Dynamic detection method and system of mitochondrial fusion and fission based on image recognition

Abstract

The present invention discloses a dynamic detection method and system of mitochondrial fusion and fission based on image recognition, which belongs to the field of image recognition technology. The method comprises: marking the acquisition time for each cell image, generating a time image sequence based on the acquisition time; assigning a first serial number to each cell image, and performing contour recognition on each cell image based on a contour detection algorithm to locate the first contour of each mitochondria; assigning a second serial number to each first contour in the cell image, and comparing and integrating the first contours in each cell image; intercepting the first contour containing the same second serial number, and integrating to generate a morphological change sequence of each mitochondria; analyzing the morphological change sequence to obtain various mitochondrial forms and corresponding appearance times; analyzing the morphology of mitochondria to locate abnormal mitochondria. The present invention can track the morphology of mitochondria, thereby accurately and effectively identifying the dynamic changes of mitochondria during the entire observation period.

Description

Technical Field

[0001] The present invention belongs to the technical field of image recognition, and in particular relates to a dynamic detection method and system for mitochondrial fusion and fission based on image recognition. Background Art

[0002] Mitochondria are key organelles present in most cells. Mitochondrial dynamics refers to the process in which mitochondria continuously undergo dynamic fusion and fission under the regulation of related proteins, thereby completing the dynamic changes in their morphology and maintaining the overall stability of the mitochondrial network structure. In addition, excessive fusion of mitochondria may lead to abnormal expansion of the mitochondrial network, while excessive fission may lead to mitochondrial fragmentation, which may be related to the occurrence of diseases. Currently, in order to reduce the pressure of manual recognition and speed up the identification of mitochondria in cell super-resolution images, a variety of image processing solutions have been proposed in the prior art.

[0003] For example, the Chinese patent document with publication number CN116797791A discloses a mitochondrial segmentation and classification method based on deep learning. This method uses a super-resolution fluorescence microscope to take super-resolution fluorescence images of mitochondria, and then uses the deep learning algorithm CNN-11 and integrates the XGBoost model to establish an image segmentation model. The image segmentation model is trained to achieve accurate recognition and segmentation of mitochondrial contours in super-resolution fluorescence images. However, this method can only achieve the segmentation of mitochondrial contours, and cannot track the fission and fusion process of mitochondria, so it cannot accurately and effectively identify the dynamic changes of mitochondria in cells. Summary of the invention

[0004] To solve the above problems, the present invention provides a method and system for dynamic detection of mitochondrial fusion and fission based on image recognition, so as to realize dynamic tracking of mitochondrial fusion and fission process.

[0005] In order to achieve the above-mentioned purpose of the invention, the present invention proposes a dynamic detection method of mitochondrial fusion and fission based on image recognition, comprising:

[0006] Using a high-resolution microscope to collect cell images, marking the collection time for each of the cell images, and generating a time image sequence of the cell images based on the collection time;

[0007] Assigning a first serial number to each of the cell images in the time image sequence, and performing contour recognition on each of the cell images based on a contour detection algorithm to locate the first contour of each mitochondria in the cell image;

[0008] assigning a second serial number to each of the first contours in the cell image, and comparing and integrating the first contours in each of the cell images so that the first contours corresponding to the same mitochondria have the same second serial number in different cell images;

[0009] intercepting the first contour containing the same second serial number in each of the cell images, and integrating and generating a morphological change sequence of each mitochondria according to the corresponding first serial number;

[0010] Analyzing the morphological change sequence to obtain various mitochondrial morphologies and corresponding appearance times, the mitochondrial morphologies including fission state, fusion state and stable state;

[0011] Mitochondrial morphology is analyzed to locate abnormal mitochondria.

[0012] Further, in the morphological change sequence, if no abnormal segmentation line appears inside the first contour, the state of the corresponding mitochondria is classified as the stable state;

[0013] If the abnormal segmentation line appears in the first contour, and the abnormal segmentation line divides the first contour into a plurality of sub-areas, the first contour is defined as a second contour;

[0014] Extracting the first contours whose second serial numbers are located before and after the second contour and defining them as a third contour and a fourth contour respectively, acquiring contour shapes of each of the sub-regions in the second contour, and locating contour curves corresponding to the contour shapes in the third contour and the fourth contour;

[0015] If the sub-region exists, there is no contour curve corresponding to the contour shape in the third contour, and there is a contour curve corresponding to the contour shape in the fourth contour, the state of the mitochondria corresponding to the second contour is classified as the fusion state, and the sub-region is defined as a fusion region;

[0016] If the sub-region exists, the contour curve corresponding to the contour shape exists in the third contour, and the contour curve corresponding to the contour shape does not exist in the fourth contour, the state of the mitochondria corresponding to the second contour is divided into the fission state, and the sub-region is defined as a fission region;

[0017] If all the sub-regions do not have the contour curve corresponding to the contour shape thereof in the third contour and the fourth contour, the state of the mitochondria corresponding to the second contour is positioned as the stable state.

[0018] Further, locating the contour curve corresponding to the contour shape comprises the following steps:

[0019] Define the cell image where the third contour is located as the first image, obtain the first coordinate of the centroid of the contour shape of the sub-area to be compared, map the first coordinate to the first image, generate a detection area in the first image with the first coordinate as the center, intercept the contour portion of the third contour located in the detection area, select multiple calibration points on the third contour and the contour portion, calculate the curvature of each calibration point, and generate a first curvature distribution of the contour shape and a second curvature distribution of the contour portion, calculate the first curvature distribution and the second curvature distribution to obtain distribution similarity, and if the distribution similarity is greater than a first threshold, determine that the contour curve corresponding to the contour shape exists in the third contour.

[0020] Furthermore, comparing and integrating the first profiles includes the following steps:

[0021] The cell images adjacent to the first serial number are selected and defined as the second image and the third image respectively, the first contour is selected in the second image and defined as the target contour, the second serial number and the second coordinate of the center of mass corresponding to the target contour are obtained, the third coordinate of the center of mass of the first contour of each mitochondria in the third image is located, the second coordinate is mapped in the third image, the third coordinate in the third image whose distance from the second coordinate is less than a second threshold is defined as a neighboring coordinate, the first contour with the neighboring coordinate as the center of mass is defined as a neighboring contour, the shape similarity between the target contour and each of the neighboring contours is compared, the second serial number of the neighboring contour with the largest shape similarity is corrected to be the same as that of the target contour, and this step is repeated to traverse all the first contours and the cell images.

[0022] Further, calculating the shape similarity between the target contour and the adjacent contour comprises the following steps:

[0023] The target contour is overlaid on the adjacent contour with the centroid as the reference point, and the shape similarity P between the target contour and the adjacent contour is calculated by a comparison formula, and the comparison formula is: , where M0 is the number of overlapping pixels after the target contour covers the adjacent contour, M1 and M2 are the number of pixels included in the target contour and the adjacent contour respectively, Max (M1, M2) returns the larger value of M1 or M2, and L1 and L2 are the corresponding contour line lengths of the target contour and the adjacent contour respectively.

[0024] Further, the morphological analysis of mitochondria includes the following steps:

[0025] The number of fusion areas in the first contour is defined as the number of single fusions, and the number of fission areas is defined as the number of single fissions. If the number of single fusions exceeds the third threshold, it is determined that the mitochondria have excessive fusion; if the number of single fissions exceeds the fourth threshold, it is determined that the mitochondria have excessive fission.

[0026] Furthermore, after analyzing the morphological change sequence, the first number of times that all mitochondria appear in the fission state and the second number of times that all mitochondria appear in the fusion state are counted based on the analysis results.

[0027] Further, after acquiring the first contour, the morphological features of the mitochondria are calculated based on the first contour, the morphological features include the area, length and shape of the mitochondria, and the shape includes linear and granular shapes.

[0028] Furthermore, the contour detection algorithm is a deep learning algorithm based on CNN.

[0029] The present invention also provides a dynamic detection system for mitochondrial fusion and fission based on image recognition, which is used to implement the above-mentioned dynamic detection method for mitochondrial fusion and fission based on image recognition, and the system comprises:

[0030] An acquisition module, which uses a high-resolution microscope to acquire cell images, marks an acquisition time for each cell image, and generates a time image sequence of the cell images based on the acquisition time;

[0031] a segmentation module, configured to assign a first serial number to each of the cell images in the time image sequence, perform contour recognition on each of the cell images based on a contour detection algorithm to locate a first contour of each mitochondria in the cell image, assign a second serial number to each of the first contours in the cell image, and compare and integrate the first contours in each of the cell images so that the first contour corresponding to the same mitochondria has the same second serial number in different cell images;

[0032] A capture module, used for capturing the first contour containing the same second serial number in each of the cell images, and integrating and generating a morphological change sequence of each mitochondria according to the corresponding first serial number;

[0033] The analysis module is used to analyze the morphological change sequence to obtain various mitochondrial morphologies and corresponding appearance times. The mitochondrial morphologies include fission state, fusion state and stable state. The mitochondrial morphologies are analyzed to locate abnormal mitochondria.

[0034] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0035] The present invention integrates the collected high-resolution cell images into a time image sequence according to the acquisition time, and then combines the deep learning technology to perform contour segmentation and recognition on each cell image, and locates the first contour belonging to the mitochondria in each cell image. In order to facilitate the subsequent mitochondrial morphological analysis, after completing the contour segmentation, the mitochondrial contours in each cell image are numbered, and the contours belonging to the same mitochondria are made to have the same second sequence number in all cell images through contour matching. On this basis, the mitochondrial contours with the same second sequence number are intercepted to form a morphological change sequence, and finally the morphological change sequence is analyzed to achieve dynamic tracking of the mitochondrial fission state and fusion state. In addition, the present invention also counts the number of mitochondrial divisions and fusions during the entire observation period, as well as the specific circumstances of each division and fusion, so as to automatically locate the mitochondria with abnormal division and fusion, thereby further assisting researchers in statistical analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 The present invention is a flowchart of the steps of the dynamic detection method of mitochondrial fusion and fission based on image recognition;

[0037] Figure 2 It is a schematic diagram of contour segmentation of the present invention;

[0038] Figure 3 This is a schematic diagram of the mitochondrial state detection principle of the present invention;

[0039] Figure 4 A schematic diagram of generating the detection area of ​​the present invention;

[0040] Figure 5 It is a structural diagram of the dynamic detection system of mitochondrial fusion and fission based on image recognition of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0042] It is understood that the terms "first", "second", etc. used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, without departing from the scope of this application, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script.

[0043] like Figure 1 As shown, a dynamic detection method for mitochondrial fusion and fission based on image recognition comprises:

[0044] S1: Use a high-resolution microscope to collect cell images, mark the acquisition time for each cell image, and generate a time image sequence of the cell images based on the acquisition time.

[0045] Specifically, a high-resolution cell image can be collected every 5 seconds, or it can be set to other values ​​based on experience; each time a cell image is collected, the collection time is marked for it, and then the cell images can be integrated into a time image sequence according to the collection time. In the time image sequence, the cell images are sorted in the order of collection time, so as to facilitate subsequent analysis.

[0046] Step S2: assigning a first serial number to each cell image in the time image sequence, and performing contour recognition on each cell image based on a contour detection algorithm to locate the first contour of each mitochondria in the cell image.

[0047] Among them, the contour detection algorithm is a deep learning algorithm based on CNN.

[0048] Since the cell images in the time image sequence have been sorted according to the acquisition time, the first serial number can be assigned according to the order of precedence, and the first serial number is in the form of 1, 2, 3, etc. Then, the contour of each cell image is segmented by a contour detection algorithm, so as to separate the mitochondria, cell nucleus, etc., and the contour detection algorithm used can be a HED detection algorithm based on CNN, or a fusion algorithm combined with deep learning proposed in the background technology; Figure 2 , is the segmentation effect diagram after using the contour detection algorithm, through Figure 2 The first outline of mitochondria in the cell image can be accurately identified.

[0049] S3: assigning a second serial number to each first contour in the cell image, and comparing and integrating the first contours in each cell image, so that the first contours corresponding to the same mitochondria have the same second serial number in different cell images.

[0050] Specifically, first mark the second serial number for the first contour in each cell image, and the second serial number is in the form of A1, A2, A3, etc. Then, by comparing the cell images with adjacent second serial numbers, the same mitochondria have the same second serial number in different cell images. For example, for the same mitochondria, the second serial number in cell image 1 is A1, and the second serial number in cell image 2 is A2. After comparison and integration, the second serial number of the mitochondria in cell image 2 is modified to A1, which makes it easier to track the fission and fusion status of a single mitochondria.

[0051] S4: intercepting the first contours containing the same second serial number in each cell image, and integrating and generating a morphological change sequence of each mitochondria according to the corresponding first serial number.

[0052] After completing the integration of the second sequence number, the contour image of the mitochondria with the same second sequence number is captured from the cell image, and a morphological change sequence is generated according to the acquisition time combination of the corresponding cell image. The purpose is to split the time image sequence into a shape change sequence only for each mitochondria based on the mitochondria. In this way, when performing subsequent morphological analysis on the mitochondria, the analysis pressure of the processor can be reduced and the analysis speed can be increased.

[0053] S5: Analyze the morphological change sequence to obtain various mitochondrial morphologies and their corresponding appearance times. Mitochondrial morphologies include fission state, fusion state and stable state.

[0054] S6: Analyze the mitochondrial morphology to locate abnormal mitochondria.

[0055] This embodiment locates the morphological changes of each mitochondria by analyzing the morphological change sequence, and the specific morphological analysis process will be described in detail later. Through morphological analysis, it can be known at what time period the mitochondria are in a fission state when fission occurs, or in a fusion state when fusion occurs, or in a stable state when neither fission nor fusion occurs; then based on the number of occurrences of the fission state and the stable state, the number of mitochondrial fissions and fusions during the entire observation period are further counted, as well as how many mitochondria are split into each time of fission, and how many mitochondria are involved in each fusion, ultimately helping researchers to locate the location and number of abnormal mitochondria, reducing their statistical pressure, and achieving the purpose of auxiliary research.

[0056] The present invention first integrates the collected high-resolution cell images into a time image sequence according to the acquisition time, and then combines the deep learning technology to perform contour segmentation and recognition on each cell image, and locates the first contour belonging to the mitochondria in each cell image. In order to facilitate the subsequent morphological analysis of mitochondria, after completing the contour segmentation, the mitochondrial contours in each cell image are numbered, and the contours belonging to the same mitochondria are matched to have the same second sequence number in all cell images. On this basis, the mitochondrial contours with the same second sequence number are intercepted to form a morphological change sequence, and finally the morphological change sequence is analyzed to achieve dynamic tracking of the mitochondrial fission state and fusion state. In addition, the present invention also counts the number of mitochondrial divisions and fusions during the entire observation period, as well as the specific circumstances of each division and fusion, so as to automatically locate the mitochondria with abnormal division and fusion, thereby further assisting researchers in statistical analysis.

[0057] It is particularly noteworthy that the present invention can track the morphology of mitochondria, thereby accurately and effectively identifying the dynamic changes of mitochondria during the entire observation period.

[0058] This embodiment analyzes the morphological change sequence including the following steps:

[0059] In the morphological change sequence, if no abnormal segmentation line appears inside the first contour, the state of the corresponding mitochondria is classified as a stable state.

[0060] If an abnormal segmentation line appears in the first contour, and the abnormal segmentation line divides the first contour into a plurality of sub-regions, the first contour is defined as the second contour.

[0061] According to the following Figure 3 The morphological analysis process of the mitochondria of the present invention is introduced. First, the image of each first contour of the mitochondria is located. If there is no abnormal segmentation line, such as Figure 3 As shown in stage a, mitochondria 1 and 2 are included, and the corresponding contours are the first contour A1 and the first contour A2. The image of the first contour A2 is added additionally to clearly introduce this step. In fact, it has been eliminated during the previous contour interception and will not exist in the morphological change sequence of mitochondria 1. If it is in the state shown by the first contour A1, that is, there is no abnormal dividing line, then the mitochondria in the image are considered to be in a stable state. If there is an abnormal dividing line, as shown in stage b and stage c, an abnormal dividing line appears in the first contour A1 to divide it into sub-region 1 and sub-region 2, and an abnormal dividing line appears in stage c to divide the first contour A1 into sub-region 1 and sub-region 3. In order to avoid confusion in the description, the first contour with an abnormal dividing line is redefined as the second contour. For the sake of clear description, this embodiment still uses the first contour A1 for subsequent description.

[0062] The first contours with second serial numbers before and after the second contour are extracted and defined as the third contour and the fourth contour respectively, the contour shapes of each sub-area in the second contour are obtained, and the contour curves corresponding to the contour shapes are located in the third contour and the fourth contour.

[0063] Since the fusion or fission process of a single mitochondria usually lasts for a long time, in order to extract mitochondrial images in different forms, in this embodiment, the Mth first contour located before and after the second contour is extracted as the third contour and the fourth contour respectively, such as the first serial number of the cell image corresponding to stage a is 1, and the first serial number of the cell image corresponding to stage b is 4. The contour shapes of sub-regions 1 and 2 in the first contour A1 in stage b are obtained, and then contour curves similar to the contour shapes of sub-regions 1 and 2 in stage b are located in stages a and c respectively.

[0064] If there is a sub-region, there is no contour curve corresponding to its contour shape in the third contour, and there is a contour curve corresponding to its contour shape in the fourth contour, the state of the mitochondria corresponding to the second contour is classified as a fusion state, and the sub-region is defined as a fusion region;

[0065] If a sub-region exists, there is a contour curve corresponding to its contour shape in the third contour, and there is no contour curve corresponding to its contour shape in the fourth contour, the state of the mitochondria corresponding to the second contour is divided into a fission state, and the sub-region is defined as a fission region;

[0066] If all sub-regions have no contour curves corresponding to their contour shapes in the third contour and the fourth contour, the state of the mitochondria corresponding to the second contour is located as a stable state.

[0067] For sub-region 1 in stage b, its contour shape is relatively similar to the upper half of the first contour A1 in stage a and stage c, so sub-region 1 is not applicable to the above judgment conditions. For sub-region 2 in stage b, there is no contour curve similar to it in the first contour A1 in stage a and stage c, so the mitochondrial state corresponding to the first contour A in stage b is set as a stable state here; its actual meaning is that the mitochondria 2 floating in stage a is in an overlapping state in stage b and is not actually fused, and is separated from mitochondria 1 in stage c, so the state of the mitochondria in stage b is a stable state.

[0068] In addition, it is assumed that the following situation exists: for sub-region 2 in stage b, if the contour curve of the lower half of the first contour A1 in stage a is not similar to the contour shape of sub-region 2, and the contour curve of the lower half of the first contour A1 in stage c is similar to the contour shape of sub-region 2, then it is considered that mitochondria 1 in stage b is in a fusion state. That is, mitochondria 2 floating in stage a fuses with mitochondria 1 and stays therein.

[0069] Reference Figure 3 In stages c and d, for the first contour A1 in stage c, the contour shape of its sub-region 3 is similar to the contour curve of the upper half of the first contour A1 in stage b, and no similar contour curve is found in stage d, then the state of mitochondria 1 in stage c is classified as a fission state. That is, sub-region 3 in stage c separates and leaves mitochondria 1.

[0070] Locating the contour curve corresponding to the contour shape includes the following steps:

[0071] Define the cell image where the third contour is located as the first image, obtain the first coordinate of the centroid of the contour shape of the sub-region to be compared, map the first coordinate to the first image, generate a detection area in the first image with the first coordinate as the center, intercept the contour part of the third contour located in the detection area, select multiple calibration points on the third contour and the contour part, calculate the curvature of each calibration point, and generate a first curvature distribution of the contour shape and a second curvature distribution of the contour part. Calculate the first curvature distribution and the second curvature distribution to obtain distribution similarity. If the distribution similarity is greater than a first threshold, determine that there is a contour curve corresponding to the contour shape in the third contour.

[0072] Here, the third contour is used as an example for explanation. The comparison process of the contour shape and the fourth contour is the same as this process. Under the premise of determining the contour shape of the sub-region, the centroid of the contour shape can be obtained, and the same coordinate system is established in each cell image. The first coordinate of the centroid in the original cell image is obtained according to the coordinate system, and then the first coordinate is mapped to the first image where the third contour is located, and then the detection area is generated with the first coordinate as the center. Figure 4 As shown, the centroid Y of the contour shape is projected from the original cell image to the third image, and a rectangular detection area V is generated with the centroid Y as the center. After that, similarly, multiple calibration points are selected on the contour shape of the sub-area and the curvature at the calibration points is calculated, and a first curvature distribution is generated in a clockwise order of the calibration points. Multiple calibration points are selected on the contour part of the third contour within the detection area and the curvature is calculated, and a second curvature distribution is generated in a clockwise order of the calibration points.

[0073] The first curvature distribution and the second curvature distribution are specifically a set of data series. As a way to calculate the distribution similarity, the variance, standard deviation, mean value, etc. can be used as the distribution similarity, or the following method can be used. Assuming that the first curvature distribution has N curvatures and the second curvature distribution has 2N curvatures, align the positions of the 1st to Nth digits of the first curvature distribution with the 1st to Nth digits of the second curvature distribution, and then calculate the sum of the Euclidean distances of the digits in the same position in the two data series as the first calculation result. After that, align the positions of the 1st to Nth digits of the first curvature distribution with the 2nd to N+1th digits of the second curvature distribution, and then recalculate the sum of the Euclidean distances, and repeat the process until the Nth digit of the first curvature distribution is aligned with the 2Nth digit of the first curvature distribution. Finally, obtain the Euclidean distance X with the smallest value, and use 1 / e X as distribution similarity.

[0074] Since the movement speed of mitochondria slows down when they divide, their positions will not change significantly. When two mitochondria fuse, the positions of the two mitochondria involved in the fusion will not change significantly. This makes the position of the center of mass of the divided mitochondria or the mitochondria involved in the fusion not too far from the position of the division point and the fusion point. Therefore, through this step, the search range of the contour curve in the third contour can be narrowed, thereby further speeding up the processing speed.

[0075] In this embodiment, the comparison and integration of the first profile includes the following steps:

[0076] Select cell images adjacent to the first serial number and define them as the second image and the third image respectively, select the first contour in the second image and define it as the target contour, obtain the second serial number and the second coordinate of the center of mass corresponding to the target contour, locate the third coordinate of the center of mass of each mitochondrial first contour in the third image, map the second coordinate in the third image, define the third coordinate in the third image whose distance to the second coordinate is less than the second threshold as the adjacent coordinate, define the first contour with the adjacent coordinate as the center of mass as the adjacent contour, compare the shape similarity of the target contour and each adjacent contour, correct the second serial number of the adjacent contour with the largest shape similarity to be the same as that of the target contour, and repeat this step to traverse all first contours and cell images.

[0077] Here, cell image 1 and cell image 2 are used as examples for explanation. Cell image 1 is defined as the second image, and cell image 2 is defined as the third image. A first contour is selected from the second image as the target contour, and the second serial number of the target contour is A3. Then the second coordinate of the centroid of the target contour is obtained, and the second coordinate is similarly mapped from the second image to the third image. In the third image, the third coordinate of the centroid of each first contour contained is obtained, and the coordinates that are closer to the second coordinate after projection are selected as adjacent coordinates. Finally, the first contour with these adjacent coordinates as the centroid is obtained, which is defined as the adjacent contour. The target contour is compared with each adjacent contour respectively to find the adjacent contour that is most similar to the shape of the target contour, and the second serial number of the adjacent contour is corrected to A3. When the acquisition time interval is short, the position and shape of the same mitochondria in adjacent cell images will not change significantly. Therefore, through this step, there is no need to compare the target contour with each first contour in the third image, but only need to compare with a certain range of first contours, thereby narrowing the comparison range, and finally realizing the rapid integration of mitochondria in different cell images.

[0078] In this embodiment, calculating the shape similarity between the target contour and the adjacent contours includes the following steps:

[0079] The target contour is covered on the adjacent contour with the centroid as the reference point, and the shape similarity P between the target contour and the adjacent contour is calculated by the comparison formula. The comparison formula is: , where M0 is the number of overlapping pixels after the target contour covers the adjacent contour, M1 and M2 are the number of pixels included in the target contour and the adjacent contour respectively, Max (M1, M2) returns the larger value of M1 or M2, and L1 and L2 are the corresponding contour line lengths of the target contour and the adjacent contour respectively.

[0080] When the target contour is overlaid on the adjacent contour, the reference points of the two contours overlap, and then the number of overlapping pixels in the two contour areas after the coverage and the length of the contour lines of the two contours are counted. From the comparison formula, it can be seen that when the target contour is overlaid on the adjacent contour, the more the number of overlapping pixels, the closer the contour line lengths of the two contours, and the greater the calculated shape similarity.

[0081] In this embodiment, the morphological analysis of mitochondria includes the following steps:

[0082] The number of fusion regions in the first contour is defined as the number of single fusions, and the number of fission regions is defined as the number of single fissions. If the number of single fusions exceeds the third threshold, it is determined that the mitochondria are over-fused; if the number of single fissions exceeds the fourth threshold, it is determined that the mitochondria are over-fissioning.

[0083] For example, if there are four fission regions in the first contour, the number of single fissions is 4, which means that this fission causes the mitochondria to split from one to four. If the third threshold is 3, it is considered that the mitochondria are over-fissioned, and the process of judging the over-fusion of mitochondria is similar. This step can further assist researchers in locating mitochondria that are over-fissioned and over-fused.

[0084] After analyzing the morphological change sequence, this embodiment counts the first number of times all mitochondria appear in the fission state and the second number of times all mitochondria appear in the fusion state based on the analysis results, so as to facilitate researchers to understand whether mitochondria frequently fission or frequently fuse.

[0085] After acquiring the first contour, this embodiment calculates the morphological features of the mitochondria based on the first contour. The morphological features include the area, length and shape of the mitochondria. The shape includes linear and granular.

[0086] Specifically, the area of ​​mitochondria is calculated by the number of pixels included in the first contour, the length of mitochondria is determined by obtaining the distance between the two farthest pixels on the first contour, and the shape of mitochondria is determined based on the distribution of distances by obtaining the distance from the center of mass of the first contour to multiple points on the first contour. If the distribution of distances is relatively uniform, the mitochondria are determined to be linear, otherwise they are granular.

[0087] like Figure 5As shown, the present invention also provides a dynamic detection system for mitochondrial fusion and fission based on image recognition, which is used to implement the above-mentioned dynamic detection method for mitochondrial fusion and fission based on image recognition, and the system includes:

[0088] The acquisition module uses a high-resolution microscope to acquire cell images, marks the acquisition time for each cell image, and generates a time image sequence of the cell images based on the acquisition time.

[0089] The segmentation module is used to assign a first serial number to each cell image in the time image sequence, perform contour recognition on each cell image based on a contour detection algorithm to locate the first contour of each mitochondria in the cell image, assign a second serial number to each first contour in the cell image, and compare and integrate the first contours in each cell image so that the first contours corresponding to the same mitochondria have the same second serial number in different cell images.

[0090] The interception module is used to intercept the first contour containing the same second serial number in each cell image, and integrate and generate the morphological change sequence of each mitochondria according to the corresponding first serial number.

[0091] The analysis module is used to analyze the morphological change sequence to obtain various mitochondrial forms and the corresponding appearance time. The mitochondrial morphology includes the fission state, the fusion state and the stable state. The mitochondrial morphology is analyzed to locate the abnormal mitochondria.

[0092] It should be understood that the various technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0093] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A dynamic detection method for mitochondrial fusion and fission based on image recognition, characterized in that: include: Using a high-resolution microscope to collect cell images, marking the collection time for each cell image, and generating a time image sequence of the cell images based on the collection time; Assigning a first serial number to each of the cell images in the time image sequence, and performing contour recognition on each of the cell images based on a contour detection algorithm to locate the first contour of each mitochondria in the cell image; assigning a second serial number to each of the first contours in the cell image, and comparing and integrating the first contours in each of the cell images so that the first contours corresponding to the same mitochondria have the same second serial number in different cell images; intercepting the first contour containing the same second serial number in each of the cell images, and integrating and generating a morphological change sequence of each mitochondria according to the corresponding first serial number; Analyzing the morphological change sequence to obtain various mitochondrial morphologies and corresponding appearance times, the mitochondrial morphologies including fission state, fusion state and stable state; Mitochondrial morphology is analyzed to locate abnormal mitochondria.

2. The method according to claim 1, characterized in that Analyzing the morphological change sequence comprises the following steps: In the morphological change sequence, if no abnormal segmentation line appears inside the first contour, the state of the corresponding mitochondria is classified as the stable state; If the abnormal segmentation line appears in the first contour, and the abnormal segmentation line divides the first contour into a plurality of sub-areas, the first contour is defined as a second contour; Extracting the first contours whose second serial numbers are located before and after the second contour and defining them as a third contour and a fourth contour respectively, acquiring contour shapes of each of the sub-regions in the second contour, and locating contour curves corresponding to the contour shapes in the third contour and the fourth contour; If the sub-region exists, there is no contour curve corresponding to the contour shape in the third contour, and there is a contour curve corresponding to the contour shape in the fourth contour, the state of the mitochondria corresponding to the second contour is classified as the fusion state, and the sub-region is defined as a fusion region; If the sub-region exists, the contour curve corresponding to the contour shape exists in the third contour, and the contour curve corresponding to the contour shape does not exist in the fourth contour, the state of the mitochondria corresponding to the second contour is divided into the fission state, and the sub-region is defined as a fission region; If all the sub-regions do not have the contour curve corresponding to the contour shape thereof in the third contour and the fourth contour, the state of the mitochondria corresponding to the second contour is positioned as the stable state.

3. The method according to claim 2, characterized in that Locating the contour curve corresponding to the contour shape comprises the following steps: Define the cell image where the third contour is located as the first image, obtain the first coordinate of the centroid of the contour shape of the sub-area to be compared, map the first coordinate to the first image, generate a detection area in the first image with the first coordinate as the center, intercept the contour portion of the third contour located in the detection area, select multiple calibration points on the third contour and the contour portion, calculate the curvature of each calibration point, and generate a first curvature distribution of the contour shape and a second curvature distribution of the contour portion, calculate the first curvature distribution and the second curvature distribution to obtain distribution similarity, and if the distribution similarity is greater than a first threshold, determine that the contour curve corresponding to the contour shape exists in the third contour.

4. The method according to claim 1, characterized in that: Comparing and integrating the first profiles comprises the following steps: The cell images adjacent to the first serial number are selected and defined as the second image and the third image respectively, the first contour is selected in the second image and defined as the target contour, the second serial number and the second coordinate of the center of mass corresponding to the target contour are obtained, the third coordinate of the center of mass of the first contour of each mitochondria in the third image is located, the second coordinate is mapped in the third image, the third coordinate in the third image whose distance from the second coordinate is less than a second threshold is defined as a neighboring coordinate, the first contour with the neighboring coordinate as the center of mass is defined as a neighboring contour, the shape similarity between the target contour and each of the neighboring contours is compared, the second serial number of the neighboring contour with the largest shape similarity is corrected to be the same as that of the target contour, and this step is repeated to traverse all the first contours and the cell images.

5. The method according to claim 4, characterized in that Calculating the shape similarity between the target contour and the adjacent contour comprises the following steps: The target contour is overlaid on the adjacent contour with the centroid as the reference point, and the shape similarity P between the target contour and the adjacent contour is calculated by a comparison formula, and the comparison formula is: , where M0 is the number of overlapping pixels after the target contour covers the adjacent contour, M1 and M2 are the number of pixels included in the target contour and the adjacent contour respectively, Max (M1, M2) returns the larger value of M1 or M2, and L1 and L2 are the corresponding contour line lengths of the target contour and the adjacent contour respectively.

6. The method according to claim 2, characterized in that Morphological analysis of mitochondria involves the following steps: The number of fusion areas in the first contour is defined as the number of single fusions, and the number of fission areas is defined as the number of single fissions. If the number of single fusions exceeds the third threshold, it is determined that the mitochondria have excessive fusion; if the number of single fissions exceeds the fourth threshold, it is determined that the mitochondria have excessive fission.

7. The method according to claim 1, characterized in that After analyzing the morphological change sequence, the first number of times that all mitochondria appear in the fission state and the second number of times that all mitochondria appear in the fusion state are counted based on the analysis results.

8. The method according to claim 1, characterized in that After acquiring the first contour, the morphological features of the mitochondria are calculated based on the first contour, the morphological features including the area, length and shape of the mitochondria, and the shape including linear and granular.

9. The method according to claim 1, characterized in that: The contour detection algorithm is a deep learning algorithm based on CNN.

10. A dynamic detection system for mitochondrial fusion and fission based on image recognition, used to implement the method according to any one of claims 1 to 9, characterized in that: include: An acquisition module, which uses a high-resolution microscope to acquire cell images, marks an acquisition time for each cell image, and generates a time image sequence of the cell images based on the acquisition time; a segmentation module, configured to assign a first serial number to each of the cell images in the time image sequence, perform contour recognition on each of the cell images based on a contour detection algorithm to locate a first contour of each mitochondria in the cell image, assign a second serial number to each of the first contours in the cell image, and compare and integrate the first contours in each of the cell images so that the first contour corresponding to the same mitochondria has the same second serial number in different cell images; A capture module, used for capturing the first contour containing the same second serial number in each of the cell images, and integrating and generating a morphological change sequence of each mitochondria according to the corresponding first serial number; The analysis module is used to analyze the morphological change sequence to obtain various mitochondrial morphologies and corresponding appearance times. The mitochondrial morphologies include fission state, fusion state and stable state. The mitochondrial morphologies are analyzed to locate abnormal mitochondria.

Images