based on attention mechanism | |
| CN111178197A (en) | Mass R-CNN and Soft-NMS fusion based group-fed adherent pig example segmentation method | |
| CN112288706A (en) | Automatic chromosome karyotype analysis and abnormality detection method | |
| JP6000699B2 (en) | Cell division process tracking device and cell division process tracking program | |
| CN110517311A (en) | Pest and disease monitoring method based on leaf spot lesion area | |
| CN109271848B (en) | Face detection method, face detection device and storage medium | |
| CN116051560B (en) | Embryo dynamics intelligent prediction system based on embryo multidimensional information fusion | |
| CN115641335B (en) | Embryo abnormity multi-cascade intelligent comprehensive analysis system based on time difference incubator | |
| CN114092699B (en) | Method and system for segmenting group pig images based on transfer learning | |
| US12051191B2 (en) | Systems and methods for acquiring and inspecting lens images of ophthalmic lenses | |
| CN116740426A (en) | Classification prediction system for functional magnetic resonance images | |
| Isa et al. | CNN transfer learning of shrimp detection for underwater vision system | |
| EP3676750A1 (en) | Detection of biological cells and tracing of cell lineage | |
| Zhang et al. | TPMv2: An end-to-end tomato pose method based on 3D key points detection | |
| JP2019088243A (en) | Image analysis device | |
| Ding et al. | Classification of chromosome karyotype based on faster-rcnn with the segmatation and enhancement preprocessing model | |
| CN108021894B (en) | Wheat disease detection method based on deep strategy learning | |
| CN111881748B (en) | Lane line visual identification method and system based on VBAI platform modeling | |
| CN116884036A (en) | Live pig posture detection method, device, equipment and medium based on YOLOv5DA | |
| CN116109840A (en) | Cherry spore identification method based on machine vision | |
| CN114066920B (en) | Harvester visual navigation method and system based on improved Segnet image segmentation | |
| CN113627255B (en) | Method, device and equipment for quantitatively analyzing mouse behaviors and readable storage medium | |
| CN114120318A (en) | Dark field image target point accurate extraction method based on integrated decision tree |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |