A method, device, and storage medium for automatically detecting the sarcomere region of cardiomyocytes

Abstract

The present invention relates to a method, device and storage medium for automatically detecting the sarcomere region of cardiomyocytes, and is applied to the technical field of cardiomyocyte detection. Specifically, it includes: fusing the original video data into an image to represent the region of cells during rhythmic contraction and relaxation; calculating the structural strength of the image based on the fused and sharpened image, and generating a high-structural-strength region mask according to the structural strength, which represents the region where sarcomeres exist; finally, iteratively optimizing the high-structural-strength region mask image by the centroid-driven polar ray method to obtain the sarcomere region map inside the cardiomyocyte; by calculating the structural features of the image after fusing multiple frames of images, the present application obtains the general sarcomere region, and through the centroid-driven polar ray method, obtains a smooth inscribed contour of the sarcomere region. Compared with the prior art, it does not need to locate cardiomyocytes, directly locates sarcomeres through the structural strength, and the obtained sarcomere ROI is more suitable for subsequent calculation of sarcomere motion parameters and is not affected by human subjective factors.

Description

Technical Field

[0001] The present invention belongs to the technical field of cardiomyocyte detection, and particularly relates to a method, device and storage medium for automatically detecting the sarcomere region of cardiomyocytes. Background Art

[0002] Cardiomyocytes are the basic functional units of the heart. Their rhythmic contraction and relaxation characteristics are closely related to heart diseases. One of the main methods for studying cardiomyocyte activity is to analyze the movement law of sarcomeres in cardiomyocytes. Sarcomeres are the basic contraction units of cardiomyocytes, and their movement laws can be used to evaluate the contraction function of cardiomyocytes, and then for the research of cardiovascular diseases and drug screening; in actual research, accurately selecting the region of interest (ROI) where the sarcomere is located is the premise for data analysis, and its accuracy directly affects the subsequent calculation results, thus determining the quality of experimental data and the reliability of research results.

[0003] Currently, the analysis of the contraction characteristics of cardiomyocytes mainly uses microscopic video data analysis technology. Its standard process usually includes the following steps: (1) isolating primary cardiomyocytes from rat or mouse heart tissue; (2) observing them through an optical microscope pixel system and continuously shooting microscopic video data of the rhythmic contraction of cardiomyocytes; (3) researchers need to select target cardiomyocytes from the video and manually delimit static ROIs (usually limited to single-frame images) in their sarcomere regions; (4) calculating sarcomere movement parameters such as its length, movement trace, power spectrum, and velocity based on the frequency domain characteristics of the sarcomere ROI region; the existing technology mainly relies on manual annotation of the sarcomere region, and there are the following problems: (1) lack of capture of dynamic characteristics: the sarcomere region shows temporal dynamic changes along with the rhythmic contraction and relaxation of cardiomyocytes, while the traditional manual annotation method only considers single-frame static images and cannot find the region corresponding to the most relaxed sarcomere during the continuous change of the sarcomere, resulting in inaccurate selection of the sarcomere region; (2) inefficiency and non-scalability: the sample processing is time-consuming and cannot form high-throughput and automated analysis, making it difficult to meet the requirements for detecting the sarcomere region of large-scale cardiomyocyte data; (3) subjective bias: manual annotation depends on manual operation and is affected by human subjective factors, with inter-individual and intra-individual annotation biases, poor consistency and low repeatability. Summary of the Invention

[0004] In view of this, an object of the present invention is to provide a method, apparatus, and storage medium for automatically detecting the sarcomere region of cardiomyocytes, so as to solve the problems in the prior art that only single-frame static images are considered, the corresponding region of the most diastolic sarcomere cannot be found during the continuous change process of the sarcomere, resulting in inaccurate selection of the sarcomere region, high time consumption for sample processing, inability to form high-throughput and automated analysis, difficulty in meeting the sarcomere region detection requirements of large-scale cardiomyocyte data, and manual annotation relying on manual operations, being affected by human subjective factors, having inter-individual and intra-individual annotation biases, poor consistency, and low repeatability.

[0005] According to the first aspect of the embodiments of the present invention, a method for automatically detecting the sarcomere region of cardiomyocytes is provided, and the method includes:

[0006] Obtain microscope video data of the rhythmic contraction of cardiomyocytes, perform fusion and sharpening processing on multiple frames of images in the microscope video data to obtain a sharpened enhanced image;

[0007] Calculate the gradients of the sharpened enhanced image in the X and Y directions respectively through the Sobel operator to obtain an X-direction gradient image and a Y-direction gradient image; obtain an X-direction gradient intensity image, a Y-direction gradient intensity image, and an XY-direction correlation image respectively through the X-direction gradient image and the Y-direction gradient image;

[0008] Perform filtering processing on the X-direction gradient intensity image, the Y-direction gradient intensity image, and the XY-direction correlation image through a Gaussian filter and extract the structural intensity to obtain a structural intensity image;

[0009] Obtain a high-structural-intensity region mask image based on the structural intensity image;

[0010] Perform iterative optimization on the high-structural-intensity region mask image through the centroid-driven polar ray method to obtain an internal sarcomere region image of cardiomyocytes.

[0011] Preferably,

[0012] The performing fusion and sharpening processing on multiple frames of images in the microscope video data to obtain a sharpened enhanced image includes:

[0013] Obtain the average image of multiple frames of images in the microscope video data;

[0014] Perform normalization processing on the average image, and perform convolution processing on the normalized average image through a Laplace sharpening filter to obtain an enhanced edge information image;

[0015] Perform normalization processing on the enhanced edge information image to obtain the sharpened enhanced image.

[0016] Preferably,

[0017] Obtaining a high structural strength region mask map based on the structural strength map includes:

[0018] Normalize the structural strength map, extract a binary mask map of the high structural feature region in the normalized structural strength map through Gaussian adaptive threshold segmentation, and invert the pixel values of the binary mask map of the high structural feature region to obtain an inverted binary mask map;

[0019] Perform dilation and erosion on the inverted binary mask map, retain the largest region, and eliminate other small regions to obtain the high structural strength region mask map.

[0020] Preferably,

[0021] Iteratively optimizing the high structural strength region mask map by the centroid-driven polar ray method to obtain a myocardial cell internal sarcomere region map includes:

[0022] Extract the centroid from the high structural strength region mask map through the spatial moment analysis method;

[0023] Based on the centroid, use the centroid-driven polar ray method to obtain an optimized myocardial ROI map; obtain the centroid of the optimized myocardial ROI map, and based on the new centroid, use the centroid-driven polar ray method again to obtain an optimized myocardial ROI map; repeat the iteration, continuously move the centroid and optimize the boundary until the preset target number of iterations is completed or the intersection over union ratio of the optimized myocardial ROI map for three consecutive times to its previous optimized myocardial ROI map is greater than the preset target percentage, then exit the iteration to obtain the final myocardial ROI map;

[0024] Perform centroid regression on the final myocardial ROI map to obtain the myocardial cell internal sarcomere region map.

[0025] Preferably,

[0026] Obtaining the centroid of the optimized myocardial ROI map includes:

[0027] Obtain the distances between all boundary points in the optimized myocardial ROI map and the centroid, and select the point with the largest distance as the new centroid.

[0028] Preferably,

[0029] Based on the centroid, using the centroid-driven polar ray method to obtain an optimized myocardial ROI map includes:

[0030] Centered at the centroid and based on the image coordinate system direction, one ray is emitted every 1° clockwise, for a total of 360 rays. Each ray starts from the centroid and radiates pixel by pixel within the mask until it encounters a mask boundary point, which is recorded as the farthest valid boundary point in that direction. Then, an optimized myocardial ROI map is generated based on the 360 farthest valid boundary points obtained from the current centroid.

[0031] Preferably,

[0032] Performing centroid regression on the final myocardial ROI map to obtain the internal sarcomere region map of the myocardial cells includes:

[0033] Calculating the centroid of the final myocardial ROI map through spatial moments, and obtaining the internal sarcomere region map of the myocardial cells by using the centroid-driven polar ray method based on the centroid of the final myocardial ROI map.

[0034] According to a second aspect of an embodiment of the present invention, there is provided an apparatus for automatically detecting the sarcomere region of myocardial cells, the apparatus including:

[0035] An image fusion module: configured to obtain microscopic video data of the rhythmic contraction of myocardial cells, perform fusion and sharpening processing on multiple frames of images in the microscopic video data to obtain a sharpened enhanced map;

[0036] A structure tensor acquisition module: configured to calculate the gradients of the sharpened enhanced map in the X and Y directions respectively through a Sobel operator to obtain an X-direction gradient map and a Y-direction gradient map; respectively obtain an X-direction gradient intensity map, a Y-direction gradient intensity map, and an XY-direction correlation map through the X-direction gradient map and the Y-direction gradient map;

[0037] A structure intensity acquisition module: configured to perform filtering processing on the X-direction gradient intensity map, the Y-direction gradient intensity map, and the XY-direction correlation map through a Gaussian filter and extract the structure intensity to obtain a structure intensity map;

[0038] A mask map acquisition module: configured to obtain a high-structure-intensity region mask map based on the structure intensity map;

[0039] A sarcomere region map acquisition module: configured to perform iterative optimization on the high-structure-intensity region mask map through the centroid-driven polar ray method to obtain the internal sarcomere region map of the myocardial cells.

[0040] According to a third aspect of an embodiment of the present invention, there is provided a storage medium storing a computer program, which when executed by a main controller, implements each step in the above method.

[0041] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0042] In this application, by acquiring microscopic video data of the rhythmic contraction of cardiomyocytes, multiple frames of images in the microscopic video data are fused and sharpened to obtain a sharpened enhanced image; the gradients of the sharpened enhanced image in the X and Y directions are calculated respectively by the Sobel operator to obtain an X-direction gradient map and a Y-direction gradient map; an X-direction gradient intensity map, a Y-direction gradient intensity map, and an XY-direction correlation map are obtained respectively from the X-direction gradient map and the Y-direction gradient map; the X-direction gradient intensity map, the Y-direction gradient intensity map, and the XY-direction correlation map are filtered by a Gaussian filter and the structural intensity is extracted to obtain a structural intensity map; a high structural intensity region mask map is obtained based on the structural intensity map; the high structural intensity region mask map is iteratively optimized by the centroid-driven polar ray method to obtain a sarcomere region map inside the cardiomyocytes. By calculating the structural features of the image after fusing multiple frames of images, this application obtains a pan-sarcomere region, and by the centroid-driven polar ray method, it obtains an inscribed contour of the smooth sarcomere region. Compared with the prior art, it does not require positioning of cardiomyocytes, directly locates the sarcomere by the structural intensity, and the obtained sarcomere ROI is more suitable for subsequent calculation of sarcomere motion parameters. The calculation process is efficient and fully automatic, and is not affected by human subjective factors.

[0043] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The drawings herein are incorporated into the specification and constitute a part of this specification, showing embodiments consistent with the present invention, and are used together with the specification to explain the principles of the present invention.

[0045] Figure 1 is a schematic flowchart of a method for automatically detecting the sarcomere region of cardiomyocytes shown according to an exemplary embodiment;

[0046] Figure 2 is a schematic diagram for obtaining a sharpened enhanced image shown according to another exemplary embodiment;

[0047] Figure 3 is a schematic diagram for obtaining a structure tensor shown according to another exemplary embodiment;

[0048] Figure 4 is a schematic diagram for obtaining a high structural intensity region mask shown according to another exemplary embodiment;

[0049] Figure 5 is a schematic diagram of iterative optimization shown according to another exemplary embodiment;

[0050] Figure 6 is a schematic diagram for obtaining a sarcomere region shown according to another exemplary embodiment;

[0051] Figure 7 It is a system schematic diagram of a device for automatically detecting the sarcomere region of cardiomyocytes shown according to another exemplary embodiment;

[0052] In the drawings: 1 - Image fusion module, 2 - Structure tensor acquisition module, 3 - Structure strength acquisition module, 4 - Mask map acquisition module, 5 - Sarcomere region map acquisition module. Detailed implementation mode

[0053] Here, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation manners described in the following exemplary embodiments do not represent all implementation manners consistent with the present invention. On the contrary, they are only examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0054] Embodiment 1

[0055] Figure 1 It is a flowchart of a method for automatically detecting the sarcomere region of cardiomyocytes shown according to an exemplary embodiment, as Figure 1 shown, the method includes:

[0056] S1. Obtain microscopic video data of the rhythmic contraction of cardiomyocytes, perform fusion and sharpening processing on multiple frames of images in the microscopic video data to obtain a sharpened enhanced map;

[0057] S2. Calculate the gradients of the sharpened enhanced map in the X and Y directions respectively through the Sobel operator to obtain an X-direction gradient map and a Y-direction gradient map; obtain an X-direction gradient intensity map, a Y-direction gradient intensity map, and an XY-direction correlation map respectively through the X-direction gradient map and the Y-direction gradient map;

[0058] S3. Perform filtering processing on the X-direction gradient intensity map, the Y-direction gradient intensity map, and the XY-direction correlation map through a Gaussian filter and extract the structure strength to obtain a structure strength map;

[0059] S4. Obtain a high-structure-strength region mask map based on the structure strength map;

[0060] S5. Perform iterative optimization on the high-structure-strength region mask map through the centroid-driven polar ray method to obtain an internal sarcomere region map of cardiomyocytes;

[0061] It can be understood that as attached Figure 2As shown, in this application, the rhythmic contraction of cardiomyocytes is observed through an optical microscope imaging pixel system, and microscopic video data is continuously captured. The average image of multiple frames in the microscopic video data is calculated; the average image is normalized so that its pixel range is changed from (0 - 255) to (0 - 1). The normalized average image is convolved using a Laplacian sharpening filter to obtain an enhanced edge information image. This operation can highlight the details in the image and improve the resolution ability for tiny structures such as the sarcomere region; the enhanced image is normalized again, and the pixel range is adjusted back to (0 - 255) to obtain a sharpened enhanced image.

[0062] As shown in the Figure 3 appendix, based on the above sharpened enhanced image, the Sobel operator is used to calculate the gradients in the X and Y directions, obtaining an X-direction gradient map and a Y-direction gradient map. The structure tensor is calculated, including the X-direction gradient intensity J11, the Y-direction gradient intensity J22, and the XY-direction correlation J12. The specific calculation method is as follows:

[0063] J11 = sobel_x * sobel_x

[0064] J22 = sobel_y * sobel_y

[0065] J12 = sobel_x * sobel_y

[0066] As shown in the Figure 4 appendix, the Gaussian filter is used to filter J11, J22, and J12 respectively to make them smoother, and the structure strength structure_strength is calculated. The specific calculation method is as follows:

[0067] Lambda1 = (J11 + J22) / 2 + sqrt((J11 - J22)*(J11 - J22) / 4 + J12 * J12);

[0068] Lambda2 = (J11 + J22) / 2 - sqrt((J11 - J22)*(J11 - J22) / 4 + J12 * J12);

[0069] structure_strength = Lambda1 - Lambda2;

[0070] Normalize structure_strength to the range of (0 - 255). Through Gaussian adaptive threshold segmentation, extract the binary mask of the region with high structural features, and invert the pixel values to obtain the inverted binary mask image. Perform dilation and erosion on the region with high structural features in the inverted binary mask image, retain the largest region, and eliminate other small regions to obtain the mask image of the region with high structural strength;

[0071] As shown in the appendix Figure 5 Extract the initial centroid of the mask image of the region with high structural strength through the spatial moment analysis method, and use the centroid-driven polar ray method to iteratively optimize the myocardial ROI (ROI represents the region of interest) to eliminate the influence of the edge structure. The centroid-driven polar ray method is specifically as follows: Taking the centroid as the center and the image coordinate system direction as the reference, emit 1 ray every 1° clockwise, a total of 360 rays are emitted. Each ray starts from the centroid and radiates pixel by pixel within the mask until it encounters the mask boundary point, which is recorded as the farthest valid boundary point in this direction. Then, generate a new myocardial ROI based on the 360 farthest valid boundary points obtained from the current centroid. The first generated myocardial ROI is called the initial myocardial ROI; Based on the initial myocardial ROI, calculate the distance between all boundary points and the centroid, select the point with the maximum distance, and use it as the new centroid. Then, use the centroid-driven polar ray method again to shift the centroid towards the external structure of the ROI, thereby optimizing the boundary position and generating a new myocardial ROI; Calculate the distance between all boundary points and the centroid based on the new myocardial ROI, select the point with the maximum distance, and use it as the new centroid. Adopt the centroid-driven polar ray method and repeat this iterative process, continuously moving the centroid and optimizing the boundary. The shape of the myocardial ROI after each optimization gradually becomes smooth, and it gradually fits the actual sarcomere region until the target number of iterations is completed, or, the intersection over union of the myocardial ROI and its previous myocardial ROI is greater than the target percentage for three consecutive times, then exit the iteration, and take the myocardial ROI obtained in the last iteration as the final myocardial ROI map;

[0072] As shown in the appendix Figure 6 Calculate the centroid of the final myocardial ROI map through spatial moments to make it return to a more balanced position. Based on the centroid of the final myocardial ROI map, use the centroid-driven polar ray method to generate the final mask and accurately extract the sarcomere region inside the myocardial cells.

[0073] Embodiment 2:

[0074] Figure 7 is a system schematic diagram of a device for automatically detecting the sarcomere region of myocardial cells shown according to another exemplary embodiment. The device includes:

[0075] Image fusion module 1: Used to obtain the microscopic video data of the rhythmic contraction of myocardial cells, fuse and sharpen multiple frames of images in the microscopic video data to obtain a sharpened enhanced image;

[0076] Structure tensor acquisition module 2: It is used to calculate the gradients of the sharpened enhanced image in the X and Y directions respectively through the Sobel operator to obtain the X-direction gradient image and the Y-direction gradient image; and to obtain the X-direction gradient intensity image, the Y-direction gradient intensity image, and the XY-direction correlation image respectively through the X-direction gradient image and the Y-direction gradient image;

[0077] Structure intensity acquisition module 3: It is used to perform filtering processing on the X-direction gradient intensity image, the Y-direction gradient intensity image, and the XY-direction correlation image through a Gaussian filter and extract the structure intensity to obtain a structure intensity image;

[0078] Mask image acquisition module 4: It is used to obtain a high-structure-intensity region mask image based on the structure intensity image;

[0079] Sarcomere region image acquisition module 5: It is used to perform iterative optimization on the high-structure-intensity region mask image through the centroid-driven polar ray method to obtain the sarcomere region image inside cardiomyocytes.

[0080] Embodiment 3:

[0081] This embodiment provides a storage medium that stores a computer program. When the computer program is executed by a main controller, it implements each step in the above method;

[0082] It can be understood that the above-mentioned storage medium can be a read-only memory, a disk, an optical disc, etc.

[0083] It can be understood that the same or similar parts in the above embodiments can be referred to each other, and the content not described in detail in some embodiments can be referred to the same or similar content in other embodiments.

[0084] It should be noted that in the description of the present invention, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, the meaning of "sparsely distributed in small amounts" means at least two.

[0085] Any process or method description shown in the flowchart or described in other ways herein can be understood as representing a module, segment, or part of executable instructions including one or more sparsely distributed steps for implementing a specific logical function or process. The scope of the preferred embodiments of the present invention includes additional implementations, where the functions can be executed in a substantially simultaneous manner or in the reverse order according to the involved functions, rather than in the order shown or discussed, which should be understood by those skilled in the technical field to which the embodiments of the present invention belong.

[0086] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, a small number of sparsely distributed steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), and the like.

[0087] Those of ordinary skill in the art can understand that all or part of the steps carried by the method of the above embodiments can be completed by instructing relevant hardware through a program. The said program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0088] In addition, in each of the embodiments of the present invention, the functional units can be integrated into a processing module, or each unit can exist physically alone, or two or more units can be integrated into one module. The above integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0089] The above-mentioned storage medium can be a read-only memory, a magnetic disk, an optical disc, or the like.

[0090] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or a small number of sparsely distributed embodiments or examples.

[0091] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for automatically detecting the sarcomere region of cardiomyocytes, characterized in that, The method includes: Obtaining microscopic video data of the rhythmic contraction of cardiomyocytes, fusing and sharpening multiple frames of images in the microscopic video data to obtain a sharpened enhanced image; Calculating the gradients of the sharpened enhanced image in the X and Y directions respectively through a Sobel operator to obtain an X-direction gradient map and a Y-direction gradient map; obtaining an X-direction gradient intensity map, a Y-direction gradient intensity map, and an XY-direction correlation map respectively through the X-direction gradient map and the Y-direction gradient map; Filtering the X-direction gradient intensity map, the Y-direction gradient intensity map, and the XY-direction correlation map through a Gaussian filter and extracting the structural intensity to obtain a structural intensity map; Obtaining a high structural intensity region mask map based on the structural intensity map; Performing iterative optimization on the high structural intensity region mask map through a centroid-driven polar ray method to obtain a sarcomere region map inside the cardiomyocyte; The performing iterative optimization on the high structural intensity region mask map through a centroid-driven polar ray method to obtain a sarcomere region map inside the cardiomyocyte includes: Extracting the centroid from the high structural intensity region mask map through a spatial moment analysis method; Obtaining an optimized myocardial ROI map by using the centroid-driven polar ray method based on the centroid; obtaining the centroid of the optimized myocardial ROI map, and obtaining an optimized myocardial ROI map again by using the centroid-driven polar ray method based on the new centroid; repeating the iteration, continuously moving the centroid and optimizing the boundary until the preset target iteration number is completed or the intersection over union of the optimized myocardial ROI map and its previous optimized myocardial ROI map is greater than the preset target percentage, and then exiting the iteration to obtain the final myocardial ROI map; Performing centroid regression on the final myocardial ROI map to obtain the sarcomere region map inside the cardiomyocyte; The obtaining the centroid of the optimized myocardial ROI map includes: Obtaining the distances between all boundary points and the centroid in the optimized myocardial ROI map, and selecting the point with the largest distance as the new centroid.

2. The method according to claim 1, wherein The fusing and sharpening multiple frames of images in the microscopic video data to obtain a sharpened enhanced image includes: Obtaining the average image of multiple frames of images in the microscopic video data; Performing normalization processing on the average image, and performing convolution processing on the normalized average image through a Laplacian sharpening filter to obtain an enhanced edge information image; Performing normalization processing on the enhanced edge information image to obtain the sharpened enhanced image.

3. The method according to claim 2, wherein The obtaining a high structural intensity region mask map based on the structural intensity map includes: Performing normalization processing on the structural intensity map, extracting a binary mask map of the high structural feature region in the normalized structural intensity map through Gaussian adaptive threshold segmentation, and inverting the pixel values of the binary mask map of the high structural feature region to obtain an inverted binary mask map; Performing dilation and erosion on the inverted binary mask map, retaining the largest region, and eliminating other small regions to obtain the high structural intensity region mask map.

4. The method according to claim 3, wherein the optimized myocardial ROI map obtained by the centroid-driven polar ray method based on the centroid includes: Taking the centroid as the center and based on the image coordinate system direction, 1 ray is emitted every 1° clockwise, a total of 360 rays are emitted. Each ray starts from the centroid and radiates pixel by pixel within the mask until it encounters a mask boundary point, which is recorded as the farthest valid boundary point in that direction. Then, an optimized myocardial ROI map is generated according to the 360 farthest valid boundary points obtained from the current centroid.

5. The method according to claim 4, wherein the centroid regression of the final myocardial ROI map to obtain the sarcomere region map inside the myocardial cell includes: Calculating the centroid of the final myocardial ROI map through spatial moments, and obtaining the sarcomere region map inside the myocardial cell by the centroid-driven polar ray method based on the centroid of the final myocardial ROI map.

6. An apparatus for automatically detecting the sarcomere region of cardiomyocytes, characterized in that, The device includes: An image fusion module: configured to acquire microscopic video data of the rhythmic contraction of myocardial cells, fuse and sharpen multiple frames of images in the microscopic video data to obtain a sharpened enhanced map; A structure tensor acquisition module: configured to calculate the gradients of the sharpened enhanced map in the X and Y directions respectively through a Sobel operator to obtain an X-direction gradient map and a Y-direction gradient map; obtain an X-direction gradient intensity map, a Y-direction gradient intensity map, and an XY-direction correlation map respectively through the X-direction gradient map and the Y-direction gradient map; A structure intensity acquisition module: configured to filter and extract the structure intensity of the X-direction gradient intensity map, the Y-direction gradient intensity map, and the XY-direction correlation map through a Gaussian filter to obtain a structure intensity map; A mask map acquisition module: configured to acquire a high-structure-intensity region mask map based on the structure intensity map; A sarcomere region map acquisition module: configured to iteratively optimize the high-structure-intensity region mask map by the centroid-driven polar ray method to obtain a sarcomere region map inside the myocardial cell; The iteratively optimizing the high-structure-intensity region mask map by the centroid-driven polar ray method to obtain a sarcomere region map inside the myocardial cell includes: Extracting the centroid from the high-structure-intensity region mask map through a spatial moment analysis method; Obtaining an optimized myocardial ROI map by the centroid-driven polar ray method based on the centroid; obtaining the centroid of the optimized myocardial ROI map, and obtaining an optimized myocardial ROI map again by the centroid-driven polar ray method based on the new centroid; repeating the iteration, continuously moving the centroid and optimizing the boundary until the preset target iteration number is completed or the intersection over union of the optimized myocardial ROI map and its previous optimized myocardial ROI map is greater than the preset target percentage for three consecutive times, exiting the iteration, and obtaining the final myocardial ROI map; Performing centroid regression on the final myocardial ROI map to obtain the sarcomere region map inside the myocardial cell; The obtaining the centroid of the optimized myocardial ROI map includes: Obtaining the distances between all boundary points in the optimized myocardial ROI map and the centroid, and selecting the point with the largest distance as the new centroid.

7. A storage medium, characterized in that, The storage medium stores a computer program, and when the computer program is executed by the main controller, each step in a method for automatically detecting the sarcomere region of cardiomyocytes as described in any one of claims 1-5 is implemented.

Images