A method, apparatus, device and medium for determining the phase of the cardiac cycle

Abstract

The application discloses a heart cycle phase determination method and device, equipment and medium, and relates to the technical field of computers, which comprises the following steps: determining the target mask of each cardiac chamber in a to-be-processed echocardiogram; determining the ventricular area and the atrial area in the to-be-processed echocardiogram by using the contour information of each cardiac chamber determined based on the target mask; and determining the heart cycle phase corresponding to the to-be-processed echocardiogram based on the change value corresponding to the area difference between the ventricular area and the atrial area. The application determines the contour information of each cardiac chamber in the to-be-processed echocardiogram by using the target mask, and then calculates the area difference between the ventricular area and the atrial area according to the contour information, so as to determine the heart cycle phase of the to-be-processed echocardiogram based on the change value corresponding to the area difference. The above scheme can determine the heart cycle phase of the to-be-processed echocardiogram in real time and more intelligently.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method, apparatus, device, and medium for determining the phase of the cardiac cycle. Background Technology

[0002] Currently, when determining the cardiac cycle phase in echocardiogram images, physicians can manually acquire the cardiac cycle by activating PW (Pulsed-wave Doppler) mode. This process requires physician involvement throughout, thus incurring significant human resource costs.

[0003] In addition, in order to obtain cardiac cycle images, it is necessary to use the ultrasound video stored in the ultrasound equipment, which cannot achieve real-time performance.

[0004] In summary, how to determine the cardiac cycle phase of the echocardiogram image to be processed in a real-time and more intelligent manner is a problem that needs to be solved. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a method, apparatus, device, and medium for determining the cardiac cycle phase, which can determine the cardiac cycle phase of the echocardiogram image to be processed in real time and more intelligently. The specific solution is as follows:

[0006] In a first aspect, this application discloses a method for determining the phase of the cardiac cycle, including:

[0007] Determine the target mask for each cardiac chamber in the cardiac ultrasound image to be processed;

[0008] Using the contour information of each heart chamber determined based on the target mask, the ventricular area and atrial area in the cardiac ultrasound image to be processed are determined;

[0009] The cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined based on the change value corresponding to the area difference between the ventricular area and the atrial area.

[0010] Optionally, determining the target mask for each cardiac chamber in the cardiac ultrasound image to be processed includes:

[0011] The cardiac ultrasound image to be processed is processed using a pre-built chamber structure segmentation model based on a preset convolutional neural network to obtain the target mask for each cardiac chamber.

[0012] Optionally, the preset convolutional neural network includes a CNN backbone network and a feature decoder;

[0013] Accordingly, the process of using a pre-built chamber structure segmentation model based on a preset convolutional neural network to process the cardiac ultrasound image to obtain the target mask for each cardiac chamber includes:

[0014] High-level semantic features in the cardiac ultrasound image to be processed are extracted using the CNN backbone network in the chamber structure segmentation model.

[0015] The high-level semantic features are decoded using the feature decoder to obtain the target mask for each heart chamber.

[0016] Optionally, the step of extracting high-level semantic features from the cardiac ultrasound image to be processed using the CNN backbone network in the chamber structure segmentation model includes:

[0017] The CNN backbone network in the chamber structure segmentation model is used to extract low-level semantic features from the cardiac ultrasound image to be processed; the low-level semantic features include edge feature information, contour feature information and gray-level feature information.

[0018] The lower-level semantic features are fused and encoded to obtain higher-level semantic features.

[0019] Optionally, the feature decoder includes a first feature decoder and a second feature decoder;

[0020] Accordingly, the step of decoding the high-level semantic features using the feature decoder to obtain the target mask for each cardiac chamber includes:

[0021] The location and index information of each heart chamber are obtained by decoding the high-level semantic features using the first feature decoder.

[0022] The high-level semantic features are decoded using the second feature decoder to obtain the three-dimensional mask of each heart chamber;

[0023] The two-dimensional masks corresponding to the index information are filtered from the three-dimensional masks, and the two-dimensional masks are corrected based on the position information to obtain the target masks for each heart chamber.

[0024] Optionally, the location information is the minimum bounding rectangle corresponding to each of the heart chambers;

[0025] Accordingly, the step of correcting the two-dimensional mask based on the location information to obtain the target mask for each cardiac chamber includes:

[0026] The two-dimensional mask is cropped using the minimum bounding rectangle so that the part of the two-dimensional mask located inside the minimum bounding rectangle is used as the target mask.

[0027] Optionally, the training set corresponding to the chamber structure segmentation model includes training images obtained by enhancing historical echocardiogram images and corresponding training labels; the training labels include the boundary range of each cardiac chamber structure in the training images.

[0028] Optionally, before determining the ventricular and atrial areas in the cardiac ultrasound image to be processed, the method further includes:

[0029] The target mask is converted into initial contour information for each of the heart chambers using computer vision algorithms;

[0030] The initial contour information is post-processed using a preset post-processing algorithm to obtain the contour information of each heart chamber.

[0031] Optionally, the preset post-processing algorithm includes any one or more of the following algorithms: image denoising algorithm, image smoothing algorithm, and image edge cutting algorithm.

[0032] Optionally, determining the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the change value corresponding to the area difference between the ventricular area and the atrial area includes:

[0033] The area difference between the ventricular area and the atrial area is recorded in a preset area difference change graph; the preset area difference change graph is used to record the area difference between the ventricular area and the atrial area in consecutive different cardiac ultrasound images;

[0034] Determine the gradient rate of change of the area difference corresponding to the cardiac ultrasound image to be processed in the current preset area difference change map;

[0035] The cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined based on the gradient change rate.

[0036] Optionally, determining the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the gradient change rate includes:

[0037] If the gradient change rate is positive, then the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be diastolic.

[0038] If the gradient change rate is negative, then the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be the systolic phase.

[0039] If the gradient change rate is zero, the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined using the frame number of the cardiac ultrasound image to be processed and the cardiac cycle length determined based on the preset area difference change map.

[0040] Optionally, determining the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed using the frame number of the cardiac ultrasound image to be processed and the cardiac cycle length determined based on the preset area difference change map includes:

[0041] Periodic fitting is performed on the preset area difference change map to determine the cardiac cycle length;

[0042] Obtain the first target frame number corresponding to the previous extreme point adjacent to the position point of the cardiac ultrasound image to be processed in the preset area difference change map;

[0043] Based on the cardiac cycle length and the first target frame number, predict the second target frame number corresponding to the next extreme point adjacent to the previous extreme point in the preset area difference change map;

[0044] The location type of the location point corresponding to the cardiac ultrasound image to be processed is determined by comparing the frame number of the cardiac ultrasound image to be processed with the second target frame number and the type of the previous extreme point.

[0045] The cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined based on the location type.

[0046] Optionally, determining the location type of the location point corresponding to the cardiac ultrasound image to be processed using the comparison result between the frame number of the cardiac ultrasound image to be processed and the second target frame number, and the type of the previous extreme point, includes:

[0047] If the difference between the frame number of the cardiac ultrasound image to be processed and the second target frame number is within a preset difference range, then the position type of the corresponding position point of the cardiac ultrasound image to be processed is determined according to the type of the previous extreme point; wherein, if the type of the previous extreme point is a maximum value, then the position type is a trough position; if the type of the previous extreme point is a minimum value, then the position type is a peak position.

[0048] Optionally, determining the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the location type includes:

[0049] If the location type is a peak position, then the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be diastolic.

[0050] If the location type is a trough location, then the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be the systolic phase.

[0051] Optionally, determining the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the change value corresponding to the area difference between the ventricular area and the atrial area includes:

[0052] Determine the difference between the area difference between the ventricular area and the atrial area and the area difference of the previous frame; the area difference of the previous frame is the area difference between the ventricular area and the atrial area in the previous frame of processed cardiac ultrasound image relative to the cardiac ultrasound image to be processed.

[0053] The cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined based on the difference.

[0054] Optionally, determining the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the difference includes:

[0055] If the difference is greater than zero, the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be diastolic.

[0056] If the difference is less than zero, the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be the systolic phase.

[0057] If the difference is zero, then the cardiac cycle phase of the previous processed cardiac ultrasound image is taken as the cardiac cycle phase of the cardiac ultrasound image to be processed.

[0058] Secondly, this application discloses a cardiac cycle phase determination device, comprising:

[0059] The mask determination module is used to determine the target mask for each cardiac chamber in the cardiac ultrasound image to be processed;

[0060] An area determination module is used to determine the ventricular area and atrial area in the cardiac ultrasound image to be processed by using the contour information of each of the heart chambers determined based on the target mask.

[0061] The cycle phase determination module is used to determine the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the change value corresponding to the area difference between the ventricular area and the atrial area.

[0062] Thirdly, this application discloses an electronic device, including:

[0063] Memory, used to store computer programs;

[0064] A processor is configured to execute the computer program to implement the steps of the aforementioned disclosed method for determining cardiac cycle phases.

[0065] Fourthly, this application discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the aforementioned disclosed method for determining the phase of the cardiac cycle.

[0066] As can be seen, this application determines the target mask of each heart chamber in the echocardiogram image to be processed; uses the contour information of each heart chamber determined based on the target mask to determine the ventricular area and atrial area in the echocardiogram image to be processed; and determines the cardiac cycle phase corresponding to the echocardiogram image to be processed based on the change value corresponding to the area difference between the ventricular area and the atrial area. Therefore, this application first needs to determine the target mask of each heart chamber in the echocardiogram image to be processed, and then use the target mask to determine the contour information of each heart chamber. Further, the ventricular area and atrial area are calculated based on the contour information, and finally, the cardiac cycle phase of the echocardiogram image to be processed is determined based on the change value corresponding to the area difference between the ventricular area and the atrial area. The echocardiogram image to be processed can be a real-time acquired current frame echocardiogram image or a historical echocardiogram image. If it is a current frame echocardiogram image, the cardiac cycle phase of the current frame echocardiogram image can be determined in real time. In addition, this application determines the cardiac cycle phase of the cardiac ultrasound image to be processed by automatically calculating the difference in ventricular and atrial area. The entire process does not require the intervention of a doctor, thus making it more intelligent. Attached Figure Description

[0067] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0068] Figure 1 This is a flowchart of a method for determining the phase of the cardiac cycle disclosed in this application;

[0069] Figure 2 This is a flowchart of a specific method for determining cardiac cycle phases disclosed in this application;

[0070] Figure 3 This is a schematic diagram of the network structure of a cavity structure segmentation model disclosed in this application;

[0071] Figure 4 This is a flowchart illustrating the process of a chamber structure segmentation model disclosed in this application;

[0072] Figure 5 This is a flowchart of the cardiac cycle phase determination method based on gradient rate of change disclosed in this application;

[0073] Figure 6 This is a flowchart of the method for dividing cardiac cycle phases based on gradient change rate disclosed in this application;

[0074] Figure 7 This is a flowchart of the cardiac cycle phase determination method based on the area difference between two adjacent frames disclosed in this application;

[0075] Figure 8 This is a schematic diagram of a cardiac cycle phase determination device disclosed in this application;

[0076] Figure 9 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0077] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0078] Currently, determining the cardiac cycle phase in echocardiogram images requires physician participation throughout the entire process, consuming significant manpower. Furthermore, related technologies rely on stored ultrasound video to acquire cardiac cycle images, making real-time determination impossible. Therefore, this application discloses a method, apparatus, device, and medium for determining the cardiac cycle phase in echocardiogram images, enabling real-time and more intelligent determination of the cardiac cycle phase in the images to be processed.

[0079] See Figure 1 As shown in the embodiment of this application, a method for determining the phase of the cardiac cycle is disclosed, the method comprising:

[0080] Step S11: Determine the target mask for each cardiac chamber in the cardiac ultrasound image to be processed.

[0081] In this embodiment, a cardiac ultrasound image to be processed is acquired, and a target mask for each cardiac chamber in the cardiac ultrasound image is determined. The cardiac ultrasound image to be processed can be a real-time acquired current frame cardiac ultrasound image or a historical cardiac ultrasound image stored in the ultrasound equipment; this embodiment does not impose any limitations on this. The cardiac ultrasound image to be processed can be a single image or an image set composed of multiple images. If the cardiac ultrasound image to be processed is an image set, each cardiac ultrasound image can be processed separately.

[0082] Furthermore, the image format of the cardiac ultrasound image to be processed can be RGB format. Alternatively, the cardiac ultrasound image to be processed can be a grayscale image output by the ultrasound equipment.

[0083] The cardiac ultrasound image to be processed can be a cardiac ultrasound image containing a preset chamber, for example, an ultrasound image containing four complete chambers.

[0084] Step S12: Using the contour information of each heart chamber determined based on the target mask, determine the ventricular area and atrial area in the cardiac ultrasound image to be processed.

[0085] In this embodiment, the contour information of each heart chamber is determined using a target mask, and the ventricular area and atrial area are further calculated based on the contour information. Here, the ventricular area includes the left ventricular area and the right ventricular area, which can be the sum of the left ventricular area and the right ventricular area; the atrial area includes the left atrial area and the right atrial area, which can be the sum of the left atrial area and the right atrial area.

[0086] It should be noted that before determining the ventricular and atrial areas in the cardiac ultrasound image to be processed, the process further includes: converting the target mask into initial contour information of each cardiac chamber using a computer vision algorithm; and post-processing the initial contour information using a preset post-processing algorithm to obtain the contour information of each cardiac chamber. That is, after obtaining the target mask, this embodiment uses a traditional computer vision algorithm (CV algorithm) to convert the target mask into initial contour information of each cardiac chamber. Furthermore, a preset post-processing algorithm is used to post-process the initial contour information to obtain the contour information of each cardiac chamber. It is understood that the initial contours obtained directly from the target mask may have edge noise, rough edges, etc., therefore, a preset post-processing algorithm is needed to perform post-processing operations to obtain the contour information of each cardiac chamber.

[0087] The aforementioned pre-defined post-processing algorithms include, but are not limited to, image denoising algorithms, image smoothing algorithms, and image edge cropping algorithms. It should be noted that when using image edge cropping algorithms for edge cropping, the initial contour information can be optimized in conjunction with the original image information to ensure that the optimized contour information does not overflow the actual boundary range.

[0088] Step S13: Determine the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the change value corresponding to the area difference between the ventricular area and the atrial area.

[0089] In this embodiment, the area difference between the ventricular area and the atrial area is calculated by subtracting the sum of the left and right atrial areas from the sum of the left and right ventricular areas. It is understood that the ventricular area is usually larger than the atrial area; therefore, this embodiment uses the method of subtracting the atrial area from the ventricular area to ensure that the obtained area difference is positive.

[0090] Furthermore, after obtaining the area difference between the ventricular and atrial areas, the change value corresponding to the area difference is used to determine the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed. In this embodiment, the main focus is on determining whether the cardiac ultrasound image to be processed is in the diastolic or systolic phase of the cardiac cycle; specifically, when in diastole, the ventricular area continuously increases while the atrial area continuously decreases; when in systole, the ventricular area continuously decreases while the atrial area continuously increases.

[0091] As can be seen, this application determines the target mask of each heart chamber in the echocardiogram image to be processed; uses the contour information of each heart chamber determined based on the target mask to determine the ventricular area and atrial area in the echocardiogram image to be processed; and determines the cardiac cycle phase corresponding to the echocardiogram image to be processed based on the change value corresponding to the area difference between the ventricular area and the atrial area. Therefore, this application first needs to determine the target mask of each heart chamber in the echocardiogram image to be processed, and then use the target mask to determine the contour information of each heart chamber. Further, the ventricular area and atrial area are calculated based on the contour information, and finally, the cardiac cycle phase of the echocardiogram image to be processed is determined based on the change value corresponding to the area difference between the ventricular area and the atrial area. The echocardiogram image to be processed can be a real-time acquired current frame echocardiogram image or a historical echocardiogram image. If it is a current frame echocardiogram image, the cardiac cycle phase of the current frame echocardiogram image can be determined in real time. In addition, this application determines the cardiac cycle phase of the cardiac ultrasound image to be processed by automatically calculating the difference in ventricular and atrial area. The entire process does not require the intervention of a doctor, thus making it more intelligent.

[0092] See Figure 2 As shown, this application discloses a specific method for determining the phase of the cardiac cycle. Compared to the previous embodiment, this embodiment further explains and optimizes the technical solution. Specifically, it includes:

[0093] Step S21: Process the cardiac ultrasound image to be processed using a pre-built chamber structure segmentation model based on a preset convolutional neural network to obtain the target mask for each cardiac chamber.

[0094] In this embodiment, a chamber structure segmentation model for cardiac chamber structure segmentation is pre-constructed based on a preset convolutional neural network. Therefore, the cardiac ultrasound image to be processed is processed using this chamber structure segmentation model to obtain the target mask of each cardiac chamber.

[0095] Specifically, the training set corresponding to the aforementioned chamber structure segmentation model includes training images obtained by enhancing historical echocardiogram images and corresponding training labels; the training labels include the boundary ranges of each cardiac chamber structure in the training images. That is, during model training, a large number of training images labeled with the boundary ranges of each cardiac chamber structure are used. These training images can specifically be historical echocardiogram images, and different enhancement methods are employed to enhance them. Enhancement methods include, but are not limited to, horizontal flipping, vertical flipping, mirror-fill rotation, mosaic enhancement, affine transformation, random contrast enhancement, random brightness enhancement, and image cropping. Furthermore, to enhance the learned chamber structure segmentation model's learning and generalization abilities, targeted training can be conducted for different target audiences. For example, if the goal is to determine fetal cardiac phases, the training data could include fetal echocardiogram images covering different gestational ages and regions.

[0096] Furthermore, for details on the network structure of the chamber structure segmentation model, please refer to [link to relevant documentation]. Figure 3 As shown, according to Figure 3 As can be seen, the aforementioned pre-defined convolutional neural network specifically includes a CNN backbone network and a feature decoder; see the relevant documentation for details. Figure 4 As shown, step S21 may specifically include:

[0097] Step S211: Extract high-level semantic features from the cardiac ultrasound image to be processed using the CNN backbone network in the chamber structure segmentation model.

[0098] In this embodiment, the preset convolutional neural network specifically includes a CNN (Convolutional Neural Networks) backbone network and a feature decoder. The CNN backbone network is mainly used for image feature extraction. Therefore, when acquiring the cardiac ultrasound image to be processed and processing it using the chamber structure segmentation model, the first step is to use the CNN backbone network in the chamber structure segmentation model to extract high-level semantic features from the cardiac ultrasound image to be processed.

[0099] In a specific implementation, the extraction of high-level semantic features from the cardiac ultrasound image to be processed using the CNN backbone network in the chamber structure segmentation model includes: extracting low-level semantic features from the cardiac ultrasound image to be processed using the CNN backbone network in the chamber structure segmentation model; the low-level semantic features include edge feature information, contour feature information, and grayscale feature information; and performing feature fusion and encoding processing on the low-level semantic features to obtain high-level semantic features. That is, the CNN backbone network first extracts low-level semantic features from the cardiac ultrasound image to be processed, including but not limited to edge feature information, contour feature information, and grayscale feature information, and then performs feature fusion and encoding processing on these low-level semantic features in a high-level convolutional network to generate high-level semantic features.

[0100] Step S212: Use the feature decoder to decode the high-level semantic features to obtain the target mask for each heart chamber.

[0101] In this embodiment, after extracting the high-level semantic features of the cardiac ultrasound image to be processed using a CNN backbone network, the high-level semantic features are then decoded using a feature decoder to obtain the target mask for each cardiac chamber.

[0102] In specific implementations, such as Figure 3As shown, the aforementioned feature decoder includes a first feature decoder and a second feature decoder. Correspondingly, the step of decoding the high-level semantic features using the feature decoder to obtain the target mask for each heart chamber includes: decoding the high-level semantic features using the first feature decoder to obtain the location information and index information of each heart chamber; decoding the high-level semantic features using the second feature decoder to obtain a three-dimensional mask for each heart chamber; filtering the two-dimensional masks corresponding to the index information from the three-dimensional masks, and correcting the two-dimensional masks based on the location information to obtain the target mask for each heart chamber. That is, the feature decoder used to decode the high-level semantic features includes a first feature decoder and a second feature decoder, wherein the first feature decoder is used to decode the high-level semantic features to obtain the location information and index information of each heart chamber, and the second feature decoder is used to decode the high-level semantic features to obtain a three-dimensional mask for each heart chamber. It should be noted that the 3D mask here is composed of several stacked channels. Each channel can be understood as 2D mask information, and only one channel among the several channels contains valid 2D mask information. The index information output by the first feature decoder can accurately locate this channel from the several channels. Here, the index information refers to the index value of the channel dimension where the 2D mask is located in the 3D mask output by the network. Therefore, in this embodiment, the first feature decoder and the second feature decoder will transmit information. That is, the index information output by the first feature decoder is used to filter out the valid 2D mask from the 3D mask output by the second feature decoder, and then the position information output by the first feature decoder is used to correct the 2D mask to obtain the target mask for each heart chamber. In other words, in this embodiment, one decoder outputs index information and the other decoder outputs the 3D mask. By combining the two decoders, the valid 2D mask can be accurately and quickly located from the 3D mask based on the index information, and the remaining invalid 2D mask information can be filtered out, which improves the accuracy of the target mask and reduces the amount of data computation.

[0103] For example, suppose the index information output by the first decoder is 8, and the three-dimensional mask output by the second decoder is composed of 32 channels stacked together, and the index values ​​of these 32 channels are marked as 1 to 32 respectively; therefore, the two-dimensional mask with an index value of 8 is selected from the three-dimensional mask, and the position information output by the first feature decoder is used to correct the two-dimensional mask to obtain the target mask of each heart chamber.

[0104] In a specific implementation, the aforementioned location information refers to the minimum bounding rectangle corresponding to each of the heart chambers. Correspondingly, the step of correcting the two-dimensional mask based on the location information to obtain the target mask for each heart chamber includes: cropping the two-dimensional mask using the minimum bounding rectangle to use the portion of the two-dimensional mask located within the minimum bounding rectangle as the target mask. It can be understood that the location information can be represented by generating a corresponding minimum bounding rectangle based on the actual structure of each heart chamber in the image to determine the approximate location and boundary of each heart chamber, and then cropping the two-dimensional mask using the minimum bounding rectangle to use the portion of the two-dimensional mask located within the minimum bounding rectangle as the target mask. That is, this embodiment removes masks that exceed the range of the minimum bounding rectangle, thereby reducing errors and making the target mask obtained after correction more closely match the actual structure of the heart chamber, facilitating more accurate measurement of the heart chamber area in subsequent measurements.

[0105] Step S22: Using the contour information of each heart chamber determined based on the target mask, determine the ventricular area and atrial area in the cardiac ultrasound image to be processed.

[0106] Step S23: Determine the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the change value corresponding to the area difference between the ventricular area and the atrial area.

[0107] For more detailed processing procedures of steps S22 and S23, please refer to the corresponding content disclosed in the foregoing embodiments, which will not be repeated here.

[0108] As can be seen, in this embodiment, a chamber structure segmentation model for cardiac chamber structure segmentation is pre-constructed based on a preset convolutional neural network, and this model is used to process the cardiac ultrasound image to be processed to obtain the target mask for each cardiac chamber. Specifically, the preset convolutional neural network includes a CNN backbone network and a feature decoder. The CNN backbone network is used to extract low-level semantic features from the cardiac ultrasound image to be processed, and these low-level semantic features are fused and encoded in a high-level convolutional network to generate high-level semantic features. The feature decoder includes a first feature decoder and a second feature decoder. The first feature decoder is used to decode the high-level semantic features to obtain the location information and index information of each cardiac chamber. The location information is specifically the minimum bounding rectangle corresponding to each cardiac chamber. The second feature decoder is used to decode the high-level semantic features to obtain the three-dimensional mask of each cardiac chamber, and then the index information is used to filter out the corresponding two-dimensional mask from the three-dimensional mask. Two-dimensional masks exceeding the range of the minimum bounding rectangle are then removed to reduce errors and obtain the target mask for each cardiac chamber. In other words, this embodiment provides a method for determining cardiac cycle phases based on a CNN network model.

[0109] See Figure 5 As shown, this application discloses a specific method for determining the phase of the cardiac cycle. Compared to the previous embodiment, this embodiment further explains and optimizes the technical solution. Specifically, it includes:

[0110] Step S31: Determine the target mask for each cardiac chamber in the cardiac ultrasound image to be processed.

[0111] Step S32: Using the contour information of each heart chamber determined based on the target mask, determine the ventricular area and atrial area in the cardiac ultrasound image to be processed.

[0112] Step S33: Record the area difference between the ventricular area and the atrial area in a preset area difference change graph; the preset area difference change graph is used to record the area difference between the ventricular area and the atrial area in continuous different cardiac ultrasound images.

[0113] In this embodiment, after obtaining the area difference between the ventricular area and the atrial area, the area difference is recorded in a preset area difference change map. This preset area difference change map is used to record the area difference between the ventricular area and the atrial area in consecutive different echocardiogram images. It is understood that change maps can be plotted separately for the ventricular area and the atrial area in consecutive different echocardiogram images, and then the difference between the two can be used to obtain the preset area difference change map composed of the area difference between the ventricular area and the atrial area.

[0114] Step S34: Determine the gradient rate of change of the area difference corresponding to the cardiac ultrasound image to be processed in the current preset area difference change map.

[0115] In this embodiment, the area difference is recorded in a preset area difference change map, and the gradient change rate of the area difference corresponding to the cardiac ultrasound image to be processed in the current preset area difference change map is determined. That is, the gradient change rate of the area difference in the current preset area difference change map is determined.

[0116] Step S35: Determine the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the gradient change rate.

[0117] In this embodiment, the specific steps for determining the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the specific value of the gradient change rate can be found in [reference needed]. Figure 6 As shown, step S35 may specifically include:

[0118] Step S351: If the gradient change rate is positive, then the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be diastolic.

[0119] In other words, if the gradient rate of change is positive, it indicates that the area difference between the ventricle and the atrium is increasing. That is, the ventricular area is continuously increasing while the atrial area is continuously decreasing, thus causing the area difference to continuously increase. Therefore, when the gradient rate of change is positive, the cardiac cycle phase corresponding to the echocardiogram image being processed is determined to be diastole.

[0120] Step S352: If the gradient change rate is negative, then the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be the systolic phase.

[0121] In this embodiment, if the gradient change rate is negative, it indicates that the area difference between the ventricular and atrial areas is decreasing. That is, the ventricular area is continuously decreasing while the atrial area is continuously increasing, thus causing the area difference to continuously decrease. Therefore, when the gradient change rate is negative, the cardiac cycle phase corresponding to the cardiac ultrasound image being processed is determined to be the systolic phase.

[0122] Step S353: If the gradient change rate is zero, the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined using the frame number of the cardiac ultrasound image to be processed and the cardiac cycle length determined based on the preset area difference change map.

[0123] In this embodiment, if the gradient change rate is zero, it indicates that the current cardiac cycle phase is at its end. Therefore, when the gradient change rate is zero, there are two possibilities: 1. The ventricular area is at its maximum and the atrial area is at its minimum, indicating the end of diastole; 2. The ventricular area is at its minimum and the atrial area is at its maximum, indicating the end of systole. Therefore, it is not possible to directly determine that the cardiac cycle phase corresponding to the processed echocardiogram image is systole. Instead, it is necessary to further determine the cardiac cycle phase corresponding to the processed echocardiogram image using the frame number of the processed echocardiogram image and the cardiac cycle length determined based on a preset area difference change map.

[0124] In a specific implementation, the method of determining the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed using the frame number of the cardiac ultrasound image to be processed and the cardiac cycle length determined based on the preset area difference change map includes: performing periodic fitting on the preset area difference change map to determine the cardiac cycle length; obtaining the first target frame number corresponding to the previous extreme point adjacent to the position point corresponding to the cardiac ultrasound image to be processed in the preset area difference change map; predicting the second target frame number corresponding to the next extreme point adjacent to the previous extreme point in the preset area difference change map based on the cardiac cycle length and the first target frame number; determining the position type of the position point corresponding to the cardiac ultrasound image to be processed using the comparison result between the frame number of the cardiac ultrasound image to be processed and the second target frame number and the type of the previous extreme point; and determining the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the position type.

[0125] In other words, firstly, a periodic fitting is performed on the preset area difference change map to determine the cardiac cycle length. It's important to note that to ensure better periodic fitting and a more accurate cardiac cycle length, the preset area difference change map must include at least two cardiac cycles. Next, the first target frame number is obtained from the previous extreme point adjacent to the location point in the preset area difference change map corresponding to the cardiac ultrasound image to be processed. This previous extreme point corresponds to the end of systole or end of diastole in the cardiac cycle. Then, the cardiac cycle length and the first target frame number can be used to predict the second target frame number corresponding to the next extreme point adjacent to the previous extreme point in the preset area difference change map. The frame number of the cardiac ultrasound image to be processed is compared with the second target frame number to obtain a comparison result. Finally, the location type of the location point corresponding to the location point in the cardiac ultrasound image to be processed is determined by combining this comparison result with the type of the previous extreme point.

[0126] Furthermore, the method of determining the location type of the corresponding location point in the cardiac ultrasound image to be processed by comparing the frame number of the cardiac ultrasound image to be processed with the second target frame number and the type of the previous extreme point includes: if the difference between the frame number of the cardiac ultrasound image to be processed and the second target frame number is within a preset difference range, then the location type of the corresponding location point in the cardiac ultrasound image to be processed is determined according to the type of the previous extreme point; wherein, if the type of the previous extreme point is a maximum value, then the location type is a trough location; if the type of the previous extreme point is a minimum value, then the location type is a peak location. Understandably, when comparing the frame number of the cardiac ultrasound image to be processed with the second target frame number, the main focus is on determining whether the difference between the two falls within a preset difference range. If the difference is within the preset range, the position type of the corresponding point in the cardiac ultrasound image to be processed is determined based on the type of the previous extreme point. If the type of the previous extreme point is a maximum value, the position type of the corresponding point in the cardiac ultrasound image to be processed is a trough position; if the type of the previous extreme point is a minimum value, the position type of the corresponding point in the cardiac ultrasound image to be processed is a peak position. Thus, determining the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the position type includes: if the position type is a peak position, the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be diastolic; if the position type is a trough position, the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be systolic. In other words, if the location type of the corresponding point in the cardiac ultrasound image to be processed is a peak position, then the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be diastole; if the location type of the corresponding point in the cardiac ultrasound image to be processed is a trough position, then the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be systole.

[0127] For example, assuming the previous extreme point is a maximum, the corresponding first target frame number is 100, indicating that the echocardiogram image corresponding to the previous extreme point is in the end of diastole, and assuming the cardiac cycle length is 30 frames, then based on the first target frame number and the cardiac cycle length, the second target frame number corresponding to the next extreme point adjacent to the previous extreme point can be predicted to be 115 frames. This is a minimum, meaning that the echocardiogram image should be in the end of systole half a cycle later. The frame number of the echocardiogram image to be processed is then compared with the second target frame number to determine if the difference between them is within a preset difference range, for example, within 2 frames. If the frame number of the echocardiogram image to be processed is 114, then the difference between the two is 1, which is within the preset difference range. Therefore, the cardiac cycle phase corresponding to the echocardiogram image to be processed can be determined to be systole.

[0128] In another specific embodiment, the aforementioned previous extreme point may refer only to the previous maximum or minimum value. In this case, only the frame number of the maximum or minimum value needs to be recorded each time. For example, taking the previous maximum value as an example, the second target frame number corresponding to the next maximum value can be predicted by adding the cardiac cycle length to the first target frame number of the previous maximum value. If the frame number of the cardiac ultrasound image to be processed is close to the second target frame number, and the difference between the two is within a preset difference range, then the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be diastolic. If the frame number of the cardiac ultrasound image to be processed differs from the second target frame number by approximately half a cardiac cycle, then the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be systolic. Furthermore, in the above process, the second target frame number corresponding to the next minimum value can also be predicted by adding half a cardiac cycle length to the first target frame number. Subsequent judgment steps are similar to the aforementioned content and will not be repeated here.

[0129] For more detailed processing procedures regarding steps S31 and S32, please refer to the corresponding content disclosed in the foregoing embodiments, which will not be repeated here.

[0130] As can be seen, in this embodiment, after obtaining the area difference between the ventricular and atrial areas, the area difference is recorded in a preset area difference change map. This preset area difference change map records the area difference between the ventricular and atrial areas in consecutive different echocardiogram images and determines the gradient rate of change of the area difference corresponding to the echocardiogram image to be processed in the current preset area difference change map. If the gradient rate of change is positive, the cardiac cycle phase corresponding to the echocardiogram image to be processed is determined to be diastolic; if the gradient rate of change is negative, the cardiac cycle phase corresponding to the echocardiogram image to be processed is determined to be systolic; if the gradient rate of change is zero, the cardiac cycle phase corresponding to the echocardiogram image to be processed needs to be determined using the frame number of the echocardiogram image to be processed and the cardiac cycle length determined based on the preset area difference change map. In other words, this embodiment provides a method for determining the cardiac cycle phase corresponding to an echocardiogram image to be processed based on the gradient rate of change of the area difference in a preset area difference change map.

[0131] See Figure 7 As shown, this application discloses a specific method for determining the phase of the cardiac cycle. Compared to the previous embodiment, this embodiment further explains and optimizes the technical solution. Specifically, it includes:

[0132] Step S41: Determine the target mask for each cardiac chamber in the cardiac ultrasound image to be processed.

[0133] Step S42: Using the contour information of each heart chamber determined based on the target mask, determine the ventricular area and atrial area in the cardiac ultrasound image to be processed.

[0134] Step S43: Determine the difference between the area difference between the ventricular area and the atrial area and the area difference of the previous frame; the area difference of the previous frame is the area difference between the ventricular area and the atrial area in the previous frame of processed cardiac ultrasound image relative to the cardiac ultrasound image to be processed.

[0135] In this embodiment, after obtaining the area difference between the ventricular area and the atrial area, the difference between this area difference and the area difference of the previous frame is determined. The area difference of the previous frame refers to the area difference between the ventricular area and the atrial area in the previous frame of processed cardiac ultrasound image relative to the cardiac ultrasound image to be processed.

[0136] Step S44: Determine the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the difference.

[0137] In this embodiment, the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined based on the difference between the area difference of the current frame and the area difference of the previous frame. Further, determining the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the difference includes: if the difference is greater than zero, determining that the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is diastolic; if the difference is less than zero, determining that the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is systolic; if the difference is equal to zero, using the cardiac cycle phase of the previously processed cardiac ultrasound image as the cardiac cycle phase of the cardiac ultrasound image to be processed.

[0138] That is, in the first specific embodiment, if the difference between the area difference of the current frame and the area difference of the previous frame is greater than zero, it means that the area difference between the ventricular area and the atrial area is increasing compared to the previous frame. In other words, the ventricular area is continuously increasing and the atrial area is continuously decreasing. Therefore, the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be the diastolic phase.

[0139] In the second specific embodiment, if the difference between the area difference of the current frame and the area difference of the previous frame is less than zero, it means that the area difference between the ventricular area and the atrial area is decreasing compared to the previous frame. That is, the ventricular area is continuously decreasing and the atrial area is continuously increasing. Therefore, the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed is determined to be the systolic phase.

[0140] In the third specific embodiment, if the difference between the area difference of the current frame and the area difference of the previous frame is equal to zero, it indicates that the current frame is at the end of a certain cardiac cycle phase. In this case, the cardiac cycle phase of the previously processed echocardiogram image is taken as the cardiac cycle phase of the echocardiogram image to be processed. That is, if the cardiac cycle phase of the previously processed echocardiogram image is diastolic, then the cardiac cycle phase of the echocardiogram image to be processed is also diastolic; if the cardiac cycle phase of the previously processed echocardiogram image is systolic, then the cardiac cycle phase of the echocardiogram image to be processed is also systolic.

[0141] For more detailed processing of steps S41 and S42, please refer to the relevant content disclosed in the foregoing embodiments, which will not be repeated here.

[0142] As can be seen, in this embodiment, after obtaining the current frame area difference between the current ventricular area and the atrial area, the difference between the current frame area difference and the previous frame area difference is determined. The previous frame area difference refers to the area difference between the ventricular area and the atrial area in the previous processed frame of the echocardiogram image relative to the image to be processed. If the difference between the current frame area difference and the previous frame area difference is greater than zero, the cardiac cycle phase corresponding to the echocardiogram image to be processed is determined to be diastolic. If the difference between the current frame area difference and the previous frame area difference is less than zero, the cardiac cycle phase corresponding to the echocardiogram image to be processed is determined to be systolic. If the difference between the current frame area difference and the previous frame area difference is equal to zero, the cardiac cycle phase of the previous processed frame of the echocardiogram image is taken as the cardiac cycle phase of the echocardiogram image to be processed. That is, this embodiment provides a method for determining the cardiac cycle phase corresponding to the echocardiogram image to be processed based on the difference between the current ventricular area and the atrial area and the previous frame area difference.

[0143] See Figure 8 As shown in the figure, this application discloses a cardiac cycle phase determination device, which includes:

[0144] The mask determination module 11 is used to determine the target mask for each cardiac chamber in the cardiac ultrasound image to be processed.

[0145] The area determination module 12 is used to determine the ventricular area and atrial area in the cardiac ultrasound image to be processed by using the contour information of each of the heart chambers determined based on the target mask.

[0146] The cycle phase determination module 13 is used to determine the cardiac cycle phase corresponding to the cardiac ultrasound image to be processed based on the change value corresponding to the area difference between the ventricular area and the atrial area.

[0147] Since the embodiments of the device part correspond to the embodiments described above, please refer to the embodiments described in the method part for the embodiments of the device part, and will not be repeated here.

[0148] As can be seen, this application determines the target mask of each heart chamber in the echocardiogram image to be processed; uses the contour information of each heart chamber determined based on the target mask to determine the ventricular area and atrial area in the echocardiogram image to be processed; and determines the cardiac cycle phase corresponding to the echocardiogram image to be processed based on the change value corresponding to the area difference between the ventricular area and the atrial area. Therefore, this application first needs to determine the target mask of each heart chamber in the echocardiogram image to be processed, and then use the target mask to determine the contour information of each heart chamber. Further, the ventricular area and atrial area are calculated based on the contour information, and finally, the cardiac cycle phase of the echocardiogram image to be processed is determined based on the change value corresponding to the area difference between the ventricular area and the atrial area. The echocardiogram image to be processed can be a real-time acquired current frame echocardiogram image or a historical echocardiogram image. If it is a current frame echocardiogram image, the cardiac cycle phase of the current frame echocardiogram image can be determined in real time. In addition, this application determines the cardiac cycle phase of the cardiac ultrasound image to be processed by automatically calculating the difference in ventricular and atrial area. The entire process does not require the intervention of a doctor, thus making it more intelligent.

[0149] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Specifically, it may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the cardiac cycle phase determination method disclosed in any of the foregoing embodiments.

[0150] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0151] The processor 21 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 21 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 21 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 21 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 21 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0152] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored on it include operating system 221, computer program 222 and data 223, etc., and the storage method can be temporary storage or permanent storage.

[0153] The operating system 221 manages and controls the various hardware devices and computer programs 222 on the electronic device 20 to enable the processor 21 to perform calculations and processing on the massive amounts of data 223 in the memory 22. The operating system 221 can be Windows, Unix, Linux, etc. The computer program 222, in addition to including a computer program capable of performing the cardiac cycle timing determination method disclosed in any of the foregoing embodiments, may further include computer programs capable of performing other specific tasks. The data 223 may include data received by the electronic device from external devices, as well as data collected by its own input / output interface 25.

[0154] Furthermore, embodiments of this application also disclose a computer-readable storage medium storing a computer program. When the computer program is loaded and executed by a processor, it implements the steps of the cardiac cycle phase determination method disclosed in any of the foregoing embodiments.

[0155] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0156] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0157] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0158] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0159] The foregoing has provided a detailed description of the method, apparatus, device, and storage medium for determining cardiac cycle phases provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method of cardiac cycle phase determination, characterized by, The method comprises the following steps: determining a target mask of each cardiac chamber in a to-be-processed cardiac ultrasound image; determining a ventricular area and an atrial area in the to-be-processed cardiac ultrasound image by using contour information of each cardiac chamber determined based on the target mask; wherein the ventricular area is the sum of a left ventricular area and a right ventricular area; and the atrial area is the sum of a left atrial area and a right atrial area; determining a cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image based on a change value corresponding to an area difference between the ventricular area and the atrial area; wherein the determination of the cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image based on the change value corresponding to the area difference between the ventricular area and the atrial area comprises: recording the area difference between the ventricular area and the atrial area into a preset area difference change graph; the preset area difference change graph is used to record the area difference between the ventricular area and the atrial area in continuous different cardiac ultrasound images; determining a gradient change rate of the area difference corresponding to the to-be-processed cardiac ultrasound image in the current preset area difference change graph; determining the cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image based on the gradient change rate.

2. The cardiac cycle phase determination method of claim 1, wherein, The determination of the target mask of each cardiac chamber in the to-be-processed cardiac ultrasound image comprises: processing the to-be-processed cardiac ultrasound image by using a chamber structure segmentation model constructed in advance based on a preset convolutional neural network to obtain the target mask of each cardiac chamber.

3. The cardiac cycle phase determination method of claim 2, wherein, The preset convolutional neural network comprises a CNN backbone network and a feature decoder. Correspondingly, the processing of the to-be-processed cardiac ultrasound image by using the chamber structure segmentation model constructed in advance based on the preset convolutional neural network to obtain the target mask of each cardiac chamber comprises: extracting high-level semantic features in the to-be-processed cardiac ultrasound image by using the CNN backbone network in the chamber structure segmentation model; decoding the high-level semantic features by using the feature decoder to obtain the target mask of each cardiac chamber.

4. The cardiac cycle phase determination method of claim 3, wherein, The extraction of the high-level semantic features in the to-be-processed cardiac ultrasound image by using the CNN backbone network in the chamber structure segmentation model comprises: extracting low-level semantic features in the to-be-processed cardiac ultrasound image by using the CNN backbone network in the chamber structure segmentation model; the low-level semantic features comprise edge feature information, contour feature information and grayscale feature information; performing feature fusion and coding processing on the low-level semantic features to obtain high-level semantic features.

5. The cardiac cycle phase determination method of claim 3, wherein, The feature decoder comprises a first feature decoder and a second feature decoder. Correspondingly, the decoding of the high-level semantic features by using the feature decoder to obtain the target mask of each cardiac chamber comprises: decoding the high-level semantic features by using the first feature decoder to obtain position information and index information of each cardiac chamber; decoding the high-level semantic features by using the second feature decoder to obtain a three-dimensional mask of each cardiac chamber; screening a two-dimensional mask corresponding to the index information from the three-dimensional mask, and correcting the two-dimensional mask based on the position information to obtain the target mask of each cardiac chamber.

6. The cardiac cycle phase determination method of claim 5, wherein, The position information is a minimum circumscribed rectangular frame corresponding to each of the cardiac chambers; Correspondingly, the position information is a minimum circumscribed rectangular frame corresponding to each of the cardiac chambers. The method further includes:

7. The cardiac cycle phase determination method of claim 2, wherein, The training set corresponding to the chamber structure segmentation model comprises training images obtained by performing enhancement processing on historical cardiac ultrasound images and corresponding training labels; the training labels comprise boundary ranges of each cardiac chamber structure in the training images.

8. The cardiac cycle phase determination method of claim 1, wherein, The method further includes: The computer vision algorithm is used to convert the target mask into initial contour information of each of the cardiac chambers. The preset post-processing algorithm comprises any one or several of an image denoising algorithm, an image smoothing algorithm, and an image edge cutting algorithm.

9. The cardiac cycle phase determination method of claim 8, wherein, The method further includes:

10. The cardiac cycle phase determination method of claim 1, wherein, If the gradient change rate is positive, it is determined that the cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image is a diastolic phase. If the gradient change rate is negative, it is determined that the cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image is a systolic phase. If the gradient change rate is zero, the frame number of the to-be-processed cardiac ultrasound image and the cardiac cycle length determined based on the preset area difference change graph are used to determine the cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image. The method further includes:

11. The cardiac cycle phase determination method of claim 10, wherein, Period fitting is performed on the preset area difference change graph to determine a cardiac cycle length. A first target frame number corresponding to a last extreme value point adjacent to a position point corresponding to the to-be-processed cardiac ultrasound image in the preset area difference change graph is obtained. A second target frame number corresponding to a next extreme value point adjacent to the last extreme value point is predicted based on the cardiac cycle length and the first target frame number. A position type of the position point corresponding to the to-be-processed cardiac ultrasound image is determined based on a comparison result between the frame number of the to-be-processed cardiac ultrasound image and the second target frame number and a type of the last extreme value point. The position type is used to determine the cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image. The position type of the position point corresponding to the to-be-processed cardiac ultrasound image is determined based on a comparison result between the frame number of the to-be-processed cardiac ultrasound image and the second target frame number and a type of the last extreme value point.

12. The cardiac cycle phase determination method of claim 11, wherein, ​ If a difference between the frame number of the to-be-processed heart ultrasound image and the second target frame number is within a preset difference range, a position type of a position point corresponding to the to-be-processed heart ultrasound image is determined according to a type of the last extreme point; wherein if the type of the last extreme point is a maximum value, the position type is a wave trough position; if the type of the last extreme point is a minimum value, the position type is a wave peak position.

13. The cardiac cycle phase determination method of claim 11 or 12, characterized by, The determining of the cardiac cycle phase corresponding to the to-be-processed heart ultrasound image based on the position type comprises: If the position type is the wave peak position, it is determined that the cardiac cycle phase corresponding to the to-be-processed heart ultrasound image is a diastolic phase; If the position type is the wave trough position, it is determined that the cardiac cycle phase corresponding to the to-be-processed heart ultrasound image is a systolic phase.

14. A method for determining the phase of the cardiac cycle, characterized in that, It comprises: determining target masks of each heart chamber in a to-be-processed heart ultrasound image; determining ventricular area and atrial area in the to-be-processed heart ultrasound image by using contour information of each heart chamber determined based on the target masks; wherein the ventricular area is the sum of left ventricular area and right ventricular area; the atrial area is the sum of left atrial area and right atrial area; determining the cardiac cycle phase corresponding to the to-be-processed heart ultrasound image based on a change value corresponding to an area difference between the ventricular area and the atrial area; The determining of the cardiac cycle phase corresponding to the to-be-processed heart ultrasound image based on the change value corresponding to the area difference between the ventricular area and the atrial area comprises: determining a difference value between the area difference between the ventricular area and the atrial area and a last frame area difference; the last frame area difference is an area difference between a ventricular area and an atrial area in a last frame processed heart ultrasound image relative to the to-be-processed heart ultrasound image; determining the cardiac cycle phase corresponding to the to-be-processed heart ultrasound image based on the difference value.

15. The cardiac cycle phase determination method of claim 14, wherein, The determining of the target masks of each heart chamber in the to-be-processed heart ultrasound image comprises: processing the to-be-processed heart ultrasound image by using a chamber structure segmentation model constructed in advance based on a preset convolutional neural network to obtain the target masks of each heart chamber.

16. The cardiac cycle phase determination method of claim 15, wherein, The preset convolutional neural network comprises a CNN backbone network and a feature decoder; Correspondingly, the processing of the to-be-processed heart ultrasound image by using the chamber structure segmentation model constructed in advance based on the preset convolutional neural network to obtain the target masks of each heart chamber comprises: extracting high-level semantic features in the to-be-processed heart ultrasound image by using the CNN backbone network in the chamber structure segmentation model; decoding the high-level semantic features by using the feature decoder to obtain the target masks of each heart chamber.

17. The cardiac cycle phase determination method of claim 16, wherein, The extracting of the high-level semantic features in the to-be-processed heart ultrasound image by using the CNN backbone network in the chamber structure segmentation model comprises: extracting low-level semantic features in the to-be-processed heart ultrasound image by using the CNN backbone network in the chamber structure segmentation model; the low-level semantic features comprise edge feature information, contour feature information and gray scale feature information; The bottom layer semantic features are fused and encoded to obtain high layer semantic features.

18. The cardiac cycle phase determination method of claim 16, wherein, The feature decoder comprises a first feature decoder and a second feature decoder. Correspondingly, the decoding of the high layer semantic features by the feature decoder to obtain the target mask of each cardiac chamber comprises: Decoding the high layer semantic features by the first feature decoder to obtain the position information and index information of each cardiac chamber; Decoding the high layer semantic features by the second feature decoder to obtain the three-dimensional mask of each cardiac chamber; Filtering the two-dimensional mask corresponding to the index information from the three-dimensional mask, and correcting the two-dimensional mask based on the position information to obtain the target mask of each cardiac chamber.

19. The cardiac cycle phase determination method of claim 18, wherein, The position information is the minimum circumscribed rectangle corresponding to each cardiac chamber; Correspondingly, the correction of the two-dimensional mask based on the position information to obtain the target mask of each cardiac chamber comprises: Clipping the two-dimensional mask by the minimum circumscribed rectangle to take the mask inside the minimum circumscribed rectangle in the two-dimensional mask as the target mask.

20. The cardiac cycle phase determination method of claim 15, wherein, The training set corresponding to the chamber structure segmentation model comprises training images obtained by enhancing historical cardiac ultrasound images and corresponding training labels; the training labels comprise the boundary range of each cardiac chamber structure in the training images.

21. The cardiac cycle phase determination method of claim 14, wherein, Before determining the ventricular area and atrial area in the to-be-processed cardiac ultrasound image, the method further comprises: Converting the target mask into initial contour information of each cardiac chamber by using a computer vision algorithm; Post-processing the initial contour information by using a preset post-processing algorithm to obtain contour information of each cardiac chamber.

22. The cardiac cycle phase determination method of claim 21, wherein, The preset post-processing algorithm comprises any one or several of an image denoising algorithm, an image smoothing algorithm and an image edge cutting algorithm.

23. The cardiac cycle phase determination method of claim 14, wherein, The determination of the cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image based on the difference value comprises: If the difference value is greater than zero, it is determined that the cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image is a diastolic phase; If the difference value is less than zero, it is determined that the cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image is a systolic phase; If the difference value is equal to zero, the cardiac cycle phase of the previous frame of processed cardiac ultrasound image is taken as the cardiac cycle phase of the to-be-processed cardiac ultrasound image.

24. A device for determining the phase of a cardiac cycle, characterized in that, Comprise: A mask determination module is configured to determine a target mask of each cardiac chamber in a to-be-processed cardiac ultrasound image; An area determination module is configured to determine a ventricular area and an atrial area in the to-be-processed cardiac ultrasound image by using contour information of each cardiac chamber determined based on the target mask; wherein the ventricular area is the sum of a left ventricular area and a right ventricular area; and the atrial area is the sum of a left atrial area and a right atrial area; A cycle phase determination module is configured to determine a cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image based on a change value corresponding to an area difference between the ventricular area and the atrial area. The cycle phase determination module is specifically configured to record the area difference between the ventricular area and the atrial area into a preset area difference change graph; the preset area difference change graph is used to record the area difference between the ventricular area and the atrial area in continuous different cardiac ultrasound images; determine the gradient change rate of the area difference corresponding to the to-be-processed cardiac ultrasound image in the preset area difference change graph; and determine the cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image based on the gradient change rate.

25. A device for determining the phase of a cardiac cycle, characterized in that, Comprise: A mask determination module is configured to determine a target mask of each cardiac chamber in a to-be-processed cardiac ultrasound image; An area determination module is configured to determine a ventricular area and an atrial area in the to-be-processed cardiac ultrasound image by using the contour information of each cardiac chamber determined based on the target mask; wherein the ventricular area is the sum of a left ventricular area and a right ventricular area; and the atrial area is the sum of a left atrial area and a right atrial area; A cycle phase determination module is configured to determine a cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image based on a change value corresponding to the area difference between the ventricular area and the atrial area; The cycle phase determination module is specifically configured to determine a difference value between the area difference between the ventricular area and the atrial area and a previous frame area difference; the previous frame area difference is the area difference between the ventricular area and the atrial area in a previous frame processed cardiac ultrasound image relative to the to-be-processed cardiac ultrasound image; and determine the cardiac cycle phase corresponding to the to-be-processed cardiac ultrasound image based on the difference value.

26. An electronic device, comprising: Comprise: A memory is configured to save a computer program; A processor is configured to execute the computer program to implement the steps of the cardiac cycle phase determination method according to any one of claims 1 to 23.

27. A computer-readable storage medium, characterized in that, A computer program is configured to be executed by a processor to implement the steps of the cardiac cycle phase determination method according to any one of claims 1 to 23.

Images