Method of assisting in assessing the state of cardiac motion

Abstract

The present application provides a method for assisting in evaluating a heart motion state, comprising: obtaining a plurality of continuous heart ultrasound images corresponding to a heart, and estimating a plurality of left ventricular volumes corresponding to the plurality of heart ultrasound images; finding a plurality of specific extreme values in the plurality of left ventricular volumes, and estimating a plurality of time differences between the plurality of specific extreme values; estimating a statistical property value of the plurality of time differences based on the plurality of time differences; and determining that the heart has an abnormal motion state in response to determining that at least one of the plurality of time differences deviates from the statistical property value by a preset threshold value. Thus, relevant medical personnel can more easily grasp the condition of the heart, thereby reducing the probability of making a false evaluation (e.g., calculating a false ejection fraction).

Description

Technical Field

[0001] This invention relates to a medical image assessment method, and more particularly to a method for assisting in the assessment of cardiac motion status. Background Technology

[0002] Currently, the most widely used and economical method for examining heart structure is echocardiography. When a patient needs an ultrasound examination of their heart, the ultrasound technician or doctor holds the ultrasound probe and scans the heart at a specific angle to check for any abnormalities in the patient's heart structure.

[0003] Currently, there are many methods for assessing cardiac function in clinical practice, one of which is measuring the left ventricular ejection fraction (LVEF). LVEF measures how much blood is ejected from the left ventricle with each heartbeat. In hospitals, manual or semi-automatic methods are still used to identify end-diastolic (ED) and end-systolic (ES) phases of the heart cycle. After identifying ED and ES, the built-in tools on the ultrasound machine can calculate the end-diastolic volume (EDV) and end-systolic volume (ESV), and then calculate the ejection fraction (EF).

[0004] In the current technological field, there is a growing trend towards using AI deep learning models to identify the end-diastole and end-systole of the heart during the heart rate cycle, reducing the tedious work of manually identifying whether each image is an ED or ES.

[0005] However, in actual clinical practice, there are two main reasons why AI deep learning models have not yet replaced human identification. First, when patients have cardiac conditions, such as atrial fibrillation causing arrhythmia, AI deep learning models may be unable to effectively identify ED and ES, leading to incorrect ejection fraction calculations. Second, if the echocardiogram images obtained from the ultrasound probe are not clear enough, the AI ​​deep learning model will make poor judgments, failing to effectively identify ED and ES, resulting in incorrect ejection fraction calculations.

[0006] Therefore, for those skilled in the art, if a mechanism can be designed to determine whether a patient's heart has an abnormal motion state (such as arrhythmia), it should help the aforementioned AI deep learning model to more effectively identify the end-diastole and end-contraction of the heart in the heart rate cycle, thereby obtaining a more accurate ejection fraction. Summary of the Invention

[0007] In view of this, the present invention provides a method for assisting in the assessment of cardiac motion status, which can be used to solve the above-mentioned technical problems.

[0008] This invention provides a method for assisting in assessing cardiac motion status, suitable for an electronic device, comprising: acquiring a plurality of consecutive left ventricular shielded images corresponding to a heart, and estimating a plurality of left ventricular volumes corresponding to the plurality of cardiac ultrasound images; identifying a plurality of specific extreme values ​​among the plurality of left ventricular volumes, and estimating a plurality of time differences between the plurality of specific extreme values; estimating a statistical characteristic value of the plurality of time differences based on the plurality of time differences; and determining that the heart has an abnormal motion state in response to determining that at least one of the plurality of time differences deviates from the statistical characteristic value to a preset threshold value. Attached Figure Description

[0009] The accompanying drawings are included to further illustrate the invention, and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention.

[0010] Figure 1 This is a schematic diagram of an electronic device illustrated according to an embodiment of the present invention;

[0011] Figure 2 This is a flowchart illustrating a method for estimating ventricular volume according to an embodiment of the present invention;

[0012] Figure 3 This is an application scenario diagram illustrating an embodiment of the present invention;

[0013] Figure 4 This is a schematic diagram illustrating the identification of first, second, and third reference point pixels based on the distance between reference point pixels according to an embodiment of the present invention;

[0014] Figure 5A It is based on Figure 3 The illustration shows how to find the reference point pixel corresponding to the apex of the heart.

[0015] Figure 5B It is based on Figure 5A The diagram illustrates how to find the reference point pixel corresponding to the left mitral valve.

[0016] Figure 5C It is based on Figure 5B The diagram illustrates how to find the reference point pixel corresponding to the right mitral valve.

[0017] Figure 6 This is a flowchart illustrating a method for assisting in the assessment of cardiac motion status according to an embodiment of the present invention;

[0018] Figure 7A This is an application scenario diagram illustrating an embodiment of the present invention;

[0019] Figure 7B It is based on Figure 7A The illustration shows another application scenario. Detailed Implementation

[0020] Reference will now be made in detail to exemplary embodiments of the invention, examples of which are illustrated in the accompanying drawings. Wherever possible, the same component reference numerals are used in the drawings and description to denote the same or similar parts.

[0021] Please refer to Figure 1 This is a schematic diagram of an electronic device according to an embodiment of the present invention. In different embodiments, the electronic device 100 may be, for example, various computer devices, smart devices and / or handheld devices, but is not limited thereto.

[0022] like Figure 1 As shown, the electronic device 100 includes a storage circuit 102 and a processor 104. The storage circuit 102 is, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk, or other similar device or combination of these devices, and can be used to record multiple program codes or modules.

[0023] Processor 104 is coupled to storage circuit 102 and may be a general purpose processor, special purpose processor, conventional processor, digital signal processor, multiple microprocessors, one or more microprocessors incorporating a digital signal processor core, controller, microcontroller, application specific integrated circuit (ASIC), field programmable gate array (FPGA), any other type of integrated circuit, state machine, processor based on advanced RISC machine (ARM), and the like.

[0024] In an embodiment of the present invention, the processor 104 may access the modules and program code recorded in the storage circuit 102 to implement the method for estimating ventricular volume proposed in the present invention, the details of which are described below.

[0025] Please refer to Figure 2 and Figure 3 ,in Figure 2 This is a flowchart illustrating a method for estimating ventricular volume according to an embodiment of the present invention. Figure 3This is an application scenario diagram illustrating an embodiment of the present invention. The method of this embodiment can be derived from... Figure 1 The electronic device 100 performs the following: Figure 1 The components shown and Figure 3 Context description Figure 2 Details of each step.

[0026] First, in step S210, the processor 104 can obtain a left ventricular mask image 31 corresponding to the cardiac ultrasound image 30, wherein the left ventricular mask image 31 is a binarized image. In an embodiment of the invention, the processor 104 can, for example, input the cardiac ultrasound image 30 (which is, for example, an apical view of A2C (apical two chamber) or A4C ​​(apical four chamber)) into a pre-trained machine learning model, so that the machine learning model outputs a corresponding binarized image as the left ventricular mask image 31 in response to the cardiac ultrasound image 30.

[0027] In one embodiment, to enable the machine learning model to possess the aforementioned capabilities, the designer can input various cardiac ultrasound images labeled with ventricular regions as training data into the machine learning model during training. This allows the machine learning model to learn the features corresponding to the ventricular regions, and thus, when an unknown cardiac ultrasound image is obtained, it can accordingly identify the ventricular regions within it. Subsequently, the machine learning model can set all pixels in the ventricular regions to a first value (e.g., 255) and all pixels in non-ventricular regions to a second value (e.g., 0) to generate a corresponding binarized image, but this is not limited to this method.

[0028] Subsequently, in step S220, the processor 104 can find three reference point pixels 311 to 313 in the left ventricular mask image 31.

[0029] In embodiments of the present invention, each reference point pixel 311-313 may have a first value (e.g., 255). In addition, each reference point pixel 311-313 may be surrounded by N (e.g., 8) surrounding pixels, and the aforementioned surrounding pixels of each reference point pixel 311-313 may include N1 (e.g., 3) first surrounding pixels with the first value and N2 (e.g., 5) second surrounding pixels with the second value (e.g., 0), where N, N1, and N2 are positive integers.

[0030] In one embodiment, among the pixels surrounding the reference point pixel 311, the first surrounding pixels (i.e., pixels located within range 311a) are arranged in a straight line, and the second surrounding pixels (i.e., pixels located within range 311b) are arranged in a C-shape.

[0031] In one embodiment, among the pixels surrounding the reference point pixel 312, the first surrounding pixels (i.e., pixels located within range 312a) are arranged in an L-shape, and the second surrounding pixels (i.e., pixels located within range 312b) are arranged in an L-shape.

[0032] In one embodiment, among the pixels surrounding the reference point pixel 313, the first surrounding pixels (i.e., pixels located within range 313a) are arranged in an L-shape, and the second surrounding pixels (i.e., pixels located within range 313b) are arranged in an L-shape.

[0033] In an embodiment of the present invention, since each of the reference point pixels 311 to 313 is unique in the left ventricular mask image 31, the processor 104 can examine each pixel in the left ventricular mask image 31 one by one, and define the three pixels that meet the above conditions (for example, including three first surrounding pixels with a first value and five second surrounding pixels with a second value among eight surrounding pixels, and having the first value) as reference point pixels 311 to 313.

[0034] Subsequently, in step S230, the processor 104 can estimate the left ventricular volume corresponding to the cardiac ultrasound image 30 based on reference point pixels 311-313. In one embodiment, the processor 104 can estimate the distance between the reference point pixels 311-313 and thereby identify a first reference point pixel, a second reference point pixel, and a third reference point pixel among the reference point pixels 311-313, respectively corresponding to the apex of the heart, the first mitral valve (e.g., the left mitral valve), and the second mitral valve (e.g., the right mitral valve). The processor 104 can then apply Simpson's formula based on the first reference point pixel, the second reference point pixel, and the third reference point pixel to estimate the left ventricular volume corresponding to the cardiac ultrasound image 30.

[0035] Please refer to Figure 4 This is a schematic diagram illustrating the identification of first, second, and third reference point pixels based on the distance between reference point pixels according to an embodiment of the present invention. Generally, the distance between the left and right mitral valves should be smaller than the distance between the apex of the heart and either mitral valve. Therefore, the processor 104 can identify the first, second, and third reference point pixels corresponding to the apex of the heart, the first mitral valve, and the second mitral valve, respectively, based on this principle in reference point pixels 311-313.

[0036] exist Figure 4 In the above, there may be a first distance D1 between reference point pixel 311 and reference point pixel 312, a second distance D2 between reference point pixel 311 and reference point pixel 313, and a third distance D3 between reference point pixel 312 and reference point pixel 313.

[0037] exist Figure 4 In this scenario, in response to the determination that both the first distance D1 and the second distance D2 are greater than the third distance D3, the processor can define reference point pixels 311 to 313 as the first, second, and third reference point pixels, respectively.

[0038] In another embodiment, in response to the determination that both the second distance D2 and the third distance D3 are greater than the first distance D1, the processor 104 may define reference pixel 313, reference pixel 311, and reference pixel 312 as the first, second, and third reference pixel, respectively. Furthermore, in yet another embodiment, in response to the determination that both the first distance D1 and the third distance D3 are greater than the second distance D2, the processor 103 may define reference pixel 312, reference pixel 311, and reference pixel 313 as the first, second, and third reference pixel, respectively.

[0039] Furthermore, if the cardiac ultrasound image 30 is determined to be an apical view, then the pixel with the highest height among the three identified reference points should correspond to the apex of the heart. Therefore, in Figure 4 In this process, the processor 104 can directly define the reference point pixel 311 with the highest height as the first reference point pixel corresponding to the apex of the heart, and define the remaining reference point pixels 312 and 313 as the second and third reference point pixels corresponding to the mitral valve, respectively, but it is not limited to this.

[0040] Subsequently, the processor 104 may apply Simpson's formula based on the first, second, and third reference point pixels to estimate the left ventricular volume corresponding to the cardiac ultrasound image 30, the details of which can be found in relevant prior art literature and will not be repeated here.

[0041] In addition, to improve the efficiency of finding reference point pixels 311-313, the processor 104 may be based on... Figures 5A to 5C The mechanism shown is used to find reference point pixels 311 to 313.

[0042] Please refer to Figure 5A It is based on Figure 3 The illustration shows a schematic diagram of finding the reference point pixel corresponding to the apex of the heart. As mentioned above, if the cardiac ultrasound image 30 is determined to be an apical view, then the highest of the three identified reference point pixels should correspond to the apex of the heart.

[0043] Therefore, starting from the highest pixel column in the left ventricular mask image 31, the processor 104 can scan down column by column to find pixels that meet the above conditions (e.g., including 3 first surrounding pixels with a first value and 5 second surrounding pixels with a second value among 8 surrounding pixels, and having the first value). Figure 5AWhen a pixel that meets the above conditions is found, the processor 104 can directly define this pixel as the reference point pixel 311 corresponding to the apex of the heart and pause the scanning.

[0044] Please refer to Figure 5B It is based on Figure 5A The illustration shows how to find the reference point pixel corresponding to the left mitral valve. In the case where the cardiac ultrasound image 30 is determined to be an apical view, the leftmost of the three identified reference point pixels should correspond to the left mitral valve.

[0045] Therefore, starting from the lowest pixel column in the left ventricular mask image 31, the processor 104 can scan column by column from left to right to find pixels that meet the above conditions (e.g., including 3 first surrounding pixels with a first value and 5 second surrounding pixels with a second value among 8 surrounding pixels, and having the first value). Figure 5B When a pixel that meets the above conditions is found, the processor 104 can directly define this pixel as the reference point pixel 312 corresponding to the left mitral lobe and pause the scanning process.

[0046] Please refer to Figure 5C It is based on Figure 5B The illustration shows how to find the reference point pixel corresponding to the right mitral valve. In the case where the cardiac ultrasound image 30 is determined to be an apical view, the lower rightmost of the three identified reference point pixels should correspond to the right mitral valve.

[0047] Therefore, starting from the lowest pixel column in the left ventricular mask image 31, the processor 104 can scan column by column from right to left to find pixels that meet the above conditions (e.g., including 3 first surrounding pixels with a first value and 5 second surrounding pixels with a second value among 8 surrounding pixels, and having the first value). Figure 5C When a pixel that meets the above conditions is found, the processor 104 can first determine whether this pixel has already been defined as another reference point pixel (e.g., reference point pixel 312). If not, the processor 104 can directly define this pixel as reference point pixel 313 corresponding to the right mitral lobe and pause the scanning process. On the other hand, if this pixel has already been defined as another reference point pixel (e.g., reference point pixel 312), the processor 104 can ignore this pixel and continue scanning upwards to find another pixel that meets the above conditions. When another pixel that meets the above conditions is found, the processor 104 can directly define the other pixel as reference point pixel 313 corresponding to the right mitral lobe and pause the scanning process.

[0048] In other embodiments, the processor 104 may also use other methods to find the reference point pixels 311-313 in the left ventricular mask image 31, and is not limited to these methods. Figures 5A to 5C Teaching methods.

[0049] As described above, the method for estimating ventricular volume proposed in this invention, after obtaining a left ventricular mask image corresponding to a left ventricular ultrasound, identifies three pixels that meet specific conditions (e.g., three first surrounding pixels with a first value and five second surrounding pixels with a second value, and also possessing the first value) as reference point pixels corresponding to the apex and both mitral valves. Then, the left ventricular volume can be estimated based on these multiple reference point pixels. Therefore, this invention can automatically estimate left ventricular volume with better efficiency without manually marking the apex and both mitral valves.

[0050] In other embodiments, the present invention provides a method for assisting in the assessment of cardiac motion status, which can determine whether the heart is experiencing abnormal motion based on changes in left ventricular volume. In embodiments of the present invention, the processor 104 can access modules and program code recorded in the storage circuit 102 to implement the method for assisting in the assessment of cardiac motion status proposed in the present invention, the details of which are described below.

[0051] Please refer to Figure 6 This is a flowchart illustrating a method for assisting in the assessment of cardiac motion status according to an embodiment of the present invention. The method of this embodiment can be derived from... Figure 1 The electronic device 100 performs the following: Figure 1 Component description shown Figure 6 Details of each step.

[0052] First, in step S610, the processor 104 may acquire a plurality of consecutive cardiac ultrasound images corresponding to the heart (which is, for example, the heart of a patient) and estimate a plurality of left ventricular volumes corresponding to the plurality of cardiac ultrasound images.

[0053] In one embodiment, the processor 104 may first acquire the aforementioned cardiac ultrasound images and determine whether each cardiac ultrasound image belongs to an apical view (e.g., A2C or A4C). In one embodiment, the processor 104 may, for example, determine whether each cardiac ultrasound image belongs to an apical view based on the techniques described in the document "Guidelines for Performing a Comprehensive Transthoracic Echocardiographic Examination in Adults: Recommendations from the American Society of Echocardiography". Therefore, relevant details can be found in the aforementioned document and will not be repeated here.

[0054] In response to determining that each cardiac ultrasound image belongs to the apical view, the processor 104 can extract the left ventricular mask image corresponding to the left ventricle of the heart from each cardiac ultrasound image and estimate the left ventricular volume corresponding to the cardiac ultrasound image.

[0055] In one embodiment, the processor 104 may, for example, input each cardiac ultrasound image into the previously mentioned machine learning model, wherein the machine learning model may output a corresponding left ventricular mask image in response to each cardiac ultrasound image.

[0056] In embodiments of the present invention, for each left ventricular mask image, the processor 104 can be based on Figures 2 to 5C The mechanism taught is used to estimate the corresponding left ventricular volume, so its details will not be elaborated here.

[0057] To facilitate the explanation of the concept of this invention, the following will provide further details. Figure 7A To further explain, among which Figure 7A This is an application scenario diagram illustrating an embodiment of the present invention. Figure 7A In this process, the consecutive left ventricular volumes obtained from step S610 can be illustrated as follows: Figure 7A The diagram showing the change in left ventricular volume is shown in Figure 700.

[0058] Subsequently, in step S620, the processor 104 can identify multiple specific extreme values ​​711 to 715 in the multiple left ventricular volumes and estimate multiple time differences T1 to T4 between the multiple specific extreme values ​​711 to 715.

[0059] In one embodiment, the processor 104 may, for example, use multiple specific left ventricular volumes corresponding to the EDV as the multiple specific extreme values, but is not limited to this. By definition, each EDV should correspond to the largest left ventricular volume in its respective heart rate cycle. Based on this, if the processor 104 determines that the i-th (where i is an integer) left ventricular volume is greater than the (i-1)-th and (i+1)-th left ventricular volumes, then the processor 104 can determine that the i-th ventricular volume should correspond to the EDV, and further determine that the i-th left ventricular volume belongs to one of the aforementioned specific extreme values.

[0060] exist Figure 7A In this scenario, since the left ventricular volume change graph 700 can be understood as including 5 heart rate cycles, the processor 104 can identify 5 EDVs as specific extreme values ​​711 to 715 according to the above principle. Then, the processor 104 can estimate the time differences T1 to T4 between the specific extreme values ​​711 to 715.

[0061] In general, assuming that the specific extreme values ​​found by the processor 104 include the first specific extreme value to the Kth (where K is an integer) specific extreme values, the time difference between the (j+1)th specific extreme value and the jth specific extreme value can be defined as the jth time difference, where 1≤j≤K-1.

[0062] by Figure 7A For example, time difference T1 (which can be understood as the first time difference) is, for example, the time difference between a specific extreme value 711 (which can be understood as the first specific extreme value) and a specific extreme value 712 (which can be understood as the second specific extreme value). Time difference T2 (which can be understood as the second time difference) is, for example, the time difference between a specific extreme value 712 (which can be understood as the second specific extreme value) and a specific extreme value 713 (which can be understood as the third specific extreme value). Time difference T3 (which can be understood as the third time difference) is, for example, the time difference between a specific extreme value 713 (which can be understood as the third specific extreme value) and a specific extreme value 714 (which can be understood as the fourth specific extreme value). Time difference T4 (which can be understood as the fourth time difference) is, for example, the time difference between a specific extreme value 714 (which can be understood as the fourth specific extreme value) and a specific extreme value 715 (which can be understood as the fifth specific extreme value), but it is not limited to these.

[0063] Subsequently, in step S630, the processor 104 can estimate the statistical characteristic values ​​(including but not limited to the average value of time differences T1 to T4) of the plurality of time differences T1 to T4 based on the plurality of time differences T1 to T4. Furthermore, the processor 104 can determine whether each time difference T1 to T4 deviates from the aforementioned statistical characteristic to a preset threshold value. In different embodiments, the aforementioned preset threshold value can be set to any percentage value according to the designer's needs, such as 5%, but is not limited to this.

[0064] In step S640, in response to determining that at least one of the plurality of time differences T1 to T4 deviates from the statistical characteristic value to a preset threshold, the processor 104 can determine that the heart has an abnormal motion state (e.g., arrhythmia). On the other hand, in response to determining that none of the plurality of time differences T1 to T4 deviates from the statistical characteristic value to the preset threshold, the processor 104 can determine that the heart has not experienced an abnormal motion state.

[0065] exist Figure 7A If the processor 104 determines that the time difference T1 to T4 does not deviate from the statistical characteristic value to the preset threshold, then the processor 104 can determine that the heart has not experienced abnormal motion states such as arrhythmia.

[0066] Please refer to Figure 7B It is based on Figure 7A Another application scenario diagram is shown. In this embodiment, it is assumed that the processor 104 obtains the following based on previous teaching: Figure 7BThe diagram 700a shows the changes in left ventricular volume, and we find several specific extreme values ​​711a to 715a corresponding to EDV and the corresponding time differences T1' to T4'.

[0067] exist Figure 7B If the processor 104 determines that the time difference T2' in the time difference T1' to T4' deviates from the statistical characteristic value of the time difference T1' to T4' to reach a preset threshold value, then the processor 104 can determine that the heart has an abnormal motion state (e.g., arrhythmia), but it is not limited to this.

[0068] In an embodiment of the present invention, the processor 104 may provide the judgment result of whether the heart has an abnormal motion state to relevant medical personnel as a diagnostic reference, but may not be limited thereto.

[0069] Furthermore, although the above embodiments use the left ventricular volume corresponding to the EDV as the specific extreme value to be considered, in other embodiments, the processor 104 may also use the left ventricular volume corresponding to the ESV as the specific extreme value to be considered. By definition, each ESV should correspond to the smallest left ventricular volume in its respective heart rate cycle. Based on this, if the processor 104 determines that the i-th left ventricular volume is smaller than the (i-1)-th and (i+1)-th left ventricular volumes, the processor 104 can determine that the i-th ventricular volume should correspond to the ESV, and further determine that the i-th left ventricular volume belongs to one of the aforementioned specific extreme values.

[0070] Based on this, in Figure 7B In this scenario, the processor 104 can accordingly find the left ventricular volume value corresponding to the ESV as a specific extreme value 711b to 715b, and estimate the corresponding time difference T1” to T4”.

[0071] exist Figure 7B If the processor 104 determines that the time difference T1” deviates from the statistical characteristic value of the time difference T1” to T4” when it determines that the time difference T1” deviates from the statistical characteristic value of the time difference T1” to T4”, then the processor 104 can determine that the heart has an abnormal motion state (e.g., arrhythmia) and can use it as a reference for relevant medical personnel to make a diagnosis.

[0072] In some embodiments, if the relevant medical personnel examine the corresponding Figure 7BIf, after reviewing the echocardiogram, the cardiac system is misjudged as exhibiting abnormal motion (i.e., the heart does not actually exhibit abnormal motion), relevant medical personnel can report this situation to the electronic device 100. In embodiments of the present invention, since the aforementioned misjudgment may be due to the poor recognition ability of the machine learning model for the left ventricular image region, the processor 104 can retrain the machine learning model accordingly to reduce the probability of future misjudgments, but this is not a limitation.

[0073] In summary, the method for estimating ventricular volume proposed in this invention involves obtaining a left ventricular mask image corresponding to a left ventricular ultrasound, and then identifying three pixels that meet specific conditions as reference point pixels corresponding to the apex and both mitral valves. The left ventricular volume can then be estimated based on these reference point pixels. Therefore, this invention can automatically estimate left ventricular volume with better efficiency without the need for manual marking of the apex and both mitral valves.

[0074] Furthermore, the method for assisting in the assessment of cardiac motion status proposed in this invention can identify multiple specific extreme values ​​corresponding to the EDV (or ESV) in multiple left ventricular volumes corresponding to multiple consecutive cardiac ultrasound images, and determine whether the heart has abnormal motion status such as arrhythmia based on the time difference between these specific extreme values. This allows relevant medical personnel to more easily grasp the condition of the heart, thereby reducing the probability of making incorrect assessments (such as calculating incorrect ejection fractions).

[0075] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for assisting in the assessment of cardiac motion status, suitable for electronic devices, comprising: Acquiring multiple consecutive echocardiographic images corresponding to the heart, and estimating multiple left ventricular volumes corresponding to the multiple echocardiographic images, including: Obtain the plurality of cardiac ultrasound images and determine whether each cardiac ultrasound image is an apical view; In response to determining that each of the cardiac ultrasound images belongs to the apical view, a left ventricular mask image corresponding to the left ventricle of the heart is extracted from each of the cardiac ultrasound images. This left ventricular mask image is a binary image, where pixels in the image region corresponding to the ventricle are set to a first value, and pixels in the image region not corresponding to the ventricle are set to a second value, thereby generating the corresponding binary image; and Estimating the left ventricular volume corresponding to each of the said cardiac ultrasound images based on the left ventricular mask image of each of the said cardiac ultrasound images includes: In the left ventricular mask image of each of the aforementioned cardiac ultrasound images, identify three reference point pixels, including: Each pixel in the left ventricular mask image is examined one by one, and three pixels in the left ventricular mask image that meet the conditions are defined as the three reference point pixels. Each reference point pixel has the first value. The condition is that each reference point pixel is surrounded by N surrounding pixels, and the plurality of surrounding pixels of each reference point pixel includes N1 first surrounding pixels with the first value and N2 second surrounding pixels with the second value, where N, N1, and N2 are positive integers, N1 is 3, N2 is 5, the first value is 255, and the second value is 0. The three reference point pixels include a first reference point pixel, a second reference point pixel, and a third reference point pixel. Among the plurality of surrounding pixels of the first reference point pixel, the plurality of first surrounding pixels are arranged in a straight line, and the plurality of second surrounding pixels are arranged in a C-shape. Among the plurality of surrounding pixels of the second reference point pixel and the third reference point pixel, the plurality of first surrounding pixels are arranged in an L-shape, and the plurality of second surrounding pixels are arranged in an L-shape. The left ventricular volume of each cardiac ultrasound image is estimated based on multiple reference point pixels in the left ventricular mask image of each cardiac ultrasound image. Identify multiple specific extreme values ​​among the multiple left ventricular volumes, and estimate multiple time differences between the multiple specific extreme values ​​accordingly; Estimate the statistical characteristic values ​​of the multiple time differences based on the multiple time differences; The heart is determined to be in an abnormal motion state if at least one of the plurality of time differences deviates from the statistical characteristic value to a preset threshold value.

2. The method of claim 1, wherein the step of extracting the left ventricular mask image corresponding to the left ventricle of the heart from each of the cardiac ultrasound images comprises: Each of the said cardiac ultrasound images is input into a machine learning model, wherein the machine learning model outputs the corresponding left ventricular mask image in response to each of the said cardiac ultrasound images.

3. The method according to claim 2, further comprising: In response to the determination that the heart has been misjudged as exhibiting the abnormal movement state, the machine learning model is retrained.

4. The method of claim 1, wherein the step of finding the plurality of specific extreme values ​​among the plurality of left ventricular volumes comprises: The reaction is to determine that the i-th left ventricular volume among the plurality of left ventricular volumes is greater than the (i-1)-th and (i+1)-th ventricular volumes among the plurality of left ventricular volumes, and to determine that the i-th ventricular volume belongs to one of the plurality of specific extreme values, where i is an integer.

5. The method of claim 1, wherein the step of identifying a plurality of specific extreme values ​​among the plurality of left ventricular volumes comprises: The reaction is to determine that the i-th left ventricular volume among the plurality of left ventricular volumes is less than the (i-1)-th and (i+1)-th ventricular volumes among the plurality of left ventricular volumes, and to determine that the i-th ventricular volume belongs to one of the plurality of specific extreme values.

6. The method according to claim 1, wherein the plurality of specific extreme values ​​includes the first to the Kth specific extreme values, the plurality of time differences includes the first to the (K-1)th time difference, wherein the j-th time difference among the plurality of time differences is the time difference between the (j+1)th and the j-th specific extreme values ​​among the plurality of specific extreme values. , where K is an integer.

7. The method according to claim 1, wherein the statistical characteristic value is the average of the plurality of time differences.

8. The method according to claim 1, wherein the abnormal motion state includes arrhythmia.

Images