Medical index measurement method and ultrasonic diagnostic apparatus therefor

Abstract

Disclosed are a medical index measurement method and an ultrasonic diagnostic apparatus therefor. According to one embodiment, the medical index measurement method includes the steps of: an ultrasonic diagnostic apparatus acquiring ultrasonic image data; extracting a feature point from the acquired ultrasonic image data; generating map information configured according to the extracted feature point; determining an anatomical structure according to the generated map information; and measuring a medical index based on the determined anatomical structure.

Description

Technical Field

[0001] This invention relates to diagnostic techniques using ultrasound, and more particularly, to techniques for measuring medical indicators using ultrasound.

[0002] This invention was supported by the National Research and Development Program, with authorization number 1425140041, project number S2492471, department name: Small and Medium Enterprises and Entrepreneurship Department, research management institution: Korea Advanced Institute of Technology, research project name: WC300 R&D, research title: Development of Software Beamforming Ultrasound Diagnostic Apparatus Line Up, and research supervision institution: AVINUS Medical Systems Co., Ltd. Background Technology

[0003] Various imaging devices used to image information about human tissues are used in many medical fields for the early diagnosis of various diseases or surgical procedures. Some examples of such medical imaging devices are ultrasound diagnostic devices, computed tomography (CT) devices, and magnetic resonance imaging (MRI) devices.

[0004] An ultrasound diagnostic device emits ultrasound signals generated by an ultrasound probe towards a subject and receives information about the echo signals reflected from the subject, thereby obtaining images of internal parts of the subject. Specifically, ultrasound diagnostic devices are used for medical purposes, including observing internal areas of a subject, detecting foreign bodies, and assessing damage. Compared to X-ray devices, ultrasound diagnostic devices offer high stability, real-time image display, and are safe due to the absence of radiation exposure. Therefore, ultrasound diagnostic devices are widely used along with other types of imaging diagnostic devices.

[0005] The following methods are widely used: transmitting ultrasound signals to an object using an ultrasound probe, measuring the size of body parts in ultrasound images obtained using ultrasound echoes reflected from the object, and predicting medical indicators. Semantic segmentation is one method for measuring the size of body parts. This technique segments objects in an image into meaningful units. Summary of the Invention

[0006] [Technical Issues]

[0007] According to the implementation method, a medical indicator measurement method and an ultrasound diagnostic device based on the method are proposed, which can measure medical indicators more accurately, easily and automatically.

[0008] [Technical Solutions]

[0009] A method for measuring medical indicators according to an embodiment includes the following steps: acquiring ultrasound image data using an ultrasound diagnostic device; extracting feature points from the acquired ultrasound image data; generating image information configured according to the extracted feature points; determining anatomical structures based on the generated image information; and measuring medical indicators based on the determined anatomical structures.

[0010] In the step of extracting feature points, a greater number of points than the minimum required to construct the graph can be extracted.

[0011] In the step of extracting feature points, points on the circumference of an ellipse can be extracted, points on the circumference of a circle can be extracted, the two endpoints of a line segment can be extracted, and each vertex of a polygon can be extracted.

[0012] Feature point extraction may include the following steps: learning original image data and image data of interest extracted from the original image data; extracting first feature points of the object of interest from the original image data; transforming the original image data using the extracted first feature points to generate image data of interest; extracting final feature points from the generated image data of interest; and performing an inverse transformation on the coordinates of the final feature points to match the original image data.

[0013] In the learning process, the convolutional layers and nonlinear functions of the neural network can be stacked repeatedly to facilitate learning.

[0014] In the learning process, information loss due to reducing the output size can be prevented by using padding in the neural network.

[0015] In the learning process, learning can be performed using a network structure that connects multiple resolution streams in parallel, so as to repeatedly exchange resolution information between the first resolution frame and the second resolution frame.

[0016] In the step of generating graph information, a zero-dimensional graph composed of points, a one-dimensional graph composed of lines or curves, or a two-dimensional graph composed of ellipses, circles, or polygons can be generated.

[0017] In the step of generating graph information, basic graph information and candidate graph information that can replace the basic graph information can be generated.

[0018] In the step of generating graph information, scores can be assigned to the extracted feature points, and feature points with scores higher than or equal to a preset reference can be used to generate graph information.

[0019] In the step of generating graph information, the geometric properties of the graph to be generated can be used to configure the relationships between feature points, and the relationships between the feature points can be used for correction to remove outliers.

[0020] In the step of generating graph information, the graph information can be corrected by obtaining at least one of the measurement results of another fetus in the twins from pre-stored and accumulated fetal biometric information and the measurement value of another current parameter from a fully measured sample.

[0021] In the step of generating graph information, feature points can be selected to correct the graph information by means of user input of operation signals for the generated graph information.

[0022] In the step of determining anatomical structures, the head can be determined based on an ellipse, the abdomen can be determined based on a circle, and the femur or humerus can be determined based on a line segment or a quadrilateral. In the step of measuring medical indicators, the head size can be measured using the circumference point on the ellipse, the abdominal circumference can be measured using the circumference point on the circle, and the length of the femur or humerus can be measured using the two endpoints forming the line segment or the four corners forming the quadrilateral.

[0023] In the process of measuring medical indicators, medical indicators can be measured by taking into account fetal biometric data related to the patient's pregnancy information from pre-stored and accumulated fetal biometric data.

[0024] The medical indicator measurement method may also include at least one of the following steps: displaying the generated graphical information as recognizable visual information on ultrasound image data; and displaying the measured medical indicator as digital information on a graph of ultrasound image data.

[0025] An ultrasound diagnostic device according to another embodiment may include: an ultrasound probe configured to emit ultrasound signals toward a subject and receive reflected wave signals from the subject; an image processing unit configured to generate ultrasound image data using the reflected wave signals from the ultrasound probe; a point extraction unit configured to extract feature points from the generated ultrasound image data; an image generation unit configured to generate image information configured based on the extracted feature points; a structure determination unit configured to determine anatomical structures based on the generated image information; an index measurement unit configured to measure medical indicators based on the determined anatomical structures; and an output unit configured to output the medical indicator measurement results.

[0026] [Beneficial Effects]

[0027] According to the medical indicator measurement method based on the implementation method and the ultrasound diagnostic equipment based on the method, medical indicators can be measured more accurately, easily and automatically compared with semantic segmentation technology. Attached Figure Description

[0028] Figure 1 This is a diagram illustrating the configuration of an ultrasound diagnostic device according to an embodiment of the present invention;

[0029] Figure 2 This is a flowchart illustrating a method for measuring medical indicators using an ultrasound diagnostic device according to an embodiment of the present invention;

[0030] Figure 3 This is a diagram illustrating an example of extracting feature points from ultrasound image data according to an embodiment of the present invention;

[0031] Figure 4 The illustration shows a screen displayed when measuring the size of a fetal head using key point detection technology according to an embodiment of the present invention;

[0032] Figure 5 The illustration shows a screen displayed when measuring fetal abdominal circumference using key point detection technology according to an embodiment of the present invention; and

[0033] Figure 6 The illustration shows a screen displayed when measuring the length of a fetal skeleton using a key point detection technique according to an embodiment of the present invention. Detailed Implementation

[0034] The advantages and features of the invention, as well as the ways in which these advantages and features are realized, will become apparent from the following detailed description of the embodiments in conjunction with the accompanying drawings. However, the invention can be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. These embodiments are provided so that this disclosure will be thorough and complete, and these embodiments fully convey the scope of the invention to those skilled in the art, and the invention is defined only by the scope of the appended claims. Throughout this disclosure, the same reference numerals refer to the same parts.

[0035] In the following description of embodiments of the invention, detailed descriptions of relevant known functions or configurations will be omitted if it is determined that such detailed descriptions would unnecessarily obscure the gist of the invention. The terms described below are defined in consideration of the functions in the embodiments of the invention, and these terms may vary according to the intent or habit of the user or operator. Therefore, the definitions of the terms used herein should follow the context disclosed herein.

[0036] Each block of the attached block diagram and each step of the attached flowchart can be executed by computer program instructions (execution engine), and these computer program instructions can be embedded in the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device. Therefore, these computer program instructions, executed by the processor of the computer or other programmable data processing device, produce tools for performing the functions described in each block of the block diagram or each step of the flowchart.

[0037] These computer program instructions can also be stored in computer-usable or readable memory that is oriented toward a computer or other programmable data processing device to implement functions in a particular manner. Thus, computer program instructions stored in computer-usable or readable memory can produce an article of art containing instructions for performing the functions described in each block of a block diagram or each step of a flowchart.

[0038] Furthermore, computer program instructions may also be mounted on a computer or other programmable data processing equipment. Therefore, computer program instructions for operating a computer or other programmable data processing equipment to produce computer-implemented processes by performing a series of operational steps on the computer or other programmable data processing equipment may also provide steps for performing the functions described in each block of the block diagram and each step of the flowchart.

[0039] Furthermore, each block or step may represent a module, segment, or portion of code comprising one or more executable instructions for performing a specified logical function, and it should be noted that in some alternative implementations, the functions described in the blocks or steps may occur out of order. For example, two blocks or steps shown consecutively may actually be executed substantially simultaneously, and the two blocks or steps may also be executed in the reverse order of their respective functions as needed.

[0040] In the following, embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the present invention can be implemented in various forms, and the scope of the present invention is not limited to these embodiments. Embodiments of the present invention are provided to assist those skilled in the art in explaining and understanding the present invention.

[0041] Figure 1 This is a diagram illustrating the configuration of an ultrasound diagnostic device according to an embodiment of the present invention.

[0042] Reference Figure 1 The ultrasound diagnostic device includes an ultrasound probe 102, a main body 100, an input unit 104, an output unit 106, and a learning unit 108. The ultrasound probe 102, the input unit 104, and the output unit 106 are communicatively connected to the main body 100.

[0043] The ultrasound probe 102 includes multiple piezoelectric resonators, which generate ultrasound waves based on a drive signal provided from the transmitting / receiving unit 110 of the body 100. Furthermore, the ultrasound probe 102 receives reflected waves from the object and converts these reflected waves into electrical signals. That is, the ultrasound probe 102 emits ultrasound waves towards the object and receives reflected waves from the object. The ultrasound probe 102 is detachably connected to the body 100. The object can be a fetus of a pregnant woman, but is not limited to this, making the object any patient.

[0044] The input unit 104 is implemented by: a trackball, mouse, keyboard, touchpad for setting predetermined positions (e.g., the position of tissue shape, region of interest, region other than the region of interest, etc.), a screen for input operations via touch operation surface, a screen integrating the display and touchpad, a contactless input circuit using an optical sensor, an audio input circuit, etc. The input unit 104 is connected to the control unit 130 described below, converts the input operation signal received from the user into an electrical signal, and outputs the electrical signal to the control unit 130.

[0045] Output unit 106 displays a graphical user interface (GUI) for the user of the ultrasound diagnostic device to input various setting requests using input unit 104, or to display ultrasound image data generated in the main body 100, etc. Additionally, output unit 106 displays various messages or information to notify the user of the processing status or results of the main body 100. Furthermore, output unit 106 may include a speaker and output audio. Output unit 106 can display medical indicator measurement results generated by control unit 140 on a screen. Medical indicators may be, for example, fetal indicators. Fetal indicators include the fetal abdomen, head, femur, humerus, etc. However, medical indicators are not limited to fetal indicators and can be any body organ of the object. For example, medical indicators may be the pelvis, etc.

[0046] The output unit 106 can distinguishably display the graphic information generated by the control unit 130 as recognizable visual information (e.g., color) on the ultrasound image data. The output unit 106 can also display the measured medical indicators as digital information on the ultrasound image data.

[0047] Medical indicators measured by the control unit 130 can be displayed as digital information on ultrasound image data.

[0048] The learning unit 108 uses a neural network to learn from the ultrasound image data. In this case, at least one of the components constituting the control unit 130—namely, the point extraction unit 131, the image generation unit 132, the structure determination unit 133, and the index measurement unit 134—can use the learning data. The learning unit 108 may be in the form of a server connected to the main body 100 via a network, or it may be located within the main body 100. The learning unit 108 can learn the raw image data and the image data of interest extracted from the raw image data.

[0049] The main body 100 is a device for generating ultrasound image data based on reflected wave signals received by the ultrasound probe 102. According to an embodiment, the main body 100 includes a transmitting / receiving unit 110, an image processing unit 120, a control unit 130, a storage unit 140, and a learning data receiving unit 150.

[0050] The transmitting / receiving unit 110 provides a drive signal to the ultrasonic probe 102 and repeatedly generates pulses to form transmitted ultrasonic waves at a predetermined frequency. Furthermore, the transmitting / receiving unit 110 converges the ultrasonic waves generated from the ultrasonic probe 102 using beamforming. The transmitting / receiving unit 110 may include a preamplifier, an analog-to-digital (A / D) converter, a receive delay unit, a summing unit, etc., and can generate reflected wave data by performing various processing on the echo signal received by the ultrasonic probe 102. Various forms can be selected as the output signal from the transmitting / receiving unit 110, such as a signal including phase information—referred to as a radio frequency (RF) signal—or a signal including amplitude information after envelope detection.

[0051] Image processing unit 102 receives reflected wave data from transmitting / receiving unit 110 and generates ultrasound image data about the fetus. In this case, image processing unit 120 performs logarithmic amplification, envelope detection processing, etc., to generate ultrasound image data.

[0052] Storage unit 140 stores image data generated by image processing unit 120 or control unit 130.

[0053] The controller 130 controls each component of the ultrasound diagnostic device. For example, the control unit 130 controls the processing of the transmitting / receiving unit 110 and the image processing unit 120 based on various setting requests input by the user through the input unit 130 or various control programs and data read from the storage unit 140. In addition, the control unit 130 controls the display of ultrasound image data stored in the storage unit 140 through the output unit 106.

[0054] The control unit 130 according to the embodiment includes a point extraction unit 131, a graph generation unit 132, a structure determination unit 133, and an index measurement unit 134.

[0055] The point extraction unit 131 extracts feature points from the ultrasound image data generated by the image processing unit 120. These feature points may include feature points defining boundaries, feature points indicating the overall shape, differences in pixel brightness values, etc. For example, points on the circumference of an ellipse, points on the circumference of a circle, the two endpoints of a line segment, and each vertex of a polygon may be extracted. Polygons may include triangles, quadrilaterals, pentagons, etc. In the case of an ellipse, the focal point may also be extracted, and in the case of a circle, the center point may also be extracted to improve accuracy.

[0056] The graph generation unit 132 generates graph information configured using feature points extracted by the point extraction unit 131. For example, a zero-dimensional graph composed of points, a one-dimensional graph composed of lines or curves, or a two-dimensional graph composed of ellipses, circles, or polygons.

[0057] The structure determination unit 133 determines the anatomical structure based on the graph information generated by the graph generation unit 132. For example, the anatomical structure may include the fetal abdomen, head, and skeleton. Examples of skeletons may include the femur and humerus. When determining the anatomical structure, for example, the fetal abdomen may be determined based on a circle, the fetal head based on an ellipse, and the fetal skeleton based on line segments or quadrilaterals.

[0058] The indicator measurement unit 134 measures medical indicators based on the anatomical structures determined by the structure determination unit 133. For example, it measures the head circumference based on the circumference of an ellipse, the abdominal circumference based on the circumference of a circle, and the femur length based on the length of a line segment or quadrilateral.

[0059] The following is referred to as key point detection technology: the control unit 130 extracts points to form an image, uses the extracted feature points to generate an image, and then measures fetal indicators based on the generated image. The following will refer to... Figure 2 This describes a more detailed implementation of the keypoint detection technology.

[0060] The learning data receiving unit 150 receives learning data from the learning unit 108 and sends the training data to the control unit 130. For example, the raw image data learned by the learning unit 108 and the image data of interest can be provided to the point extraction unit 131 of the control unit 130.

[0061] Figure 2 This is a flowchart illustrating a method for measuring medical indicators using an ultrasound diagnostic device according to an embodiment of the present invention.

[0062] Reference Figure 2 Ultrasound diagnostic equipment acquires ultrasound image data of the fetus (210).

[0063] The ultrasound diagnostic device then extracts feature points (220) from the acquired ultrasound image data. The feature points may include feature points defining boundaries, feature points showing the overall shape, differences in pixel brightness values, etc.

[0064] The step (220) of extracting feature points may include learning, extracting a first feature point, performing a transformation, and extracting a second feature point. For example, an ultrasound diagnostic device learns raw image data and image data of interest extracted from the raw image data through a training step. The raw image data is the original image captured by the device, and the image data of interest is an image obtained by extracting (e.g., cropping) the object of interest from the raw image data, identifying the region of interest within it. The ultrasound diagnostic device generates the raw image data and the image data of interest based on the labels of the image data and learns from all the generated image data.

[0065] Before feature point extraction, object locations can be searched from the raw image data, and the image can be cropped based on the found object locations. Accuracy is significantly improved if feature points are extracted by first searching for the object locations and then inputting raw image data cropped based on those object locations. The image size can be resized to the network's input size (e.g., 256×256).

[0066] Subsequently, the ultrasound diagnostic device extracts the first feature points of the object of interest from the raw image data through a step of extracting the first feature points.

[0067] Then, the ultrasound diagnostic device transforms the original image data using the extracted first feature points through a transformation step to generate image data of interest. In this case, the transformation may include rotation, distortion, magnification, etc.

[0068] Subsequently, the ultrasound diagnostic device extracts the final feature points from the generated image data of interest and performs an inverse transformation on the coordinates of the final feature points to match the original image data.

[0069] Based on the above processing, learning can augment the data. Error detection can be corrected through learning, extracting the first feature point, and extracting the second feature point. Target data standardization can be achieved through transformations such as image distortion. Pixel accuracy can be improved through transformations such as image magnification.

[0070] In the step (220) of extracting feature points, a greater number of points than the minimum required to construct the map can be extracted. This is to construct a map even when the object in the ultrasound image is incomplete and therefore key points cannot be detected. Therefore, more points than the information constituting the map are extracted. For example, in the case of an ellipse, although four points are needed to form the ellipse, 16 points can be extracted to improve accuracy.

[0071] In the case of an ellipse, the circumference of the ellipse is defined as the outline of the object. Five or more points (e.g., 16 points) are extracted along the circumference of the ellipse. In the case of a rectangle, four points corresponding to the corners are extracted from the boundary of the rectangle containing the object. However, when the object is curved, the rectangles at both ends of the object, rather than the rectangle containing the entire object, can be defined as valid rectangles, and the points corresponding to these valid rectangles can be extracted. Furthermore, points corresponding to the lower part of the actual object and the four vertices can be further extracted.

[0072] In the step (220) of extracting feature points, the ultrasound diagnostic device may extract points on the circumference of the circle forming an ellipse (additionally, the focus), extract points on the circumference of the circle forming a circle (additionally, the center point), extract the two endpoints of the line segment forming a line segment, and extract each vertex of the polygon forming a polygon. The polygon may include triangles, quadrilaterals, pentagons, etc.

[0073] Then, the ultrasound diagnostic device generates image information (230) configured using the extracted feature points.

[0074] In the step (230) of generating graph information, the ultrasound diagnostic device can generate a zero-dimensional graph composed of points, a one-dimensional graph composed of lines or curves, or a two-dimensional graph composed of ellipses, circles or polygons.

[0075] In step (230) of generating image information, the ultrasound diagnostic device can generate basic image information and candidate image information that can replace the basic image information. For example, an image consisting of the two upper endpoints of the leg bones is set as the basic image information, and an image consisting of the two lower endpoints of the leg bones and of similar length is set as the candidate image information.

[0076] In the step (230) of generating image information, the ultrasound diagnostic device can assign scores to the extracted feature points and use feature points with scores higher than or equal to a preset reference to generate image information.

[0077] In the step (230) of generating the graph information, the ultrasound diagnostic device may further include a step of correction based on the features of the graph to be created when generating the graph information using the extracted feature points. Since the extracted feature points are components of a graph, the relationships between the feature points can be configured using the geometric properties of the graph to be created, and outliers can be pre-removed based on these relationships. For example, in the case of a circle, outliers can be removed using the distance relationship between the center point and each feature point. In the case of an ellipse, outliers can be removed using the distance information of the foci, minor axis, and major axis. Since the distance d from each point forming the ellipse to the center point is limited to a / 2 ≤ d ≤ b / 2 (where a is the minor axis and b is the major axis), points that do not satisfy this relationship can be identified as outliers and removed.

[0078] In step (230) of generating the graph information, the graph information can be corrected by obtaining at least one of the measurement results of another fetus in a twin set of pre-stored and accumulated fetal biometric information and the measurement value of another current parameter that has been fully measured. In another example, the graph information can be corrected by selecting points using user input of an operation signal for the generated graph information.

[0079] In the step (230) of generating image information, the ultrasound diagnostic device can distinguishably display the generated image information as recognizable visual information on ultrasound image data.

[0080] Subsequently, the ultrasound diagnostic device determines the anatomical structure (240) based on the generated image information, and measures medical indicators (250) based on the determined anatomical structure.

[0081] In the step (250) of measuring medical indicators, the ultrasound diagnostic device may use a point on the circumference of an ellipse (additionally, the focal point) to measure the size of the fetal head, a point on the circumference of a circle (additionally, the center point) to measure the circumference of the fetal abdomen, and the two endpoints of a line segment or the four corners of a quadrilateral to measure the length of the fetal skeleton.

[0082] In the step (250) of measuring medical indicators, the ultrasound diagnostic device can measure medical indicators by considering fetal biometric data related to the patient's pregnancy information from pre-stored and accumulated fetal biometric data. The pregnancy information includes the patient's last menstrual period date (LMP) or date of conception (DOC).

[0083] Furthermore, the ultrasound diagnostic device can use a neural network to learn from ultrasound image data. In this case, the learned data can be used for at least one of the steps of extracting feature points (220), generating image information (230), determining anatomical structures (240), and measuring medical indicators (250).

[0084] In the learning process, ultrasound diagnostic equipment can learn by using a structure in which convolutional layers and nonlinear activation functions (ReLU) of a neural network are repeatedly stacked on top of each other.

[0085] During the learning process, the ultrasound diagnostic device can use neural network padding to prevent information loss due to reduced output size and to consistently maintain the resolution of the original layers.

[0086] In the learning process, the ultrasound diagnostic device can learn through a network structure that connects multiple resolution streams in parallel, repeatedly exchanging resolution information between a first resolution frame and a second resolution frame. The first resolution frame can be a high-resolution frame, while the second resolution frame can be a low-resolution frame. In this case, resolution information can be repeatedly exchanged between the high-resolution frame and the low-resolution frame.

[0087] Ultrasound diagnostic equipment can display measured medical indicators as digital information on an ultrasound image graph.

[0088] Figure 3 This is a diagram illustrating an example of extracting feature points from ultrasound image data according to an embodiment of the present invention.

[0089] Reference Figure 3 Ultrasound diagnostic equipment can extract points on the circumference of an ellipse to measure the size of the fetal head. Figure 3 In the study, 16 points forming an ellipse were extracted. Additionally, ultrasound diagnostic equipment can extract points on the circumference of a circle to measure the fetal abdominal circumference. Figure 3 In this example, 16 points forming the circle were extracted. In another example, the two endpoints forming the line segment can be extracted to measure the length of the bone.

[0090] Figure 4 The illustration shows a screen displayed when measuring the size of a fetal head using key point detection technology according to an embodiment of the present invention.

[0091] Reference Figure 4 Ultrasound diagnostic equipment extracts points on the circumference of an ellipse from ultrasound image data, determines the head based on the ellipse, and uses the points on the circumference of the ellipse (additionally, the focal points) to measure the size of the fetal head.

[0092] At this point, the circumference of the ellipse, as well as the lengths of its horizontal and vertical axes, can be distinguishably displayed as identifiable visual information (e.g., color). Additionally, measurements of the fetal abdominal circumference, including the lengths of the horizontal and vertical axes, can be displayed on the horizontal and vertical axes.

[0093] Figure 5 The illustration shows a screen displayed when measuring fetal abdominal circumference using key point detection technology according to an embodiment of the present invention.

[0094] Reference Figure 5 The system extracts points on the circumference of a circle from ultrasound image data, determines the fetal abdomen based on this circle, and uses the circumference points and center point of the circle to measure the fetal abdominal circumference.

[0095] In this configuration, the central circumference can be distinguishably displayed as identifiable visual information (e.g., color). Additionally, measurements of the fetal abdominal circumference can be displayed on this circle.

[0096] Figure 6 The illustration shows a screen displayed when measuring the length of a fetal skeleton using a key point detection technique according to an embodiment of the present invention.

[0097] Reference Figure 6 The two endpoints that form a line segment are extracted from ultrasound image data, and the length of the fetal skeleton is measured using the two endpoints that form the line segment.

[0098] In this case, the two endpoints of the line segment and the line segment connecting the two endpoints can be distinguishably displayed as identifiable visual information (e.g., color). Additionally, measurements of the fetal skeleton can be displayed on the skeleton.

[0099] The invention has now been described focusing on exemplary embodiments. Those skilled in the art will understand that the invention can be implemented in modified forms without departing from its essential characteristics. Therefore, the disclosed embodiments should be considered illustrative rather than definitive. The scope of the invention is defined by the appended claims rather than by the foregoing description, and all differences within the scope of its equivalents should be interpreted as included within the invention.

Claims

1. A method for measuring medical indicators performed by an ultrasound diagnostic device, comprising the following steps: Acquire ultrasound image data; A neural network is used to extract feature points from the acquired ultrasound image data for generating image information, where, The feature points include at least one of feature points defining boundaries in ultrasound image data and feature points indicating the overall shape; Graph information is generated by connecting the extracted feature points into a geometric graph; The anatomical structure is determined based on the generated image information; as well as Medical indicators are measured based on the determined anatomical structures. The medical indicator is a geometric measurement of a geometric figure corresponding to the determined anatomical structure, and includes either the length or the perimeter of the geometric figure.

2. The medical indicator measurement method according to claim 1, wherein, In the step of extracting the feature points, a greater number of points than the minimum number required to construct the graph are extracted.

3. The medical indicator measurement method according to claim 1, wherein, In the step of extracting the feature points, points on the circumference of the circle forming the ellipse are extracted, points on the circumference of the circle forming the circle are extracted, the two endpoints of the line segment are extracted, and each vertex of the polygon is extracted.

4. The medical indicator measurement method according to claim 1, wherein, Extracting the feature points includes the following steps: Learn the original image data and the image data of interest extracted from the original image data; Extract the first feature points of the object of interest from the original image data; Image data of interest is generated by transforming the original image data using the extracted first feature points; Extract the final feature points from the generated image data of interest; as well as The coordinates of the final feature points are inversely transformed to match the original image data.

5. The medical indicator measurement method according to claim 4, wherein, In the learning step, the learning is performed using a structure in which convolutional layers and nonlinear functions of a neural network are repeatedly stacked on top of each other.

6. The medical indicator measurement method according to claim 4, wherein, In the learning process, information loss due to reduced output size is prevented by using padding in the neural network.

7. The medical indicator measurement method according to claim 4, wherein, In the learning step, the learning is performed through a network structure that connects multiple resolution streams in parallel, so as to repeatedly exchange resolution information between the first resolution frame and the second resolution frame.

8. The medical indicator measurement method according to claim 1, wherein, In the step of generating the graph information, a zero-dimensional graph composed of points, a one-dimensional graph composed of lines or curves, or a two-dimensional graph composed of ellipses, circles, or polygons is generated.

9. The medical indicator measurement method according to claim 1, wherein, In the step of generating the graph information, basic graph information and candidate graph information that can replace the basic graph information are generated.

10. The medical indicator measurement method according to claim 1, wherein, In the step of generating the graph information, scores are assigned to the extracted feature points, and feature points with scores higher than or equal to a preset reference are used to generate the graph information.

11. The medical indicator measurement method according to claim 1, wherein, In the step of generating the graph information, the geometric properties of the graph to be generated are used to configure the relationships between the feature points, and the relationships between the feature points are used for correction to remove outliers.

12. The medical indicator measurement method according to claim 1, wherein, In the step of generating the graph information, the graph information is corrected by obtaining at least one of the measurement results of another fetus in the twins from pre-stored and accumulated fetal biometric information and the measurement value of another current parameter from a fully measured data set.

13. The medical indicator measurement method according to claim 1, wherein, In the step of generating the graph information, the feature points are selected to correct the graph information by means of user input of operation signals for the generated graph information.

14. The medical indicator measurement method according to claim 1, wherein, In the step of determining the anatomical structures, the head is determined based on an ellipse, the abdomen based on a circle, and the femur or humerus based on a line segment or quadrilateral. In the step of measuring the medical indicators, the head size is measured using the circumference points on the ellipse, the abdominal circumference is measured using the circumference points on the circle, and the length of the femur or humerus is measured using the two endpoints forming the line segment or the four corners forming the quadrilateral.

15. The medical indicator measurement method according to claim 1, wherein, In the step of measuring the medical indicator, the medical indicator is measured by taking into account fetal biometric data related to the patient's pregnancy information from pre-stored and accumulated fetal biometric data.

16. The medical indicator measurement method according to claim 1, further comprising at least one of the following steps: The generated image information can be distinguishably displayed as identifiable visual information on the ultrasound image data; and The measured medical indicators are displayed as digital information on a graph of the ultrasound image data.

17. An ultrasound diagnostic device, comprising: An ultrasonic probe configured to emit ultrasonic signals toward an object and receive reflected wave signals from the object; An image processing unit is configured to generate ultrasound image data using the reflected wave signal from the ultrasound probe; A point extraction unit is configured to use a neural network to extract feature points from the generated ultrasound image data for generating graph information, wherein the feature points include at least one of feature points defining boundaries in the ultrasound image data and feature points indicating the overall shape. The graph generation unit is configured to generate graph information by connecting the extracted feature points into a geometric graph; A structure-determining unit is configured to determine anatomical structures based on generated map information; An index measurement unit is configured to measure medical indicators based on a determined anatomical structure, wherein the medical indicator is a geometric measurement of a geometry corresponding to the determined anatomical structure, and includes either the length or the perimeter of the geometry; and The output unit is configured to output the results of medical indicator measurements.

Images