A prenatal ultrasound-assisted screening method and system for congenital heart disease

Abstract

The present invention relates to the technical field of ultrasonic assisted screening, and particularly relates to a prenatal ultrasonic assisted screening method and system for congenital heart disease. First, the present invention obtains a frame sequence of four-chamber view images of a fetus, matches all suspected cardiac cavity regions between different frames of images, and obtains all region matching sequences; then, based on the diastolic and systolic rules of cardiac pulsation, it evaluates the cardiac cavity confidence coefficients of each region matching sequence, so as to screen out all cardiac cavity matching sequences; further analyzes and evaluates the normal indexes of each cardiac cavity region in the cardiac cavity matching sequences, and then screens out abnormal frame images from all frame images; finally, reconstructs the four-chamber view image based on all abnormal frame images. Based on the characteristics of regular changes in cardiac cavities caused by cardiac pulsation, the present invention analyzes and screens abnormal cardiac cavities in each four-chamber view image, and then distinguishes normal cardiac cavities from abnormal cardiac cavities for reconstruction, avoiding the situation where the reconstruction result cannot reflect the abnormal structure of the fetal heart, and improving the auxiliary screening effect of prenatal ultrasound.

Description

Technical Field

[0001] The present invention relates to the technical field of ultrasonic-assisted screening, and particularly relates to a prenatal ultrasonic-assisted screening method and system for congenital heart disease. Background Art

[0002] Congenital heart disease (hereinafter referred to as CHD) refers to a structural heart disease caused by disorders or abnormal development of the heart and large blood vessels during embryonic development. It is one of the common congenital malformations. Therefore, prenatal ultrasonic screening is crucial. In prenatal ultrasonic screening, the four-chamber view of the fetus is an important reference for evaluating CHD, which can help doctors evaluate whether the fetal heart structure is normal, so as to assist doctors to take necessary intervention measures, reduce the birth risk, and ensure the health of pregnant women and newborns.

[0003] However, due to factors such as the small size of the fetal heart, the angle and frequency of the ultrasonic probe, etc., the resolution or clarity of the four-chamber view may be low, and a single-frame four-chamber view cannot show the dynamic activities of the fetal heart. Therefore, it is usually necessary to use multiple frames of four-chamber views for super-resolution reconstruction to improve the resolution and quality of the image, helping doctors to observe the structure and potential abnormalities of the fetal heart more clearly. However, during the super-resolution reconstruction process, the frame images during normal heart beats may cover the frame images during abnormal beats, resulting in the super-resolution reconstructed image being unable to reflect the true abnormal structure of the fetal heart, thereby reducing the auxiliary screening effect of prenatal ultrasound. Summary of the Invention

[0004] In order to solve the technical problem that the super-resolution reconstructed image in the prior art cannot reflect the true abnormal structure of the fetal heart, resulting in poor auxiliary screening effect of prenatal ultrasound, the purpose of the present invention is to provide a prenatal ultrasonic-assisted screening method and system for congenital heart disease, and the specific technical solutions adopted are as follows:

[0005] The present invention proposes a prenatal ultrasonic-assisted screening method for congenital heart disease, and the method includes:

[0006] Obtain a frame sequence of four-chamber views of the fetus in the pregnant woman's abdomen and all suspected cardiac cavity regions in each frame; match all suspected cardiac cavity regions between different frames in the frame sequence, and use the suspected cardiac cavity regions as sequence elements to obtain all region matching sequences in the frame sequence, where the sorting method of the region matching sequence is the same as that of the frame sequence;

[0007] Based on the area change of the suspected cardiac cavity region, each region matching sequence is divided into a suspected cardiac cavity diastolic feature sequence and a suspected cardiac cavity systolic feature sequence; in the suspected cardiac cavity diastolic feature sequence and the suspected cardiac cavity systolic feature sequence, according to the area similarity of the suspected cardiac cavity region in each feature sequence and the change of the inter-frame interval of the corresponding frame images, combined with the area difference of the suspected cardiac cavity region between different feature sequences, the cardiac cavity confidence coefficient corresponding to the region matching sequence is obtained;

[0008] According to the cardiac cavity confidence coefficient, all cardiac cavity matching sequences are screened out from all region matching sequences, and the sequence elements in the cardiac cavity matching sequences are used as cardiac cavity regions; in any feature sequence of each cardiac cavity matching sequence, according to the area of each cardiac cavity region and the inter-frame interval between the corresponding frame images of each cardiac cavity region and adjacent cardiac cavity regions, the normal index of each cardiac cavity region is obtained; in the frame image sequence, the abnormal frame images are screened out from all frame images according to the normal index;

[0009] Based on all abnormal frame images, a reconstructed four-chamber view for auxiliary prenatal ultrasound screening is obtained.

[0010] Further, the method for obtaining the suspected cardiac cavity region includes:

[0011] Each four-chamber view is binarized, and the binarized frame image is subjected to region connectivity detection to obtain all connected domains in the background; the corresponding region of each connected domain in the frame image is used as a suspected cardiac cavity region in the corresponding frame image.

[0012] Further, the method for obtaining the region matching sequence includes:

[0013] An image coordinate system is constructed with the center of each frame image as the origin, and the centroid coordinates of each suspected cardiac cavity region in each frame image are obtained; taking any frame image as the target frame image and any one of the suspected cardiac cavity regions in the target frame image as the target region, the centroid coordinates of the target region are used as the target coordinates;

[0014] In each non-target frame image in the frame image sequence, according to the difference between the centroid coordinates of each suspected cardiac cavity region and the target coordinates, the matching region of the target region in the corresponding non-target frame image is obtained; the target region and its matching regions in all non-target frame images are sorted in the order of the frame images to construct a region matching sequence.

[0015] Further, the method for obtaining the suspected cardiac cavity diastolic feature sequence and the suspected cardiac cavity systolic feature sequence includes:

[0016] In each of the region matching sequences, taking the area of each suspected cardiac cavity region as a sequence element, an area change sequence corresponding to each region matching sequence is obtained, and the sorting method of the area change sequence is the same as that of the region matching sequence;

[0017] All maximum values and all minimum values in the area change sequence are obtained. The suspected cardiac cavity regions corresponding to the maximum values are sorted in the order of obtaining the maximum values to construct a suspected cardiac cavity diastolic feature sequence, and the suspected cardiac cavity regions corresponding to the minimum values are sorted in the order of obtaining the minimum values to construct a suspected cardiac cavity diastolic feature sequence.

[0018] Further, the method for obtaining the cardiac cavity confidence coefficient includes:

[0019] In the suspected cardiac cavity diastolic feature sequence and the suspected cardiac cavity systolic feature sequence, according to the area variance of the suspected cardiac cavity regions in each feature sequence and the frame interval between the frame graphs corresponding to adjacent suspected cardiac cavity regions, the pulsation law parameters of the suspected cardiac cavity regions in the corresponding region matching sequence are obtained;

[0020] The difference between the average area of all the suspected cardiac cavity regions in the suspected cardiac cavity diastolic feature sequence and the average area of all the suspected cardiac cavity regions in the suspected cardiac cavity systolic feature sequence is used as the cardiac cavity discrimination reference weight;

[0021] The pulsation law parameters are weighted by using the cardiac cavity discrimination reference weight, and the normalized result of the weighted result is used as the cardiac cavity confidence coefficient corresponding to the region matching sequence.

[0022] Further, the method for obtaining the pulsation law parameters includes:

[0023] The variance of the areas of all the suspected cardiac cavity regions in the suspected cardiac cavity diastolic feature sequence is used as the diastolic morphology fluctuation parameter; the variance of the areas of all the suspected cardiac cavity regions in the suspected cardiac cavity systolic feature sequence is used as the systolic morphology fluctuation parameter; the diastolic morphology fluctuation parameter and the systolic morphology fluctuation parameter are fused, and the negative correlation mapping result of the fusion result is used as the cardiac cavity morphology law parameter;

[0024] The frame interval between the frame graphs to which adjacent suspected cardiac cavity regions in the suspected cardiac cavity diastolic feature sequence belong is used as the suspected diastolic interval, and the cumulative value of the differences between all adjacent suspected diastolic intervals is used as the diastolic interval change parameter; the frame interval between the frame graphs to which adjacent suspected cardiac cavity regions in the suspected cardiac cavity systolic feature sequence belong is used as the suspected systolic interval, and the cumulative value of the differences between all adjacent suspected systolic intervals is used as the systolic interval change parameter; the diastolic interval change parameter and the systolic interval change parameter are fused, and the negative correlation mapping result of the fusion result is used as the pulsation interval law parameter;

[0025] Integrate the heart cavity morphological law parameters and the pulsation interval law parameters to obtain the pulsation law parameters of the suspected heart cavity regions corresponding to the region matching sequences.

[0026] Further, the method for obtaining the heart cavity matching sequences includes:

[0027] Based on the heart cavity confidence coefficient, sort all the region matching sequences in descending order, and use the top four region matching sequences in the sorting as the heart cavity matching sequences.

[0028] Further, the method for obtaining the normal index includes:

[0029] Take the suspected heart cavity diastolic feature sequences of each heart cavity matching sequence as the target feature sequences, take the average area of all heart cavity regions in the target feature sequences as the reference area, and take the average of the frame intervals between the frame graphs to which adjacent heart cavity regions belong as the reference heart cavity pulsation period;

[0030] In each target feature sequence, take the negative correlation mapping result of the area difference between the sum of the areas of each heart cavity region and the previous adjacent heart cavity region and twice the reference area as the first pulsation normal coefficient of each heart cavity region in the target feature sequence;

[0031] In each target feature sequence, take the negative correlation mapping result of the period difference between the frame interval to which each heart cavity region and the previous adjacent heart cavity region belong and the reference heart cavity pulsation period as the second pulsation normal coefficient of each heart cavity region in the target feature sequence;

[0032] Integrate the first pulsation normal coefficient and the second pulsation normal coefficient to obtain the normal index of the corresponding heart cavity regions in the target feature sequences corresponding to the heart cavity matching sequences.

[0033] Further, the method for obtaining the abnormal frame graphs includes:

[0034] Take the heart cavity regions with the normal index less than the preset threshold as abnormal heart cavity regions, and take the frame graphs containing the abnormal heart cavity regions as abnormal frame graphs.

[0035] The present invention also provides a prenatal ultrasound assisted screening system for congenital heart diseases, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the prenatal ultrasound assisted screening method for congenital heart diseases are implemented.

[0036] The present invention has the following beneficial effects:

[0037] The present invention obtains a frame sequence of four-chamber view images of a fetus in a pregnant woman's abdomen, matches all suspected cardiac cavity regions between different frames in the frame sequence to obtain all region matching sequences, where each region matching sequence reflects the change of the same cardiac tissue structure, and prepares for subsequent analysis to screen out abnormal frames that do not conform to the normal pulsation rule and reconstruct high-quality and high-resolution four-chamber view images; then, based on the area change of the suspected cardiac cavity region, each region matching sequence is divided into a suspected cardiac cavity diastolic feature sequence and a suspected cardiac cavity systolic feature sequence. Furthermore, according to the area similarity of the suspected cardiac cavity regions in each feature sequence and the change of the frame interval of the corresponding frames, combined with the area difference of the suspected cardiac cavity regions between different feature sequences, the cardiac cavity confidence coefficient of the corresponding region matching sequence is obtained. The cardiac cavity confidence coefficient quantifies the possibility of each region matching sequence being a cardiac cavity matching sequence between different frames under the influence of excluding interference from other non-cardiac tissues such as non-cardiac cavities, so as to facilitate subsequent accurate positioning of the cardiac cavity region and analysis of the inter-frame change of the cardiac cavity; further, all cardiac cavity matching sequences are screened out according to the cardiac cavity confidence coefficient, and then in any feature sequence of each cardiac cavity matching sequence, according to the area of each cardiac cavity region and the frame interval of the corresponding frames of each cardiac cavity region and adjacent cardiac cavity regions, the normal index of each cardiac cavity region is obtained; the normal index evaluates the possibility of each cardiac cavity region having normal pulsation function and prepares for subsequent screening of abnormal frames and reconstructing high-resolution and high-quality four-chamber view images; then, abnormal frames are screened out according to the normal index; finally, a reconstructed four-chamber view image for auxiliary prenatal ultrasound screening is obtained based on all abnormal frames. Based on the characteristic that the cardiac cavity has regular changes due to cardiac pulsation, the present invention analyzes and screens abnormal cardiac cavities in each four-chamber view image, and then distinguishes normal cardiac cavities and abnormal cardiac cavities for super-resolution reconstruction, avoiding the situation where the reconstruction result cannot reflect the true abnormal structure of the fetal heart, and improving the auxiliary screening effect of prenatal ultrasound. Description of the Drawings

[0038] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for description in the embodiments or the prior art. Obviously, the following described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0039] Figure 1 Flowchart of a method for auxiliary prenatal ultrasound screening of congenital heart disease provided by an embodiment of the present invention;

[0040] Figure 2 Flowchart of a method for obtaining the cardiac cavity confidence coefficient provided by an embodiment of the present invention. Detailed Embodiments

[0041] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a prenatal ultrasound-assisted screening method and system for congenital heart disease proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0042] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0043] The specific scheme of the prenatal ultrasound-assisted screening method and system for congenital heart disease provided by the present invention is described in detail below with reference to the accompanying drawings.

[0044] See also Figure 1 , which shows a flow chart of a method for prenatal ultrasound-assisted screening of congenital heart disease provided by an embodiment of the present invention, specifically comprising:

[0045] Step S1, obtaining a frame sequence of a four-chamber heart section image of a fetus in the pregnant woman's abdomen and all suspected heart chamber areas in each frame; matching all suspected heart chamber areas between different frames in the frame sequence, and using the suspected heart chamber areas as sequence elements to obtain all region matching sequences in the frame sequence, wherein the region matching sequence is consistent with the sorting method of the frame sequence.

[0046] In one embodiment of the present invention, a frame sequence of a four-chamber heart section image of a fetus in a pregnant woman's abdomen is first obtained, and then combined with the normal beating condition of the fetal heart, abnormal frames that do not conform to the normal beating law are analyzed and screened out, thereby reconstructing a high-quality and high-resolution four-chamber heart section image, and assisting relevant medical personnel in assessing the heart structure and preparing for screening for congenital heart disease;

[0047] The specific process of obtaining the frame image sequence is as follows: relevant medical personnel guide and assist the pregnant woman to be screened to lie on her back on the examination bed. After completing the preparations before the ultrasound examination, the medical personnel hold the ultrasound probe and gently move it to scan the pregnant woman's abdomen, looking for the best scanning position to clearly observe the fetal heart in the abdomen. Then, multiple consecutive frames of four-chamber heart sections of the fetal heart are collected at the scanning position. All four-chamber heart sections are sorted based on the order of collection to construct a frame image sequence to ensure that the heart beats as the frame images change, thereby ensuring the effectiveness of subsequent analysis.

[0048] It should be noted that obtaining the four-chamber view is an existing technology well-known to those skilled in the art, and will not be elaborated here; among them, the four-chamber view is already a grayscale image, so in subsequent processing, no further grayscale processing is required. If it is a color Doppler ultrasound, the implementer can also perform certain preprocessing before analysis.

[0049] Considering that normal cardiac pulsation will cause regular changes in the cardiac chamber volume, by analyzing the changes in the cardiac chamber area in the four-chamber view, it can help evaluate whether the fetal heart structure and function are abnormal; therefore, in an embodiment of the present invention, all suspected cardiac chamber regions in each frame diagram, that is, each four-chamber view, are further obtained to prepare for subsequent analysis of cardiac chamber changes.

[0050] Preferably, in an embodiment of the present invention, considering that the cardiac chamber region is usually a low echo region and has a low grayscale value in the four-chamber view; and considering that binarization can help distinguish the cardiac chamber from the regional cardiac tissue, and then the background region corresponding to the low grayscale value region can be used as the suspected cardiac chamber region; therefore, the method for obtaining the suspected cardiac chamber region includes:

[0051] Binarize each frame of the four-chamber view, perform region connectivity detection on the binarized frame diagram, and obtain all connected domains in the background; the corresponding region of each connected domain in the frame diagram is used as a suspected cardiac chamber region in the corresponding frame diagram.

[0052] It should be noted that binarization, obtaining the background in the binarized image, and region connectivity detection are all existing technologies well-known to those skilled in the art and will not be elaborated.

[0053] Considering that even though the fetal cardiac pulsation will cause changes in the morphology of each cardiac chamber, in multiple frames of four-chamber views continuously collected at the same scanning position, the position distribution of the same cardiac chamber should be highly similar, and will not be greatly offset due to changes in the patient's vital signs such as breathing and blood flow during the ultrasound examination or errors in the ultrasound equipment, and the fetal orientation in the pregnant woman's abdomen will not change greatly within a short period of scanning, that is, it will not cause a large offset in the position distribution of the cardiac chamber. Therefore, different suspected cardiac chamber regions in the frame diagram sequence can be matched based on the position characteristics of the suspected cardiac chamber regions, so as to facilitate subsequent screening of the cardiac chamber corresponding region matching sequence, and then evaluate its abnormal conditions in combination with the cardiac pulsation characteristics.

[0054] Preferably, in an embodiment of the present invention, considering that cardiac pulsation may cause the cardiac chamber volume to increase and decrease from time to time, that is, the shape of the corresponding region of each cardiac chamber in the four-chamber view will change, but its geometric center will not undergo a major offset; based on this, the method for obtaining the region matching sequence includes:

[0055] Construct an image coordinate system with the center of each frame diagram as the origin, and obtain the centroid coordinates of each suspected heart cavity area in each frame diagram; take any frame diagram as the target frame diagram, and take any suspected heart cavity area in the target frame diagram as the target area, and use the centroid coordinates of the target area as the target coordinates;

[0056] In each non-target frame diagram in the frame diagram sequence, according to the difference between the centroid coordinates of each suspected heart cavity area and the target coordinates, obtain the matching area of the target area in the corresponding non-target frame diagram; sort the target area and its matching areas in all non-target frame diagrams in the sorting order of the frame diagrams to construct an area matching sequence.

[0057] As an example, in each non-target frame diagram in the frame diagram sequence, specifically take the suspected heart cavity area including the target coordinates as the matching area of the target area in the corresponding non-target frame diagram; then take the target area and its matching areas in each non-target frame diagram as a series of areas to be matched between different frame diagrams in the frame diagram sequence, and then sort all the areas to be matched according to the sorting order of the frame diagrams in the frame diagram sequence, and use the areas to be matched as sequence elements to construct an area matching sequence;

[0058] By changing the target area, all area matching sequences between different frame diagrams in the frame diagram sequence can be obtained. Among them, each area matching sequence can be regarded as the change sequence of the same heart structure tissue, such as the same heart cavity, between different frame diagrams, that is, each area matching sequence reflects the change situation of the same heart structure tissue.

[0059] In another embodiment of the present invention, in each non-target frame diagram in the frame diagram sequence, the suspected heart cavity area corresponding to the centroid coordinate with the smallest Euclidean distance from the target coordinate can also be used as the matching area of the target area in the corresponding non-target frame diagram.

[0060] It should be noted that each target area corresponds to only one matching area in each non-target frame diagram, that is, each sequence element in the area matching sequence is a matching area in each frame diagram.

[0061] It should be noted that the construction of the image coordinate system and the acquisition of the centroid are both prior arts and will not be elaborated here.

[0062] In another embodiment of the present invention, the implementer can also directly use the SIFT algorithm to perform feature matching on the frame diagram sequence to obtain all area matching sequences, which is already a prior art and will not be elaborated here.

[0063] Step S2: Based on the area change of the suspected cardiac cavity region, each region matching sequence is divided into a suspected cardiac cavity diastolic feature sequence and a suspected cardiac cavity systolic feature sequence; in the suspected cardiac cavity diastolic feature sequence and the suspected cardiac cavity systolic feature sequence, according to the area similarity of the suspected cardiac cavity regions in each feature sequence and the change of the frame interval between the frames to which they belong, and combining the area differences of the suspected cardiac cavity regions between different feature sequences, the cardiac cavity confidence coefficient of the corresponding region matching sequence is obtained.

[0064] Considering that the pulsation characteristics of the heart cause it to perform alternating contraction and relaxation movements, the morphology and area of each cardiac cavity will also change alternately with the pulsation of the heart, but the morphology and area of other hypoechoic tissues except the cardiac cavity usually remain unchanged; based on this feature, the possibility that the region matching sequence is the corresponding matching sequence of the cardiac cavity can be evaluated, which is convenient for subsequent screening of the cardiac cavity region; therefore, in the embodiment of the present invention, first, based on the area change of the suspected cardiac cavity region, each region matching sequence is divided into a suspected cardiac cavity diastolic feature sequence and a suspected cardiac cavity systolic feature sequence, preparing for subsequent analysis of the cardiac cavity confidence coefficient of each region matching sequence.

[0065] Preferably, in an embodiment of the present invention, considering that the area of the cardiac cavity will change alternately with the relaxation or contraction of the heart, and during relaxation, the area of the cardiac cavity region tends to be the largest, while during contraction, the area of the cardiac cavity tends to be the smallest; based on this, the methods for obtaining the suspected cardiac cavity diastolic feature sequence and the suspected cardiac cavity systolic feature sequence include:

[0066] In each region matching sequence, taking the area of each suspected cardiac cavity region as a sequence element, the corresponding area change sequence of each region matching sequence is obtained, and the sorting method of the area change sequence is the same as that of the region matching sequence;

[0067] All the maximum values and all the minimum values in the area change sequence are obtained, and the suspected cardiac cavity regions corresponding to the maximum values are sorted in the order of obtaining the maximum values to construct the suspected cardiac cavity diastolic feature sequence, and the suspected cardiac cavity regions corresponding to the minimum values are sorted in the order of obtaining the minimum values to construct the suspected cardiac cavity diastolic feature sequence.

[0068] It should be noted that the methods for obtaining the maximum value and the minimum value and the construction of the sequence are all prior arts and will not be elaborated here.

[0069] Also considering that each cardiac chamber should have certain regular change characteristics during each diastolic or systolic phase of the heart, that is, the time intervals between two adjacent diastolic or two adjacent systolic phases should be similar, and the degree of diastolic or systolic for each time should be similar, which also indicates that the area of the cardiac chamber after each diastolic or systolic phase should be similar; if in the suspected cardiac chamber diastolic feature sequence or suspected cardiac chamber systolic feature sequence, the areas of the suspected cardiac chamber regions are relatively similar and the inter-frame interval changes of the corresponding frame images also have a certain regularity, it indicates that the possibility of this region matching sequence being the corresponding matching sequence of the cardiac chamber is greater, and the corresponding cardiac chamber confidence coefficient is also greater;

[0070] At the same time, it is also considered that for cardiac tissues such as non-cardiac chambers, due to little change in morphology or area, there may also be the regular change characteristics described above in their corresponding suspected cardiac chamber diastolic feature sequences or suspected cardiac chamber systolic feature sequences. However, due to the certain area difference during cardiac chamber contraction and relaxation, based on this, the area difference of the suspected cardiac chamber regions between different feature sequences can be further combined to accurately obtain the cardiac chamber confidence coefficient of the corresponding region matching sequence.

[0071] The cardiac chamber confidence coefficient quantifies the possibility of each region matching sequence being the corresponding matching sequence of the cardiac chamber between different frame images by comprehensively analyzing the differential characteristics and regular change characteristics during cardiac chamber diastolic and systolic phases, and under the condition of excluding the interference of other cardiac tissues such as non-cardiac chambers, so as to facilitate the subsequent accurate positioning of the cardiac chamber region, analyze the inter-frame change situation of the cardiac chamber, and thus assist in screening.

[0072] Preferably, in an embodiment of the present invention, the method for obtaining the cardiac chamber confidence coefficient includes:

[0073] Please refer to Figure 2 , which shows a flowchart of a method for obtaining a cardiac chamber confidence coefficient provided by an embodiment of the present invention, specifically including:

[0074] Step S201, in the suspected cardiac chamber diastolic feature sequence and the suspected cardiac chamber systolic feature sequence, according to the area variance of the suspected cardiac chamber regions in each feature sequence and the inter-frame interval of the corresponding frame images of adjacent suspected cardiac chamber regions, obtain the pulsation regularity parameters of the suspected cardiac chamber regions in the corresponding region matching sequence.

[0075] Considering that variance can reflect the fluctuation of data and can also help evaluate the similarity characteristics of data to a certain extent, the area similarity of the suspected cardiac chamber regions can be evaluated based on the area variance; also considering that in each feature sequence, the inter-frame interval between the frame images of adjacent suspected cardiac chamber regions reflects the diastolic interval or systolic interval of the suspected cardiac chamber regions. If the interval differences of all diastolic intervals or systolic intervals are smaller, it indicates that the diastolic or systolic is more regular, the pulsation of the suspected cardiac chamber regions is more regular, and the subsequent corresponding cardiac chamber confidence coefficient is also greater; based on this, the pulsation regularity parameters of the suspected cardiac chamber regions in each region matching sequence can be obtained.

[0076] It should be noted that the characteristic sequences refer to the suspected cardiac cavity diastolic characteristic sequences and the suspected cardiac cavity systolic characteristic sequences, and the same applies to the reference to the characteristic sequences in the following text, which will not be elaborated further.

[0077] In a preferred embodiment of the present invention, the method for obtaining the pulsation regularity parameter includes:

[0078] Taking the variance of the areas of all suspected cardiac cavity regions in the suspected cardiac cavity diastolic characteristic sequence as the diastolic morphological fluctuation parameter; taking the variance of the areas of all suspected cardiac cavity regions in the suspected cardiac cavity systolic characteristic sequence as the systolic morphological fluctuation parameter; fusing the diastolic morphological fluctuation parameter and the systolic morphological fluctuation parameter, and taking the negative correlation mapping result of the fusion result as the cardiac cavity morphological regularity parameter;

[0079] Taking the inter-frame interval between the frame maps to which adjacent suspected cardiac cavity regions in the suspected cardiac cavity diastolic characteristic sequence belong as the suspected diastolic interval, and taking the cumulative value of the differences between all adjacent suspected diastolic intervals as the diastolic interval change parameter; taking the inter-frame interval between the frame maps to which adjacent suspected cardiac cavity regions in the suspected cardiac cavity systolic characteristic sequence belong as the suspected systolic interval, and taking the cumulative value of the differences between all adjacent suspected systolic intervals as the systolic interval change parameter; fusing the diastolic interval change parameter and the systolic interval change parameter, and taking the negative correlation mapping result of the fusion result as the pulsation interval regularity parameter;

[0080] Fusing the cardiac cavity morphological regularity parameter and the pulsation interval regularity parameter to obtain the pulsation regularity parameter of the suspected cardiac cavity region in the corresponding region matching sequence.

[0081] As an example, taking any region matching sequence as an example, the calculation formula for the pulsation regularity parameter is:

[0082] ; where Q is the pulsation regularity parameter of the suspected cardiac cavity region in the region matching sequence; is the exponential function with the natural constant e as the base; is the serial number of the suspected cardiac cavity diastolic characteristic sequence of the region matching sequence; is the serial number of the suspected cardiac cavity systolic characteristic sequence of the region matching sequence; is the variance of the areas of all suspected cardiac cavity regions in the suspected cardiac cavity diastolic characteristic sequence, and is also the diastolic morphological fluctuation parameter; is the variance of the areas of all suspected cardiac cavity regions in the suspected cardiac cavity systolic characteristic sequence, and is also the systolic morphological fluctuation parameter; is the cardiac cavity morphological regularity parameter; is the cumulative value of the differences between all adjacent suspected diastolic intervals, and is also the diastolic interval change parameter; is the cumulative value of the differences between all adjacent suspected systolic intervals, and is also the systolic interval change parameter; is the beat interval regularity parameter; It is a very small positive parameter. In this example, it is set to 0.001 to prevent the denominator from being 0. The implementer can also define it by themselves.

[0083] In the above formula, the diastolic morphological fluctuation parameters and the systolic morphological fluctuation parameters are added and fused. The smaller the variance is, the more similar the areas in each feature sequence are. The negative correlation of the sum is mapped to the exponential function to adjust the logic and normalize it, so that the cardiac chamber morphological regularity parameters are larger; then the diastolic interval variation parameters and the systolic interval variation parameters are multiplied and combined, so that the smaller the product is, the more similar the intervals between adjacent frames in each feature sequence are, and then the product is added with a very small positive parameter and the inverse operation is performed to adjust the logic, so that the beat interval regularity parameters are larger; finally, the cardiac chamber morphological regularity parameters and the beat interval regularity parameters are multiplied and fused to obtain the beat regularity parameters.

[0084] In other examples, implementers may also adopt other methods such as weighted summation or other negative correlation mapping methods, which are commonly used technical methods and will not be described in detail here.

[0085] Step S202: The difference between the mean area value of all suspected cardiac chamber regions in the suspected cardiac chamber diastolic feature sequence and the mean area value of all suspected cardiac chamber regions in the suspected cardiac chamber contraction feature sequence is used as a reference weight for cardiac chamber discrimination.

[0086] It should be noted that if the area of ​​the suspected cardiac chamber region in the suspected cardiac chamber diastolic feature sequence is the maximum value in the area change sequence corresponding to the regional matching sequence, then the corresponding area mean will always be greater than or equal to the area mean of all suspected cardiac chamber regions in the suspected cardiac chamber contraction feature sequence, that is, the cardiac chamber discrimination reference weight should always be greater than or equal to 0. The larger the cardiac chamber discrimination reference weight, the more it has the characteristics of alternating changes in cardiac contraction and diastole, and the greater the possibility that the corresponding regional matching sequences of the suspected cardiac chamber diastolic feature sequence and the suspected cardiac chamber contraction feature sequence are the corresponding matching sequences of the cardiac chambers.

[0087] Step S203: weighting the beat regularity parameters using the cardiac chamber discrimination reference weights, and taking the normalized weighted results as the cardiac chamber confidence coefficients of the corresponding region matching sequence.

[0088] As an example, the cardiac chamber discrimination reference weight is multiplied by the pulsation regularity parameter for weighted combination, and then the product is linearly normalized, and the normalized result is used as the cardiac chamber confidence coefficient of the corresponding regional matching sequence; in other examples, the implementer may also use other normalization methods, which are not described here.

[0089] Step S3: According to the cardiac cavity confidence coefficient, all cardiac cavity matching sequences are screened out from all region matching sequences, and the sequence elements in the cardiac cavity matching sequences are used as cardiac cavity regions; in any characteristic sequence of each cardiac cavity matching sequence, according to the area of each cardiac cavity region and the frame interval between the corresponding frame diagrams of each cardiac cavity region and adjacent cardiac cavity regions, the normal index of each cardiac cavity region is obtained; in the frame diagram sequence, the abnormal frame diagrams are screened out from all frame diagrams according to the normal index.

[0090] After obtaining the cardiac cavity confidence coefficient of each region matching sequence, all cardiac cavity matching sequences can be further screened out from all region matching sequences according to the cardiac cavity confidence coefficient, so that the sequence elements in the cardiac cavity matching sequences can be used as cardiac cavity regions, and then based on the four cardiac cavity matching sequences, the four cardiac cavity regions in each four-chamber heart cross-sectional diagram can be correspondingly marked and obtained.

[0091] Preferably, in an embodiment of the present invention, considering that the heart mainly includes four cavity regions, namely the left atrium, the right atrium, the left ventricle, and the right ventricle, the greater the possibility that the four region matching sequences with the largest cardiac cavity confidence coefficient are the cardiac cavity corresponding matching sequences; therefore, the method for obtaining the cardiac cavity matching sequences includes: sorting all region matching sequences in descending order based on the cardiac cavity confidence coefficient, and using the top four region matching sequences as the cardiac cavity matching sequences.

[0092] Considering that congenital heart disease may affect the contraction and relaxation of the heart due to structural defects or deformities, and incomplete or inefficient contraction and relaxation may occasionally occur, the area of the cardiac cavity region in the four-chamber heart cross-sectional diagram may not be able to dilate to the maximum or contract to the minimum; and due to structural problems, the contraction and relaxation of the heart may be affected by delay or abnormality, resulting in irregular diastolic and systolic times; based on this, in an embodiment of the present invention, the normal index of each cardiac cavity region can be obtained in any characteristic sequence of each cardiac cavity matching sequence; the normal index evaluates the possibility that each cardiac cavity region has a normal pulsation function, and prepares for subsequent screening of abnormal frame diagrams to reconstruct high-resolution and high-quality four-chamber heart cross-sectional diagrams.

[0093] Preferably, in an embodiment of the present invention, considering that the diastolic and systolic of the cardiac cavity are regular, then during all diastolic or systolic processes, if the corresponding cardiac cavity area during any diastolic or systolic deviates more from its average level, it indicates that it deviates more from the original regularity, and the higher the possibility of its abnormality; similarly, when the interval corresponding to adjacent diastolic or adjacent systolic deviates more from the average level, it also indicates that it deviates more from the original regularity, and the higher the possibility of its abnormality; based on this, the method for obtaining the normal index includes:

[0094] Take the suspected diastolic feature sequence of each cardiac chamber matching sequence as the target feature sequence, take the average area of all cardiac chamber regions in the target feature sequence as the reference area, and take the average of the frame intervals of the frame diagrams to which adjacent cardiac chamber regions belong as the reference cardiac chamber pulsation period;

[0095] In each target feature sequence, take the negative correlation mapping result of the area difference between the sum of the areas of each cardiac chamber region and the previous adjacent cardiac chamber region and twice the reference area as the first pulsation normal coefficient of each cardiac chamber region in the target feature sequence;

[0096] In each target feature sequence, take the negative correlation mapping result of the period difference between the frame interval of the frame diagram to which each cardiac chamber region and the previous adjacent cardiac chamber region belong and the reference cardiac chamber pulsation period as the second pulsation normal coefficient of each cardiac chamber region in the target feature sequence;

[0097] Fuse the first pulsation normal coefficient and the second pulsation normal coefficient to obtain the normal index of the corresponding cardiac chamber region in the target feature sequence of the corresponding cardiac chamber matching sequence.

[0098] It should be noted that in other embodiments of the present invention, the implementer can also take the suspected systolic feature sequence of the cardiac chamber matching sequence as the target feature sequence to analyze the normal index of the cardiac chamber region; or can weight and fuse the normal indexes obtained by the two methods to obtain the final normal index, and the implementer can choose according to actual needs.

[0099] As an example, the calculation formula of the normal index is:

[0100] ; where x is the serial number of the cardiac chamber region in the target feature sequence; is the normal index of the x-th cardiac chamber region in the target feature sequence; is the exponential function with the natural constant e as the base; is the area of the x-th cardiac chamber region in the target feature sequence; is the area of the (x - 1)-th cardiac chamber region in the target feature sequence; is the average area of the cardiac chamber regions in the target feature sequence, and is also the reference area; is the frame interval between the frame diagrams to which the x-th cardiac chamber region and the (x - 1)-th cardiac chamber region belong in the target feature sequence; is the average of the frame intervals of the frame diagrams to which adjacent cardiac chamber regions belong in the target feature sequence; is a very small positive parameter, which is taken as 0.001 in this example to prevent the denominator from being 0, and the implementer can also define it by himself.

[0101] In the calculation formula of the normal index, the absolute value of the difference is specifically used to evaluate the area difference. The smaller the absolute value of the difference, the closer the area of each cardiac chamber region is to the area of the cardiac chamber region during other diastoles, that is, the closer it is to the reference area. The negative correlation of the area difference is mapped into the exponential function to adjust the logic and normalize it, so that the larger the normal coefficient of the first beat; then the inter-frame interval between each cardiac chamber region and the frame of the previous adjacent cardiac chamber region is regarded as a diastolic interval, and the absolute value of the difference is used to evaluate the difference between it and other diastolic intervals, that is, the reference cardiac chamber beat cycle. The smaller the absolute value of the difference, the closer the diastolic interval is to other diastolic intervals. Then, after adding a very small positive parameter to the absolute value of the difference, a reciprocal operation is performed to adjust the logic, so that the larger the normal coefficient of the second beat; finally, the normal coefficient of the first beat and the normal coefficient of the second beat are multiplied and fused to obtain the normal index.

[0102] It should be noted that there is no previous adjacent cardiac chamber region for the first cardiac chamber region in the suspected cardiac chamber diastolic feature sequence. In this example, the last cardiac chamber region can be regarded as the previous adjacent cardiac chamber region of the first cardiac chamber region, or it can be not taken as an analysis object, and it does not participate in the subsequent reconstruction.

[0103] In other examples, the implementer can also directly use the negative correlation mapping result of the area difference between the area of each cardiac chamber region and the reference area to evaluate the normal coefficient of the first beat; other means such as weighted summation or other negative correlation mapping can also be used, which are all common technical means and will not be elaborated here.

[0104] After obtaining the normal index of each cardiac chamber region, it is further known whether each cardiac chamber region in the corresponding frame of each four-chamber cardiac cross-sectional view in the frame sequence has normal beating characteristics. Then, abnormal frames can be screened out from all frames according to the normal index, preparing for the subsequent reconstruction.

[0105] Preferably, in an embodiment of the present invention, the cardiac chamber region with a normal index less than the preset threshold is used as an abnormal cardiac chamber region, and the frame containing the abnormal cardiac chamber region is used as an abnormal frame; in this example, the preset threshold is specifically set to 0.7, and the implementer can also define it by himself; when there is any cardiac chamber region in any frame as an abnormal cardiac chamber region, then that frame is used as an abnormal frame. The abnormal frame reflects the abnormal beating characteristics of the heart and can help analyze and evaluate whether there are defects and abnormalities in the cardiac structure.

[0106] Step S4, obtaining a reconstructed four-chamber cardiac cross-sectional view for assisting prenatal ultrasound screening based on all abnormal frames.

[0107] In an embodiment of the present invention, after obtaining all abnormal frame images, super-resolution reconstruction can be further performed based on all the abnormal frame images to obtain a reconstructed four-chamber view image with high resolution and high quality, which can assist in prenatal ultrasound screening. The implementer can also perform super-resolution reconstruction on the remaining normal frame images in the frame image sequence except for the abnormal frame images for reference by relevant medical staff to assist them in prenatal screening of congenital heart disease.

[0108] It should be noted that super-resolution reconstruction is already a well-known prior art to those skilled in the art and will not be elaborated here.

[0109] The present invention also provides a prenatal ultrasound-assisted screening system for congenital heart disease, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a prenatal ultrasound-assisted screening method for congenital heart disease.

[0110] In summary, the present invention first obtains a frame image sequence of the four-chamber view image of the fetus, matches all suspected cardiac cavity regions between different frame images, and obtains all region matching sequences; then, based on the diastolic and systolic laws during cardiac pulsation, analyzes and evaluates the cardiac cavity confidence coefficient of each region matching sequence, so as to screen out all cardiac cavity matching sequences from all region matching sequences; further analyzes and evaluates the normal index of each cardiac cavity region in the cardiac cavity matching sequence, and then screens out abnormal frame images from all frame images; finally, obtains a reconstructed four-chamber view image for assisting prenatal ultrasound screening based on all abnormal frame images. Based on the characteristic that the cardiac cavity changes regularly due to cardiac pulsation, the present invention analyzes and screens out abnormal cardiac cavities in each frame of the four-chamber view image, and then distinguishes normal cardiac cavities and abnormal cardiac cavities for super-resolution reconstruction, avoiding the situation where the reconstruction result cannot reflect the true abnormal structure of the fetal heart, and improving the effect of prenatal ultrasound-assisted screening.

[0111] It should be noted that the above sequence of embodiments of the present invention is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0112] Each embodiment in this specification is described in a progressive manner. The same or similar parts between the embodiments can be referred to each other, and the key points of each embodiment are the differences from other embodiments.

Claims

1. A prenatal ultrasound-assisted screening method for congenital heart disease, characterized in that: The method comprises: Obtain a frame sequence of a four-chamber heart section image of a fetus in the abdomen of a pregnant woman and all suspected heart chamber regions in each frame; match all suspected heart chamber regions between different frames in the frame sequence, and use the suspected heart chamber regions as sequence elements to obtain all region matching sequences in the frame sequence, wherein the region matching sequence is consistent with the ordering method of the frame sequence; Based on the area change of the suspected cardiac chamber area, each regional matching sequence is divided into a suspected cardiac chamber diastolic feature sequence and a suspected cardiac chamber contraction feature sequence; in the suspected cardiac chamber diastolic feature sequence and the suspected cardiac chamber contraction feature sequence, according to the similarity of the area of ​​the suspected cardiac chamber area in each feature sequence and the change of the inter-frame interval of the corresponding frame image, combined with the area difference of the suspected cardiac chamber area between different feature sequences, the cardiac chamber confidence coefficient corresponding to the regional matching sequence is obtained; According to the cardiac chamber confidence coefficient, all cardiac chamber matching sequences are screened out from all region matching sequences, and sequence elements in the cardiac chamber matching sequences are used as cardiac chamber regions; in any feature sequence of each cardiac chamber matching sequence, a normal index of each cardiac chamber region is obtained according to the area of ​​each cardiac chamber region and the frame interval between each cardiac chamber region and the corresponding frame images of the adjacent cardiac chamber regions; in the frame image sequence, abnormal frame images are screened out from all frame images according to the normal index; Based on all abnormal frames, a reconstructed four-chamber heart view is obtained to assist prenatal ultrasound screening; The method for obtaining the normal index includes: The suspected cardiac chamber diastolic feature sequence of each cardiac chamber matching sequence is used as a target feature sequence, the average area of ​​all cardiac chamber regions in the target feature sequence is used as a reference area, and the average value of the inter-frame intervals of the frames to which adjacent cardiac chamber regions belong is used as a reference cardiac chamber beat cycle; In each target feature sequence, a negative correlation mapping result of the area sum value of each cardiac chamber region and the previous adjacent cardiac chamber region and the area difference between the two times of the reference area is used as the first normal pulsation coefficient of each cardiac chamber region in the target feature sequence; In each target feature sequence, a negative correlation mapping result of the inter-frame interval between each cardiac chamber region and the frame image to which the previous adjacent cardiac chamber region belongs and the period difference between the reference cardiac chamber beat period is used as the second normal beat coefficient of each cardiac chamber region in the target feature sequence; The first normal pulsation coefficient and the second normal pulsation coefficient are fused to obtain a normal index of a corresponding cardiac chamber region in a target feature sequence corresponding to the cardiac chamber matching sequence.

2. The method for prenatal ultrasound-assisted screening of congenital heart disease according to claim 1, characterized in that: The method for acquiring the suspected cardiac cavity region comprises: Each frame of the four-chamber heart section image is binarized, and the binarized frame image is subjected to regional connectivity detection to obtain all connected domains in the background; the corresponding area of ​​each connected domain in the frame image is used as a suspected heart chamber area in the corresponding frame image.

3. The method for prenatal ultrasound-assisted screening of congenital heart disease according to claim 1, characterized in that: The method for obtaining the region matching sequence includes: The corresponding image coordinate system is constructed with the center of each frame image as the origin, and the centroid coordinates of each suspected cardiac cavity area in each frame image are obtained; any frame image is taken as the target frame image, any suspected cardiac cavity area in the target frame image is taken as the target area, and the centroid coordinates of the target area are taken as the target coordinates; In each non-target frame image in the frame image sequence, the matching area of ​​the target area in the corresponding non-target frame image is obtained according to the difference between the centroid coordinates of each suspected cardiac cavity area and the target coordinates; the target area and its matching areas in all non-target frame images are sorted in the sorting order of the frame images to construct a region matching sequence.

4. The method for prenatal ultrasound-assisted screening of congenital heart disease according to claim 1, characterized in that: The method for acquiring the suspected cardiac chamber diastolic feature sequence and the suspected cardiac chamber contraction feature sequence comprises: In each of the region matching sequences, the area of ​​each suspected cardiac cavity region is used as a sequence element to obtain an area change sequence corresponding to each of the region matching sequences, wherein the area change sequence is arranged in the same manner as the region matching sequence; Acquire all the maximum values ​​and all the minimum values ​​in the area change sequence, sort the suspected cardiac chamber regions corresponding to the maximum values ​​in the order of obtaining the maximum values ​​to construct a suspected cardiac chamber diastolic feature sequence, and sort the suspected cardiac chamber regions corresponding to the minimum values ​​in the order of obtaining the minimum values ​​to construct a suspected cardiac chamber diastolic feature sequence.

5. The method for prenatal ultrasound-assisted screening of congenital heart disease according to claim 1, characterized in that: The method for obtaining the cardiac chamber confidence coefficient includes: In the suspected cardiac chamber diastolic feature sequence and the suspected cardiac chamber contraction feature sequence, according to the area variance of the suspected cardiac chamber region in each feature sequence and the inter-frame interval of the corresponding frames of the adjacent suspected cardiac chamber regions, the pulsation regularity parameter of the suspected cardiac chamber region in the region matching sequence is obtained; The difference between the average area value of all the suspected cardiac chamber regions in the suspected cardiac chamber diastolic feature sequence and the average area value of all the suspected cardiac chamber regions in the suspected cardiac chamber contraction feature sequence is used as a reference weight for cardiac chamber discrimination; The cardiac chamber discrimination reference weight is used to weight the pulsation regularity parameter, and a normalized result of the weighted result is used as a cardiac chamber confidence coefficient corresponding to the regional matching sequence.

6. The method for prenatal ultrasound-assisted screening of congenital heart disease according to claim 5, characterized in that: The method for obtaining the pulsation regularity parameters includes: The variance of the area of ​​all the suspected cardiac chamber regions in the suspected cardiac chamber diastolic feature sequence is used as a diastolic morphological fluctuation parameter; the variance of the area of ​​all the suspected cardiac chamber regions in the suspected cardiac chamber contraction feature sequence is used as a contraction morphological fluctuation parameter; the diastolic morphological fluctuation parameter and the contraction morphological fluctuation parameter are fused, and the negative correlation mapping result of the fusion result is used as a cardiac chamber morphological regularity parameter; The inter-frame interval between the frames of the suspected cardiac chamber region adjacent to each other in the suspected cardiac chamber diastolic feature sequence is taken as the suspected diastolic interval, and the difference accumulation value between all the adjacent suspected diastolic intervals is taken as the diastolic interval variation parameter; the inter-frame interval between the frames of the suspected cardiac chamber region adjacent to each other in the suspected cardiac chamber systolic feature sequence is taken as the suspected systolic interval, and the difference accumulation value between all the adjacent suspected systolic intervals is taken as the systolic interval variation parameter; the diastolic interval variation parameter and the systolic interval variation parameter are fused, and the negative correlation mapping result of the fusion result is taken as the beat interval regularity parameter; The cardiac chamber morphology regularity parameters and the beat interval regularity parameters are integrated to obtain the beat regularity parameters of the suspected cardiac chamber region in the region matching sequence.

7. The method for prenatal ultrasound-assisted screening of congenital heart disease according to claim 1, characterized in that: The method for acquiring the cardiac chamber matching sequence comprises: All the region matching sequences are sorted in descending order based on the cardiac chamber confidence coefficient, and the top four region matching sequences are used as cardiac chamber matching sequences.

8. The method for prenatal ultrasound-assisted screening of congenital heart disease according to claim 1, characterized in that: The method for obtaining the abnormal frame image includes: The cardiac chamber region whose normal index is less than a preset threshold is regarded as an abnormal cardiac chamber region, and the frame image including the abnormal cardiac chamber region is regarded as an abnormal frame image.

9. A prenatal ultrasound-assisted screening system for congenital heart disease, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of a method for prenatal ultrasound-assisted screening of congenital heart disease as described in any one of claims 1 to 8 are implemented.

Images