CN117745726A - Left ventricular ejection fraction calculating method and device based on transesophageal echocardiography - Google Patents
Left ventricular ejection fraction calculating method and device based on transesophageal echocardiography Download PDFInfo
- Publication number
- CN117745726A CN117745726A CN202410190249.7A CN202410190249A CN117745726A CN 117745726 A CN117745726 A CN 117745726A CN 202410190249 A CN202410190249 A CN 202410190249A CN 117745726 A CN117745726 A CN 117745726A
- Authority
- CN
- China
- Prior art keywords
- image
- frame
- ejection fraction
- systole
- pixel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000002861 ventricular Effects 0.000 title claims abstract description 96
- 238000013175 transesophageal echocardiography Methods 0.000 title claims abstract description 40
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000004364 calculation method Methods 0.000 claims abstract description 41
- 238000003709 image segmentation Methods 0.000 claims abstract description 26
- 238000002604 ultrasonography Methods 0.000 claims description 33
- 230000008859 change Effects 0.000 claims description 15
- 238000002592 echocardiography Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 8
- 238000003062 neural network model Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 4
- 239000004576 sand Substances 0.000 claims description 4
- 239000004480 active ingredient Substances 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 abstract description 4
- 238000012800 visualization Methods 0.000 abstract description 3
- 210000005240 left ventricle Anatomy 0.000 description 29
- 238000004422 calculation algorithm Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 8
- 239000000523 sample Substances 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 238000005259 measurement Methods 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 239000008280 blood Substances 0.000 description 2
- 210000004369 blood Anatomy 0.000 description 2
- 230000008828 contractile function Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 210000001174 endocardium Anatomy 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 210000002784 stomach Anatomy 0.000 description 2
- 238000010967 transthoracic echocardiography Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000000747 cardiac effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000002496 gastric effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Landscapes
- Ultra Sonic Daignosis Equipment (AREA)
Abstract
The invention provides a left ventricular ejection fraction calculating method and device based on transesophageal echocardiography, comprising the following steps: obtaining an interesting region image of an echocardiogram of the ejection fraction to be calculated, inputting the interesting region image into a pre-trained image segmentation model, outputting a left chamber prediction result binary image, positioning a diastole end frame image and a systole end frame image according to the left chamber prediction result binary image, determining the actual distance of the inner diameter of the diastole end of the left chamber and the actual distance of the inner diameter of the systole end of the left chamber, then calculating the volume of the diastole end of the left chamber and the volume of the systole end of the left chamber, and finally calculating the ejection fraction of the left chamber. According to the invention, the ventricular structure characteristics are obtained from the B-type ultrasonic image, the ventricular outline is extracted, and the outline is analyzed, so that the visualization of the ventricle is realized, the method has the characteristic of high segmentation accuracy, the image recognition degree of the device can be remarkably improved, and the calculation accuracy of the ejection fraction is greatly improved.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a left ventricular ejection fraction calculation method and device based on transesophageal echocardiography.
Background
There are numerous indicators in the development of echocardiography to assess left ventricular contractile function. The left ventricular (hereinafter referred to as left ventricular) ejection fraction, which is the proportion of blood pumped from the left ventricle to the end-diastole volume of the left ventricle per cardiac cycle, is currently the most common and important indicator in clinical practice to assess left ventricular contractile function. The left ventricular ejection fraction is obtained by calculating the absolute value of the difference between the left ventricular end diastole volume and the end systole volume as a percentage of the end diastole volume. The calculation formula is as follows: left ventricular ejection fraction (%) = [ (end diastole volume-end systole volume)/end diastole volume ] ×100%.
The automatic calculation of the left ventricular ejection fraction of the echocardiogram has important significance, and can overcome a plurality of defects existing in manual calculation, such as experience dependence, operator variability, time consumption and the like. Echocardiography is largely divided into two major categories, transthoracic echocardiography and transesophageal echocardiography. At present, an automatic calculation method of left ventricular ejection fraction of transthoracic echocardiography is common, but the automatic calculation method of left ventricular ejection fraction of transesophageal echocardiography is still immature.
The short axis section of the left ventricle of the stomach is one of the most common methods for measuring the ejection fraction of the left ventricle by using an esophageal echocardiography, and has the advantages of non-invasiveness, wide application range, reliable measurement result, repeatable operation, response to real-time change and the like. The calculation of echocardiography left ventricular ejection fraction through the short axis section of the left ventricular chamber of the stomach is firstly to measure two indexes of the volume of the end diastole and the volume of the end systole of the left ventricular chamber. For transgastric left ventricular short axis sectional echocardiography, the current conventional method is to calculate using M-mode ultrasound in combination with Teichholz's formula. Teichholz formula isWherein V represents left chamber volume (ml) and D represents ventricular inner diameter (cm).
The M-type ultrasound (hereinafter referred to as M-ultrasound) requires an operator to specify the position of a sampling line passing through the ultrasonic probe and the midpoint of the left chamber diameter in the short axis section of the left chamber in a two-dimensional echocardiogram, then manually mark the positions of the anterior wall and the lower wall of the left chamber at the end diastole and the end systole on the M-line, and obtain the volume at the end diastole and the volume at the end systole according to a Teichholz formula. The selection of the M supersampling line, the selection of the end diastole inner diameter and the end systole inner diameter and the like are highly dependent on subjective experience of operators, are easily affected by factors such as ultrasonic image quality and difference among operators, are low in repeatability, and consume high time and cost.
Disclosure of Invention
In view of the above, the invention provides a left ventricular ejection fraction calculating method and device based on transesophageal echocardiography, which aims to solve the problem that the accuracy of the calculation result of the existing left ventricular ejection fraction is lower.
The first aspect of the invention provides a left ventricular ejection fraction calculating method based on transesophageal echocardiography, which comprises the following steps:
step 1, acquiring an ultrasonic image of a region of interest of an echocardiogram of which the ejection fraction is to be calculated, inputting the ultrasonic image of the region of interest into a pre-trained image segmentation model, and outputting a binary image of a left room prediction result;
step 2, generating a left-room pixel area change curve according to the binary image of the left-room prediction result, and positioning an end diastole frame image and an end systole frame image by utilizing the left-room pixel area change curve;
and 3, determining the left end diastole inner diameter physical distance and the left end systole inner diameter physical distance by using the end diastole frame image and the end systole frame image, calculating the left end diastole volume and the left end systole volume according to the left end diastole inner diameter physical distance and the left end systole inner diameter physical distance, and further calculating to obtain the left chamber ejection fraction.
Further, in the above left ventricular ejection fraction calculating method based on transesophageal echocardiography, in the step 1, an ultrasound image of a region of interest in a transesophageal left ventricular short-axis tangent plane echocardiography is obtained, and the ultrasound image of the region of interest and a corresponding left ventricular binary mask image are used as a sample data set, and a deep neural network model is pre-trained to obtain an image segmentation model.
Further, in the method for calculating left ventricular ejection fraction based on transesophageal echocardiography, the pre-training process of the image segmentation model in the step 1 includes:
acquisition ofMExample echocardiography of the left ventricular short axis section of the stomachE i WhereiniRepresent the firstiIn an example of this,E i IncludedN i frame two-dimensional ultrasound images co-obtainedKFrame two-dimensional original ultrasound imageI j The method comprises the steps of carrying out a first treatment on the surface of the Wherein,M i 、N i are all positive integers;
the two-dimensional original ultrasound image for each frameI j Using a binary image having a sectorB j And (3) withI j Multiplying together to obtainKFrame region of interest ultrasound imageJ j ；
Ultrasound images of the region of interest for each frameJ j Acquisition ofKFrame left room binary mask imageL j ，L j In the left cell region, the pixel value is 255, and the pixels except the left cell region have a value of 0, which willKFrame(s)J j AndKframe(s)L j Respectively reduced to imagesAndwill beKFrame->And->As a sample dataset, an image segmentation model is generated.
Further, in the method for calculating left ventricular ejection fraction based on transesophageal echocardiography, the binary imageB j Is respectively the width and the height ofW j AndH j the pixel value of the sector area is 1, the pixel values except the sector area are 0, and the circle center coordinates of the sector area are @x j ,y j )，x j =α j W j ，0.1≤α j ≤ 0.9，y j =β j H j ，0 ≤β j Less than or equal to 0.8, the radius of the sector area isr j ，r j = γ j H j ，0.2 ≤γ j The angle of the sector area is less than or equal to 0.8θ j ，80°≤θ j ≤130°。
Further, in the above left ventricular ejection fraction calculating method based on transesophageal echocardiography, the step 1 includes:
echocardiography for ejection fraction to be calculatedG，GIncludednFrame two-dimensional ultrasound imageFUsing a binary image having a sectorB F And (3) withFMultiplying together to obtainnFrame region of interest ultrasound imageJ F The method comprises the steps of carrying out a first treatment on the surface of the Wherein, the content of the active ingredients is less than or equal to 40 percentn≤300，FIs respectively the width and the height ofW F AndH F ；
will beJ F Is reduced toLoading the image segmentation model, and taking an echocardiogram +.>A kind of electronic devicenInputting the ultrasonic images of the region of interest into the image segmentation model frame by frame, and outputtingnFrame left room prediction result binary image +.>And will->Processing the left room predicted result binary image into an enlarged left room predicted result binary imageQ，QIs respectively the width and the height ofW Q AndH Q ，W Q =W F ,H Q =H F 。
further, in the method for calculating left ventricular ejection fraction based on transesophageal echocardiography, the method comprises the steps ofB F Is respectively the width and the height ofW BF AndH BF ，W BF =W F ，H BF =H F ,B F wherein the pixel value of the sector area is 1, the pixel values except the sector area are 0, and the center coordinates of the sector area are [ ]x F ,y F )，x F =α F W BF ，0.1≤α F ≤ 0.9，y F =β F H BF ，0 ≤β F Less than or equal to 0.8, the radius of the sector area isr F ，r F = γ F H BF ，0.2 ≤γ F The angle of the sector area is less than or equal to 0.8θ F ，80°≤θ F ≤ 130°。
Further, in the above left ventricular ejection fraction calculating method based on transesophageal echocardiography, the step 2 includes:
for the left bin predictor binary imageQRespectively calculateQThe number of pixel points with the pixel value of 255 in each frame of image to obtain the left room pixel area of each frame of image, and drawing a curve of the left room pixel area changing along with the frame numberSAnd for the left chamber pixel area with the frame number curveSSmoothing to obtain a smooth curveWill smooth curve->One maximum value adjacent to the upper positionS max And a minimum value ofS min As the left chamber pixel area at end diastole and the left chamber pixel area at end systole of the left chamber respectively,S max >S min >0 maximum valueS max Corresponding abscissa isuMinimum value ofS min Corresponding abscissa isv，0<u≤n，0<v≤n，Q u AndQ v an end diastole frame image and an end systole frame image, respectively.
Further, in the above left ventricular ejection fraction calculating method based on transesophageal echocardiography, the step 3 includes:
respectively determiningQ u AndQ v minimum circumcircle of middle left chamber areaC u AndC v according toC u Center coordinates of (2)The center coordinates of the fan-shaped areax F ,y F ) Is connected with the line of (a)L u And (3) withQ u The distance between two intersection points of the edges of the left chamber region, resulting in a left chamber end diastole inside diameter pixel distanceD ED1 The method comprises the steps of carrying out a first treatment on the surface of the According toC v Center coordinates of>The center coordinates of the fan-shaped areax F ,y F ) Is connected with the line of (a)L v And (3) withQ v The distance between two intersection points of the edges of the left chamber region, resulting in a left chamber end-systolic inside diameter pixel distanceD ES1 ；
Extracting an echocardiogram of the ejection fraction to be calculatedGIs set to be equal to or smaller than the pixel pitch of (1)G p Distance of left end diastole inner diameter pixelD ED1 Distance from left end-systole inner diameter pixelD ES1 Respectively with the pixel spacingG p Multiplying to obtain the actual distance of the end diastole inner diameter of the left chamberD ED2 Actual distance from left end-systole inner diameterD ES2 According toD ED2 AndD ES2 respectively calculating to obtain left end diastole volume and left end systole volume, and further calculating to obtain left room ejection blood componentA number.
Further, in the above left ventricular ejection fraction calculation method based on transesophageal echocardiography, the left ventricular end-diastole volume, the left ventricular end-systole volume and the left ventricular ejection fraction are determined according to the following formulas (1) to (3), respectively:
（1）
wherein,EDVrepresenting the end-diastole volume of the left chamber,D ED2 representing the actual distance of the left end diastole inner diameter;
（2）
wherein,ESVindicating the end-systolic volume of the left chamber,D ES2 representing the actual distance of the left chamber end-systole inner diameter;
LVEF=[(EDV-ESV)/EDV]×100% （3）
wherein LVEF represents left ventricular ejection fraction.
The invention also provides a left ventricle ejection fraction calculation left ventricle prediction result binary image generation module based on the transesophageal echocardiogram, which is used for acquiring an ultrasonic image of a region of interest of an echocardiogram with ejection fraction to be calculated, inputting the ultrasonic image of the region of interest into a pre-trained image segmentation model, and outputting a left ventricle prediction result binary image;
the end diastole frame and end systole frame image determining module is used for generating a left room pixel area change curve according to the binary image of the left room prediction result and positioning an end diastole frame image and an end systole frame image by utilizing the left room pixel area change curve;
the left chamber ejection fraction determining module is used for determining a left chamber end diastole inner diameter physical distance and a left chamber end systole inner diameter physical distance according to the end diastole frame image and the end systole frame image, calculating a left chamber end diastole volume and a left chamber end systole volume according to the left chamber end diastole inner diameter physical distance and the left chamber end systole inner diameter physical distance, and further calculating the left chamber ejection fraction.
A third aspect of the invention provides an apparatus comprising a processor, a memory and a computer program stored on the memory and operable on the processor, the computer program when executed by the processor implementing a transesophageal echocardiography-based left ventricular ejection fraction calculation method as claimed in any of the preceding claims.
A fourth aspect of the invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements a left ventricular ejection fraction calculation method based on transesophageal echocardiography as set forth in any of the preceding claims.
According to the invention, ultrasonic video images are sequentially input into a trained deep neural network image segmentation model, a left ventricle prediction result binary image is output, a ventricular profile corresponding to each image can be obtained according to the left ventricle prediction result binary image, after a diastole end frame image and a systole end frame image are positioned according to the left ventricle prediction result binary image, a left ventricle diastole end internal diameter physical distance and a left ventricle end systole end volume are calculated by combining the diastole end frame image, the systole end frame image and a left ventricle profile simulation M super heart internal diameter calculation method, and then a left ventricle diastole end volume and a left ventricle end systole volume are calculated, so that the left ventricle ejection fraction is calculated, the operation repeatability is improved, the calculation complexity is reduced, and the calculation result accuracy is greatly improved.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to designate like parts throughout the figures. In the drawings:
FIG. 1 is a flow chart of a left ventricular ejection fraction calculation method based on transesophageal echocardiography according to an embodiment of the present invention;
fig. 2 is a process diagram of acquiring an ultrasound image of a sector region of interest in a left ventricular ejection fraction calculation method based on transesophageal echocardiography according to an embodiment of the present invention, wherein: (a) is a two-dimensional original ultrasound image, (b) is a sector area binary image, and (c) is a region of interest ultrasound image;
fig. 3 is a schematic diagram of an ultrasound image of a region of interest and a left ventricular binary mask image in a left ventricular ejection fraction calculation method based on transesophageal echocardiography according to an embodiment of the present invention, wherein: (a) An ultrasound image of the region of interest of the trans-gastric left ventricular short axis facet, (b) a left ventricular binary mask image;
FIG. 4 is a diagram of a deep neural network model in a left ventricular ejection fraction calculation method based on transesophageal echocardiography according to an embodiment of the present invention;
fig. 5 is a graph showing the change of the left ventricular pixel area with the number of frames before and after the optimization using the filtering in the left ventricular ejection fraction calculating method based on the transesophageal echocardiography according to the embodiment of the present invention, wherein: (a) A graph of the pixel area of the left chamber before optimization and the frame number, and (b) a graph of the pixel area of the left chamber after optimization and the frame number;
fig. 6 is a schematic diagram of a fitting method of ventricular inner diameters of a left ventricular end diastole frame and a left ventricular end systole frame in a left ventricular ejection fraction calculating method based on transesophageal echocardiography according to an embodiment of the present invention, wherein: (a) (c) respectively representing a binary image of a left ventricle prediction result of an end-diastole frame and a schematic diagram of a ventricular endocardium fitting method, and (b) and (d) respectively representing a binary image of a left ventricle prediction result of an end-systole frame and a schematic diagram of a ventricular endocardium fitting method.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. It should be noted that, without conflict, the embodiments of the present invention and features of the embodiments may be combined with each other. The invention will be described in detail below with reference to the drawings in connection with embodiments.
Referring to fig. 1, a left ventricular ejection fraction calculation method based on transesophageal echocardiography according to an embodiment of the present invention includes:
step S1, obtaining an ultrasonic image of a region of interest of an echocardiogram of which the ejection fraction is to be calculated, inputting the ultrasonic image of the region of interest into a pre-trained image segmentation model, and outputting a binary image of a left room prediction result.
Specifically, the ultrasonic image of the region of interest is selected from the echocardiography calculated by the ejection fraction in advance, and then the region of interest is segmented, so that the accurate identification of the ventricular profile is facilitated. In the step, the ventricular contours can be identified through the output binary images of the left ventricular predicted results, so that the visualization of ventricles is realized, the segmentation accuracy of images can be improved, and the image recognition degree is remarkably improved.
In the step, firstly, an ultrasonic image of a region of interest in an ultrasonic cardiogram of a short axis section of a gastric left ventricle is obtained, then the ultrasonic image of the region of interest and a corresponding binary mask image of the left ventricle are used as a sample data set, and a deep neural network model is pre-trained, so that an image segmentation model can be obtained.
In the step, firstly, the ultrasonic image of the region of interest is segmented, so that the processing time can be reduced, and the image processing precision can be improved.
In the specific implementation, after initializing a left-room segmented deep neural network model, training the deep neural network model by adopting a preselected ultrasonic image of a region of interest and a corresponding left-room binary mask image to obtain an image segmentation model.
And S2, generating a left-room pixel area change curve according to the binary image of the left-room prediction result, and positioning end diastole frame and end systole frame images by utilizing the left-room pixel area change curve.
Specifically, after a curve of the pixel area of the left cell with respect to the number of frames is processed, the peak on the processed curve is used as an end diastole frame image, and the trough is used as an end systole frame image. And searching the end diastole frame image and the end systole frame image from the B ultrasonic image, thereby being beneficial to reducing the calculation complexity.
And S3, determining the left end diastole inner diameter physical distance and the left end systole inner diameter physical distance by utilizing the end diastole frame image and the end systole frame image, calculating the left end diastole volume and the left end systole volume according to the left end diastole inner diameter physical distance and the left end systole inner diameter physical distance, and further calculating to obtain the left chamber ejection fraction.
In the step, the calculating method for simulating the ventricular internal diameter of the M ultrasonic calculates the left ventricular end diastole physical distance and the left ventricular end systole physical distance, further calculates the left ventricular end diastole volume and the left ventricular end systole volume, finally obtains the left ventricular ejection fraction, can realize the rapid and accurate calculation of the ejection fraction, and greatly reduces the operation time and the operation error of operators.
The above can obviously be obtained: in the embodiment of the invention, ultrasonic video images are sequentially input into a trained depth neural network image segmentation model, a left ventricle prediction result binary image is output, a ventricular profile corresponding to each image can be obtained according to the left ventricle prediction result binary image, after a diastole end frame image and a systole end frame image are positioned according to the left ventricle prediction result binary image, a left ventricle diastole end diameter physical distance and a left ventricle end volume and a left ventricle end systole volume are calculated by combining the diastole end frame image, the systole end frame image and a left ventricle profile simulation M super-endocardial calculation method, and then the left ventricle ejection fraction is calculated, so that the repeatability of operation is improved, the calculation complexity is reduced, and the accuracy of a calculation result is greatly improved.
In the above embodiment, the pre-training process of the image segmentation model in step S1 includes:
substep S11, obtainingMExample echocardiography of the left ventricular short axis section of the stomachE i WhereiniRepresent the firstiIn an example of this,E i IncludedN i frame two-dimensional ultrasound images co-obtainedKFrame two-dimensional original ultrasound imageI j Wherein, the method comprises the steps of, wherein,M i 、N i are all positive integers;，jrepresent the firstjFrame, 1-2j≤K，I j Is respectively the width and the height ofW j AndH j . Preferably, 100.ltoreq.100M i ≤10000；30≤N i And 300. Ltoreq.300, in this embodiment,M=320，W j =800，H j =600。
referring to fig. 2, substep S12, for each frame said two-dimensional raw ultrasound imageI j Using a binary image having a sectorB j And (3) withI j Multiplying together to obtainKFrame region of interest ultrasound imageJ j . Wherein: binary imageB j Is respectively the width and the height ofW j AndH j the pixel value of the sector area is 1, the pixel values except the sector area are 0, and the circle center coordinates of the sector area are @x j ,y j )，x j =α j W j ，0.1≤α j ≤ 0.9，y j =β j H j ，0 ≤β j Less than or equal to 0.8, the radius of the sector area isr j ，r j = γ j H j ，0.2 ≤γ j The angle of the sector area is less than or equal to 0.8θ j ，80°≤θ j Not more than 130 degrees. In the present embodiment of the present invention, in the present embodiment,α j =0.54，β j =0.18，γ j =0.713，θ j =93.7°。
referring to fig. 3, substep S13, for each frame, the region of interest ultrasound imageJ j Acquisition ofKFrame left room binary mask imageL j ，L j In the left cell region, the pixel value is 255, and the pixels except the left cell region have a value of 0, which willKFrame(s)J j AndKframe(s)L j Respectively reduced to imagesAnd->Will beKFrame->And->As a sample dataset, an image segmentation model is generated.
In specific implementation, interpolation algorithm is utilized to calculate the value of the interpolationKFrame(s)J j AndKframe(s)L j Respectively reduced to imagesAnd->，/>And->Is of the width and the heightl，64≤l512 or less, in this embodiment,l=256. Preferably, in this embodiment, a bilinear interpolation algorithm is used to reduce the image.
Referring to fig. 4, in the deep neural network model structure diagram in the present embodiment, the image depthD=1，Image widthH=256, image heightW=256. In order to reduce the size and the layer number of the network, the embodiment of the invention simplifies the coding layer structure, and changes the network structure based on the characteristics extracted by four convolution blocks (each convolution block comprises 23×3 convolution blocks and 1×1 convolution structure, and the number is 3,4,6,3) in the ResNet50 into a structure of double-layer convolution downsampling, which can obviously reduce the parameter quantity of the network and is beneficial to improving the calculation efficiency. The self-attention layer only adopts the coding layer structure of the transducer structure, the decoding layer structure is abandoned, and the optimal layer number of the transducer and the optimal number of the heads of the multi-head attention mechanism are found. The model has an input dimension during training of (b×0C ×1H ×2W) = (8× 3 3 ×256×256), where B, C, H and W represent the training batch, the number of channels per picture, the height and width of each picture, respectively, and an output dimension of (b×c×h×w) = (8×3×256×256). The parameters of the model are selected as a transducer layer number 12 and a multi-head attention mechanism head number 8.
In the above embodiments, step S1 further includes:
substep S21, for echocardiography in which ejection fraction needs to be calculatedG，GIncludednFrame two-dimensional ultrasound imageFUsing a binary image having a sectorB F And (3) withFMultiplying together to obtainnFrame region of interest ultrasound imageJ F The method comprises the steps of carrying out a first treatment on the surface of the Wherein, the content of the active ingredients is less than or equal to 40 percentn≤300，FIs respectively the width and the height ofW F AndH F 。
in the present embodiment of the present invention,Grefers to echocardiography that does not participate in training the deep neural network model. The saidB F Is respectively the width and the height ofW BF AndH BF ，B F wherein the pixel value of the sector area is 1, the pixel values except the sector area are 0, and the center coordinates of the sector area are [ ]x F ,y F )，x F =α F W BF ，0.1≤α F ≤ 0.9，y F =β F H BF ，0 ≤β F Less than or equal to 0.8, the radius of the sector area isr F ，r F = γ F H BF ，0.2 ≤γ F The angle of the sector area is less than or equal to 0.8θ F ，80°≤θ F Not more than 130 degrees. In the present embodiment of the present invention,n=65，W F =800，H F =600，α F =0.54，β F =0.18，γ F =0.713，θ F =93.7°。
substep S22 ofJ F Is reduced toLoading the image segmentation model, and taking an echocardiogram +.>A kind of electronic devicenInputting the ultrasonic images of the region of interest into the image segmentation model frame by frame, and outputtingnFrame left room prediction result binary image +.>Will->Processing the left room predicted result binary image into an enlarged left room predicted result binary imageQ，QIs respectively the width and the height ofW Q AndH Q ，W Q =W F ,H Q =H F 。
in specific implementation, interpolation algorithm is utilized to calculate the value of the interpolationJ F Is reduced to。/>In which the pixel value of the left chamber region is 255 and the pixel values other than the left chamber region are 0, the interpolation algorithm is used to calculate +.>Enlarged to left room prediction result binary imageQ，QThe number of image frames isn。
It should be noted that, in this embodiment, parameters for enlarging and reducing the image may be determined according to actual situations such as network conditions, model training time, and computing power of the computer.
In the above embodiments, step S2 further includes:
for the left bin predictor binary imageQRespectively calculateQThe number of pixel points with the pixel value of 255 in each frame of image to obtain the left room pixel area of each frame of image, and drawing a curve of the left room pixel area changing along with the frame numberSAnd for the left chamber pixel area with the frame number curveSSmoothing to obtain a smooth curveWill smooth curve->One maximum value adjacent to the upper positionS max And a minimum value ofS min As the left chamber pixel area at end diastole and the left chamber pixel area at end systole respectively,S max >S min >0 maximum valueS max Corresponding abscissa isuMinimum value ofS min Corresponding abscissa isv，0<u ≤n，0<v≤n，40≤n≤300，Q u AndQ v an end diastole frame image and an end systole frame image, respectively.
In specific implementation, a smooth filtering algorithm is adopted to make the left-room pixel area change curve with the frame numberSAnd (5) processing. More specifically, a Savitzky-Golay filtering algorithm is adopted to carry out data points in a window with a certain widthkAnd (5) fitting the order polynomial, thereby obtaining an algorithm of a fitted result. In this embodiment, the filter window width is 15, and the polynomial order of the fitting samples is set to 5. The area-frame number curves before and after the filter optimization are shown in fig. 5.
In this embodiment, extremum searching algorithm is used for detectionIs to locate one local maximum adjacent to the other local minimumS max And a minimum value ofS min Respectively as the end diastole left cell pixel area and end systole left cell pixel area. More specifically, the extremum searching algorithm Python toolkit scipy has a scipy.signal.find_peaks () peak detection function, the minimum horizontal distance between adjacent peaks is 25 frames,u=29，v=47。
in each of the foregoing embodiments, the step 3 includes:
substep S31 of determining respectivelyQ u AndQ v minimum circumcircle of middle left chamber areaC u AndC v according toC u Center coordinates of (2)The center coordinates of the fan-shaped areax F ,y F ) Is connected with the line of (a)L u And (3) withQ u The distance between two intersection points of the edges of the left chamber region, resulting in a left chamber end diastole inside diameter pixel distanceD ED1 The method comprises the steps of carrying out a first treatment on the surface of the According toC v Center coordinates of>The center coordinates of the fan-shaped areax F ,y F ) Is connected with the line of (a)L v And (3) withQ v The distance between two intersection points of the edges of the left chamber region, resulting in a left chamber end-systolic inside diameter pixel distanceD ES1 。
In particular, the method comprises the steps of,connectionThe center coordinates of the fan-shaped areax F ,y F ) Obtaining a straight lineL u Detection ofL u And (3) withQ u Two intersection points of the edges of the left chamber region +.>And->Calculate->And->The distance between the two is obtained to obtain the inner diameter pixel distance of the end diastole of the left chamberD ED1 . Connection->The center coordinates of the fan-shaped areax F ,y F ) Obtaining a straight lineL v Detection ofL v And (3) withQ v Two intersection points of the edges of the left chamber region +.>And->Calculate->And->The distance between the two is obtained to obtain the inner diameter pixel distance at the end of the left chamber contractionD ES1 。
In this embodiment, the minimum circumscribed circle fitting algorithm used is the minEnclosingCircle () function in the OpenCV toolkit.
Fig. 6 shows a schematic diagram of a fitting method of the ventricular inner diameters of end diastole and end systole frames, wherein,
(x F ,y F )=(428, 106)，=(429, 292)，L u two crossing points with left chamber region->And->Coordinates (429, 231), (429, 348),D ED1 =117 (pixels),. About.>=（429, 284），L v Two crossing points with left chamber region->And->Coordinates (428, 214) and (429, 370), respectively,D ES1 =156 (pixels).
Substep S32, extracting the echocardiogram for which ejection fraction is required to be calculatedGIs set to be equal to or smaller than the pixel pitch of (1)G p The end-diastole left indoor diameter pixel distanceD ED1 And end-systole left room inside diameter pixel distanceD ES1 Respectively with the pixel spacingG p Multiplying to obtain the actual distance of the end diastole inner diameter of the left chamberD ED2 Actual distance from left end-systole inner diameterD ES2 According toD ED2 AndD ES2 the left end diastole volume and the end systole volume are calculated respectively, and then the left room ejection fraction is calculated.
Specifically, the pixel pitchG p Is adjacent to the imageThe actual distance between the pixels (in cm/pixel). In the step, the M ultrasonic is simulated from the B ultrasonic image to obtain the inner diameter of the ventricle, so that the end diastole volume and the end systole volume of the left ventricle are calculated, the ventricular ejection fraction can be accurately measured in real time, and the accuracy and the repeatability of the measurement of the ventricular ejection fraction can be improved. In practice, an echocardiogram is acquired in which ejection fraction is calculatedGIs decom file of (d)G d From the slaveG d Extracting pixel pitchG p . According toD ED2 AndD ES2 the left end diastole volume and the end systole volume are calculated respectively, and the left ejection fraction is calculated according to the left end diastole volume and the end systole volume.
Further, the left end diastole volume, the left end systole volume and the left chamber ejection fraction are determined according to the following formulas (1) - (3), respectively:
（1）
wherein,EDVrepresenting the end-diastole volume of the left chamber,D ED2 representing the actual distance of the left end diastole inner diameter;
（2）
wherein,ESVindicating the end-systolic volume of the left chamber,D ES2 representing the actual distance of the left chamber end-systole inner diameter;
LVEF=[(EDV-ESV)/EDV]×100% （3）
wherein LVEF represents left ventricular ejection fraction.
In the present embodiment of the present invention,G p = 0.0268 (cm/pixel),D ED2 =4.18（cm），D ES2 =3.14（cm），EDV=77.7（mL），ESV=39（mL），LVEF=49.8%。
in summary, the invention directly inputs the B-type ultrasonic image into the image segmentation model to segment the binary image of the left chamber prediction result, namely, the ventricular structure characteristic is obtained, the ventricular contour is extracted, and the contour is analyzed, thereby realizing the visualization of the ventricle, improving the segmentation accuracy of the ultrasonic image and remarkably improving the recognition degree of the left chamber contour image; by adopting a left chamber segmentation technology based on a two-dimensional ultrasonic image, searching an end diastole frame image and an end systole frame image from a B ultrasonic image, simulating M ultrasonic to calculate the end diastole volume and the end systole volume of the left chamber so as to accurately measure the ventricular ejection fraction in real time, on one hand, the calculation complexity is reduced, and the operation time and the operation error of operators are greatly reduced; on the other hand, the accuracy and the repeatability of the left-room ejection fraction measurement can be improved, and the defect of using the M ultrasonic measurement of ejection fraction is effectively overcome.
The invention also provides a left ventricular ejection fraction calculating device based on transesophageal echocardiography, which comprises:
the left room prediction result binary image generation module is used for acquiring an ultrasonic image of a region of interest of an echocardiogram of which the ejection fraction is to be calculated, inputting the ultrasonic image of the region of interest into a pre-trained image segmentation model, and outputting a left room prediction result binary image;
the end diastole frame and end systole frame image determining module is used for generating a left room pixel area change curve according to the binary image of the left room prediction result and positioning an end diastole frame image and an end systole frame image by utilizing the left room pixel area change curve;
the left chamber ejection fraction determining module is used for determining a left chamber end diastole inner diameter physical distance and a left chamber end systole inner diameter physical distance according to the end diastole frame image and the end systole frame image, calculating a left chamber end diastole volume and a left chamber end systole volume according to the left chamber end diastole inner diameter physical distance and the left chamber end systole inner diameter physical distance, and further calculating the left chamber ejection fraction.
The invention also provides a device comprising a processor, a memory and a computer program stored on the memory and capable of running on the processor, which when executed by the processor implements the left ventricular ejection fraction calculation method based on transesophageal echocardiography of any of the above.
The present invention also provides a computer readable storage medium having stored thereon a computer program which when executed by a processor implements a left ventricular ejection fraction calculation method based on transesophageal echocardiography as described in any of the above.
The relevant parts of the embodiments of the apparatus and the device related to the embodiments of the method may be referred to each other, and are not repeated herein.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (12)
1. A method for calculating left ventricular ejection fraction based on transesophageal echocardiography, comprising:
step 1, acquiring an ultrasonic image of a region of interest of an echocardiogram of which the ejection fraction is to be calculated, inputting the ultrasonic image of the region of interest into a pre-trained image segmentation model, and outputting a binary image of a left room prediction result;
step 2, generating a left-room pixel area change curve according to the binary image of the left-room prediction result, and positioning an end diastole frame image and an end systole frame image by utilizing the left-room pixel area change curve;
and 3, determining the left end diastole inner diameter physical distance and the left end systole inner diameter physical distance by using the end diastole frame image and the end systole frame image, calculating the left end diastole volume and the left end systole volume according to the left end diastole inner diameter physical distance and the left end systole inner diameter physical distance, and further calculating to obtain the left chamber ejection fraction.
2. The left ventricular ejection fraction calculation method based on transesophageal echocardiography according to claim 1, wherein in the step 1, an ultrasound image of a region of interest in the transesophageal left ventricular short-axis planar echocardiography is acquired, and the ultrasound image of the region of interest and a corresponding left ventricular binary mask image are used as a sample data set, and a deep neural network model is pre-trained to obtain an image segmentation model.
3. The left ventricular ejection fraction calculation method based on transesophageal echocardiography according to claim 2, wherein the pre-training process of the image segmentation model in step 1 comprises:
acquisition ofMExample echocardiography of the left ventricular short axis section of the stomachE i WhereiniRepresent the firstiIn an example of this,E i IncludedN i frame two-dimensional ultrasound images co-obtainedKFrame two-dimensional original ultrasound imageI j The method comprises the steps of carrying out a first treatment on the surface of the Wherein,M i 、N i are all positive integers;
the two-dimensional original ultrasound image for each frameI j Using a binary image having a sectorB j And (3) withI j Multiplying together to obtainKFrame region of interest ultrasound imageJ j ；
Ultrasound images of the region of interest for each frameJ j Acquisition ofKFrame left room binary mask imageL j ，L j In the left cell region, the pixel value is 255, and the pixels except the left cell region have a value of 0, which willKFrame(s)J j AndKframe(s)L j Respectively reduced to imagesAnd->Will beKFrame->And->As a sample dataset, an image segmentation model is generated.
4. A left ventricular ejection fraction calculation method based on transesophageal echocardiography as claimed in claim 3, wherein said binary imageB j Is respectively the width and the height ofW j AndH j the pixel value of the sector area is 1, the pixel values except the sector area are 0, and the circle center coordinates of the sector area are @x j , y j )，x j =α j W j ，0.1≤α j ≤ 0.9，y j =β j H j ，0≤β j Less than or equal to 0.8, the radius of the sector area isr j ，r j = γ j H j ，0.2≤γ j The angle of the sector area is less than or equal to 0.8θ j ，80°≤θ j ≤130°。
5. The transesophageal echocardiography-based left ventricular ejection fraction calculation method of claim 1, wherein said step 1 comprises:
echocardiography for ejection fraction to be calculatedG，GIncludednFrame two-dimensional ultrasound imageFUsing a binary image having a sectorB F And (3) withFMultiplying together to obtainnFrame region of interest ultrasound imageJ F The method comprises the steps of carrying out a first treatment on the surface of the Wherein, the content of the active ingredients is less than or equal to 40 percentn≤300，FIs respectively the width and the height ofW F AndH F ；
will beJ F Is reduced toLoading the image segmentation model, and taking an echocardiogram +.>A kind of electronic devicenInputting the ultrasonic images of the region of interest into the image segmentation model frame by frame, and outputtingnFrame left room prediction result binary image +.>And will->Processing the left room predicted result binary image into an enlarged left room predicted result binary imageQ，QIs respectively the width and the height ofW Q AndH Q ，W Q =W F , H Q =H F 。
6. the transesophageal echocardiography-based left ventricular ejection fraction calculation method of claim 5, wherein said method comprisesB F Is respectively the width and the height ofW BF AndH BF ，W BF =W F ，H BF =H F , B F wherein the pixel value of the sector area is 1, the pixel values except the sector area are 0, and the center coordinates of the sector area are [ ]x F , y F )，
x F =α F W BF ，0.1≤α F ≤ 0.9，y F =β F H BF ，0 ≤β F Less than or equal to 0.8, the radius of the sector area isr F ，
r F = γ F H BF ，0.2≤γ F The angle of the sector area is less than or equal to 0.8θ F ，80°≤θ F ≤130°。
7. The transesophageal echocardiography-based left ventricular ejection fraction calculation method of claim 5, wherein said step 2 comprises:
for the left bin predictor binary imageQRespectively calculateQThe number of pixel points with the pixel value of 255 in each frame of image to obtain the left room pixel area of each frame of image, and drawing a curve of the left room pixel area changing along with the frame numberSAnd for the left chamber pixel area with the frame number curveSSmoothing to obtain a smooth curveWill smooth curve->One maximum value adjacent to the upper positionS max And a minimum value ofS min As the left chamber pixel area at end diastole and the left chamber pixel area at end systole of the left chamber respectively,S max > S min >0 maximum valueS max Corresponding abscissa isuMinimum value ofS min Corresponding abscissa isv，0 < u ≤n，0 < v ≤n，Q u AndQ v an end diastole frame image and an end systole frame image, respectively.
8. The transesophageal echocardiography-based left ventricular ejection fraction calculation method of claim 7, wherein said step 3 comprises:
respectively determiningQ u AndQ v minimum circumcircle of middle left chamber areaC u AndC v according toC u Center coordinates of (2)The center coordinates of the fan-shaped areax F , y F ) Is connected with the line of (a)L u And (3) withQ u The distance between two intersection points of the edges of the left chamber region, resulting in a left chamber end diastole inside diameter pixel distanceD ED1 ；
According toC v Center coordinates of (2)The center coordinates of the fan-shaped areax F , y F ) Is connected with the line of (a) L v And (3) withQ v The distance between two intersection points of the edges of the left chamber region, resulting in a left chamber end-systolic inside diameter pixel distanceD ES1 ；
Extracting an echocardiogram of the ejection fraction to be calculatedGIs set to be equal to or smaller than the pixel pitch of (1)G p Distance of left end diastole inner diameter pixelD ED1 Distance from left end-systole inner diameter pixelD ES1 Respectively with the pixel spacingG p Multiplying to obtain the actual distance of the end diastole inner diameter of the left chamberD ED2 Actual distance from left end-systole inner diameterD ES2 According toD ED2 AndD ES2 the left end diastole volume and the left end systole volume are calculated respectively, and then the left room ejection fraction is calculated.
9. The transesophageal echocardiography-based left ventricular ejection fraction calculation method of claim 8, wherein the left end diastole volume, left end systole volume and left ventricular ejection fraction are determined according to the following formulas (1) - (3), respectively:
（1）
wherein,EDVrepresenting the end-diastole volume of the left chamber,D ED2 representing the actual distance of the left end diastole inner diameter;
（2）
wherein,ESVindicating the end-systolic volume of the left chamber,D ES2 representing the actual distance of the left chamber end-systole inner diameter;
LVEF=[(EDV-ESV)/EDV]×100% （3）
wherein LVEF represents left ventricular ejection fraction.
10. A transesophageal echocardiography-based left ventricular ejection fraction calculation device, comprising:
the left room prediction result binary image generation module is used for acquiring an ultrasonic image of a region of interest of an echocardiogram of which the ejection fraction is to be calculated, inputting the ultrasonic image of the region of interest into a pre-trained image segmentation model, and outputting a left room prediction result binary image;
the end diastole frame and end systole frame image determining module is used for generating a left room pixel area change curve according to the binary image of the left room prediction result and positioning an end diastole frame image and an end systole frame image by utilizing the left room pixel area change curve;
the left chamber ejection fraction determining module is used for determining a left chamber end diastole inner diameter physical distance and a left chamber end systole inner diameter physical distance according to the end diastole frame image and the end systole frame image, calculating a left chamber end diastole volume and a left chamber end systole volume according to the left chamber end diastole inner diameter physical distance and the left chamber end systole inner diameter physical distance, and further calculating the left chamber ejection fraction.
11. An apparatus comprising a processor, a memory, and a computer program stored on the memory and executable on the processor, the computer program when executed by the processor implementing a transesophageal echocardiography-based left ventricular ejection fraction calculation method according to any one of claims 1 to 9.
12. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements the transesophageal echocardiography-based left ventricular ejection fraction calculation method according to any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410190249.7A CN117745726B (en) | 2024-02-21 | Left ventricular ejection fraction calculating method and device based on transesophageal echocardiography |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410190249.7A CN117745726B (en) | 2024-02-21 | Left ventricular ejection fraction calculating method and device based on transesophageal echocardiography |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117745726A true CN117745726A (en) | 2024-03-22 |
| CN117745726B CN117745726B (en) | 2024-06-07 |
Family
ID=
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112057109A (en) * | 2014-01-24 | 2020-12-11 | 深圳迈瑞生物医疗电子股份有限公司 | Ultrasound monitoring apparatus and method |
| CN112075956A (en) * | 2020-09-02 | 2020-12-15 | 深圳大学 | Deep learning-based ejection fraction estimation method, terminal and storage medium |
| WO2021176237A1 (en) * | 2020-03-06 | 2021-09-10 | Ultromics Limited | Assessing heart parameters using neural networks |
| CN114419499A (en) * | 2022-01-07 | 2022-04-29 | 乐普（北京）医疗器械股份有限公司 | Cardiac ultrasound video-based ejection fraction data processing method and device |
| CN114983472A (en) * | 2022-04-13 | 2022-09-02 | 苏州赫米兹健康科技有限公司 | Method for detecting ejection fraction by dividing left ventricle of echocardiogram |
| CN115526833A (en) * | 2022-08-23 | 2022-12-27 | 上海杏脉信息科技有限公司 | Ultrasonic image-based ejection fraction calculation device, method, medium, and apparatus |
| CN116130090A (en) * | 2023-02-16 | 2023-05-16 | 苏州赫米兹健康科技有限公司 | Ejection fraction measuring method and device, electronic device, and storage medium |
| CN117547306A (en) * | 2024-01-09 | 2024-02-13 | 深圳华声医疗技术股份有限公司 | Left ventricular ejection fraction measurement method, system and device based on M-type ultrasound |
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112057109A (en) * | 2014-01-24 | 2020-12-11 | 深圳迈瑞生物医疗电子股份有限公司 | Ultrasound monitoring apparatus and method |
| WO2021176237A1 (en) * | 2020-03-06 | 2021-09-10 | Ultromics Limited | Assessing heart parameters using neural networks |
| CN112075956A (en) * | 2020-09-02 | 2020-12-15 | 深圳大学 | Deep learning-based ejection fraction estimation method, terminal and storage medium |
| CN114419499A (en) * | 2022-01-07 | 2022-04-29 | 乐普（北京）医疗器械股份有限公司 | Cardiac ultrasound video-based ejection fraction data processing method and device |
| CN114983472A (en) * | 2022-04-13 | 2022-09-02 | 苏州赫米兹健康科技有限公司 | Method for detecting ejection fraction by dividing left ventricle of echocardiogram |
| CN115526833A (en) * | 2022-08-23 | 2022-12-27 | 上海杏脉信息科技有限公司 | Ultrasonic image-based ejection fraction calculation device, method, medium, and apparatus |
| CN116130090A (en) * | 2023-02-16 | 2023-05-16 | 苏州赫米兹健康科技有限公司 | Ejection fraction measuring method and device, electronic device, and storage medium |
| CN117547306A (en) * | 2024-01-09 | 2024-02-13 | 深圳华声医疗技术股份有限公司 | Left ventricular ejection fraction measurement method, system and device based on M-type ultrasound |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210177373A1 (en) | Ultrasound system with an artificial neural network for guided liver imaging | |
| CN110197713B (en) | Medical image processing method, device, equipment and medium | |
| WO2017206023A1 (en) | Cardiac volume identification analysis system and method | |
| US5360006A (en) | Automated method for digital image quantitation | |
| WO2017193251A1 (en) | Method and system for recognizing region of interest profile in ultrasound image | |
| CN101170948B (en) | Ultrasonographic device and image processing method thereof | |
| US20090226058A1 (en) | Method and apparatus for tissue border detection using ultrasonic diagnostic images | |
| US10405834B2 (en) | Surface modeling of a segmented echogenic structure for detection and measurement of anatomical anomalies | |
| US9147258B2 (en) | Methods and systems for segmentation in echocardiography | |
| CN110664435A (en) | Method and device for acquiring cardiac data and ultrasonic imaging equipment | |
| CN101617947A (en) | Cardiac muscle texture automatic analyzing method based on spot tracking technology | |
| CN106570856A (en) | Common carotid artery intima-media thickness measuring device and method combining level set segmentation and dynamic programming | |
| WO1991019457A1 (en) | Automated method for digital image quantitation | |
| CN110310323A (en) | The retinal blood vessels caliber measurement method being fitted based on Hessian matrix and dimensional Gaussian | |
| CN115035028A (en) | Left ventricular ejection fraction automatic calculation method based on ultrasonic image | |
| CN109919953B (en) | Method, system and apparatus for carotid intima-media thickness measurement | |
| WO2022062459A1 (en) | Automatic measurement method and device for fetal structural characteristic | |
| CN117017347B (en) | Image processing method and system of ultrasonic equipment and ultrasonic equipment | |
| CN117745726B (en) | Left ventricular ejection fraction calculating method and device based on transesophageal echocardiography | |
| WO2020007026A1 (en) | Segmentation model training method and apparatus, and computer-readable storage medium | |
| CN117745726A (en) | Left ventricular ejection fraction calculating method and device based on transesophageal echocardiography | |
| CN117115437A (en) | Multi-index multi-organ medical image segmentation model evaluation system based on region | |
| US8287457B2 (en) | 3D echocardiographic shape analysis | |
| CN113222985B (en) | Image processing method, image processing device, computer equipment and medium | |
| CN115775233A (en) | Processing method and device for measuring characteristic dimension based on cardiac ultrasound video |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |