Object Extraction, Measurement Method and Device in Ultrasonic Images

Abstract

The present invention relates to the field of ultrasonic image processing, and specifically provides a method and device for ultrasonic image object extraction and measurement. The method for object extraction in an ultrasonic image includes: obtaining an ultrasonic image containing a target object; performing image segmentation on the ultrasonic image to obtain a segmented image containing multiple feature regions; connecting several of the feature regions on the segmented image that satisfy a preset direction condition according to the direction information of each feature region to obtain a target image containing the complete target object; and extracting the target object in the target image. Through this method, a complete and relatively clear target object can be automatically extracted, making the subsequent measurement of the target object more accurate.

Description

Technical Field

[0001] The present invention relates to the field of ultrasonic image processing, and particularly to a method and device for extracting and measuring ultrasonic image objects. Background Art

[0002] Ultrasonic imaging is an important means for medical diagnosis, especially for emergency department examinations in hospitals. Ultrasonic imaging has been widely used in the medical field due to its advantages of being factual, inexpensive, and non-invasive. However, factors such as signal attenuation, uneven gray-scale distribution, artifacts, and speckle noise can all lead to a low signal-to-noise ratio of ultrasonic images, thus causing many difficulties in the quantitative analysis of ultrasonic images. For example, in the measurement of the length of the fetal humerus or femur by ultrasonic imaging.

[0003] The fetal skeletal system is a routine item in prenatal ultrasonic examinations. The measurement of the length of the fetal humerus or femur has important clinical value for the screening of malformations such as congenital dysplasia, short limbs, and disproportionate skeletal development caused by some chromosomal abnormalities. At the same time, it is also an indispensable parameter for estimating the fetal weight, gestational age, etc. Currently, the measurement of the fetal humerus or femur in clinical diagnosis is mainly carried out manually by an ultrasonic doctor operating a trackball. However, there are problems such as a large amount of speckle noise, artifact interference, and blurred bone edges in ultrasonic images, which will affect the accuracy of the measurement results due to random errors generated by manual measurement and visual errors of clinicians. And repetitive operations will also increase the time cost additionally.

[0004] Therefore, it is of great significance to realize the automatic measurement of target objects in the analysis of ultrasonic images. However, when using traditional image recognition methods to segment the target in ultrasonic images, since the gray-scale distribution of the target object in ultrasonic images is not completely uniform, the same target object will be split into two or more regions after segmentation, which brings difficulties to subsequent screening and calculation, resulting in a complex process and difficulty in realizing target extraction in ultrasonic images. Summary of the Invention

[0005] To solve the technical problems of complexity and difficulty in realizing target extraction in ultrasonic images by existing image recognition methods, the present invention provides a method for extracting objects in ultrasonic images that can accurately extract target objects on ultrasonic images.

[0006] At the same time, to solve the technical problem of low accuracy in manually measuring target objects on ultrasonic images in the prior art, the present invention provides a method for measuring objects in ultrasonic images that can perform automatic measurement and has a more accurate measurement result.

[0007] In a first aspect, the present invention provides a method for extracting objects in ultrasonic images, including:

[0008] Obtaining an ultrasonic image containing a target object;

[0009] Perform image segmentation on the ultrasonic image to obtain a segmented image including a plurality of feature regions, and at least one of the plurality of feature regions corresponds to the target object;

[0010] Connect the feature regions on the segmented image that meet the preset direction condition according to the direction information of the feature regions to obtain a target image including the complete target object;

[0011] Extract the target object from the target image.

[0012] In some embodiments, the performing image segmentation on the ultrasonic image to obtain a segmented image including a plurality of feature regions includes:

[0013] Perform binarization processing on the ultrasonic image to obtain a binary image including a plurality of feature regions.

[0014] In some embodiments, the connecting the feature regions on the segmented image that meet the preset direction condition according to the direction information of the feature regions includes:

[0015] Obtain the endpoints of the feature regions;

[0016] Judge whether other endpoints are included in the preset region pointed to by one endpoint,

[0017] When other endpoints are included in the preset region pointed to by the one endpoint, extract the other endpoint that is not in the same feature region as the one endpoint from the other endpoints, and the difference between the direction value of the other endpoint in the ultrasonic image direction field and the direction value of the one endpoint in the ultrasonic image direction field is within a preset range;

[0018] Connect the one endpoint and the other endpoint.

[0019] In some embodiments, the connecting the one endpoint and the other endpoint includes:

[0020] Perform pixel filling between the one endpoint and the other endpoint to obtain a connection region.

[0021] In some embodiments, the obtaining the endpoints of the feature regions includes:

[0022] Perform thinning processing on the plurality of feature regions on the binary image to obtain the endpoints of the thinned plurality of feature regions;

[0023] After connecting the one endpoint and the other endpoint, it further includes:

[0024] Perform dilation processing on the connection region, and superimpose the dilated connection region on the binary image.

[0025] In some embodiments, the image segmentation of the ultrasound image includes:

[0026] Performing filtering processing on the ultrasound image, and performing image segmentation on the filtered ultrasound image.

[0027] In some embodiments, between the image segmentation of the ultrasound image and connecting the feature regions on the segmented image according to the direction information of the feature regions, it further includes:

[0028] Filtering the segmented image based on the characteristics of the target object.

[0029] In some embodiments, the target object includes a fetal humerus and / or femur.

[0030] In a second aspect, the present invention provides a method for measuring an object in an ultrasound image, including:

[0031] Obtaining a target object on the ultrasound image, where the target object is obtained by using the above-mentioned method for extracting an object in an ultrasound image;

[0032] Measuring parameters of the target object.

[0033] In some embodiments, the measuring the parameters of the target object includes:

[0034] Performing linear fitting on the target object;

[0035] Calculating the length of the line segment between the two intersection points of the line and the target object.

[0036] In a third aspect, the present invention provides an apparatus for extracting an object in an ultrasound image, including:

[0037] An image acquisition module, configured to acquire an ultrasound image including a target object;

[0038] An image segmentation module, configured to perform image segmentation on the ultrasound image to obtain a segmented image including a plurality of feature regions, and at least one of the plurality of feature regions corresponds to the target object;

[0039] A connection module, configured to connect the feature regions on the segmented image according to the direction information of the feature regions to obtain a target image including the complete target object; and

[0040] An extraction module, configured to extract the target object from the target image.

[0041] In a fourth aspect, the present invention provides an apparatus for measuring an object in an ultrasound image, including:

[0042] An acquisition module, configured to acquire a target object on an ultrasonic image, where the target object is obtained by using the ultrasonic image object extraction method described above; and

[0043] A measurement module, configured to measure parameters of the target object.

[0044] In a fifth aspect, the present invention provides a medical device, including:

[0045] A processor; and

[0046] A memory, communicatively connected to the processor, storing computer-readable instructions executable by the processor. When the computer-readable instructions are executed, the processor executes the object extraction method in the ultrasonic image described above.

[0047] The object extraction method in the ultrasonic image provided by the present invention includes acquiring an ultrasonic image of a target object, performing image segmentation on the ultrasonic image to obtain a segmented image including a plurality of feature regions, where the feature regions include a fractured region corresponding to the target object and other interference regions. These fractured regions cause difficulties in subsequent screening and calculation. In the solution of the present invention, these feature regions are connected according to the direction information of these feature regions to obtain a target image including a complete target object, and then the target object is extracted from the target image. Through this method, a complete target object can be automatically extracted, making the subsequent measurement of the target object simpler and more accurate. Description of the Drawings

[0048] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0049] Figure 1 is a schematic diagram of an ultrasonic image according to some embodiments of the present invention;

[0050] Figure 2 is a schematic diagram of an ultrasonic image object extraction method according to an embodiment of the present invention;

[0051] Figure 3 is a schematic diagram of binary processing of an ultrasonic fetal humerus / femur image;

[0052] Figure 4 is a schematic diagram of connecting several feature regions according to some embodiments of the present invention;

[0053] Figure 5 is a schematic diagram of an ultrasonic image object measurement method according to some embodiments of the present invention;

[0054] Figure 6 is a schematic diagram of measuring a target object according to an embodiment of the present invention;

[0055] Figure 7 is a schematic diagram of the structure of an ultrasonic image object extraction device according to some embodiments of the present invention;

[0056] Figure 8 is a schematic diagram of the structure of an ultrasonic image object measurement device according to some embodiments of the present invention;

[0057] Figure 9 is a schematic diagram of a computer system structure suitable for implementing the method or processor in the embodiments of the present invention. Detailed Embodiments

[0058] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention. In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0059] The method for extracting an object in an ultrasonic image provided by the present invention can be used to extract a target object on an ultrasonic image in medical diagnosis. It should be noted that there are many problems such as uneven gray-scale distribution, artifacts, and speckle noise on the ultrasonic image. Therefore, the signal-to-noise ratio of the ultrasonic image is low. As a result, it is difficult for an operator to manually measure the target object of the ultrasonic image, leading to inaccurate measurement. At the same time, due to the above problems of the ultrasonic image, a large number of non-target regions will be generated when using the existing image recognition method to segment the ultrasonic image, which brings difficulties to target extraction.

[0060] More importantly, after segmenting the ultrasound image using traditional image recognition, segmentation breaks of the target object will occur on the segmented image. The reason for the breaks is that the gray-scale distribution of the target object in the ultrasound image is not completely uniform, and gray-scale values may be relatively low at different positions of the same target object. In this case, image recognition is likely to identify such positions as belonging to the background area, so that the same target object is segmented into two or more different sub-regions after segmentation. In subsequent calculation processes, theoretically, these broken sub-regions all belong to the target object. Therefore, these broken sub-regions will increase the complexity and inaccuracy of subsequent calculations. Moreover, due to the damage to the integrity of the target object, it cannot be well selected in subsequent screening. Therefore, it is difficult to extract the target from the ultrasound image using the existing image recognition method. The method for extracting an object from an ultrasound image provided by the present invention is precisely based on connecting the broken sub-regions generated after segmenting the ultrasound image, making the target extraction simpler and more accurate, and facilitating subsequent measurement and calculation.

[0061] Figure 1 The method for extracting an object from an ultrasound image in some embodiments of the present invention is shown. As Figure 1 shown, in some embodiments, the method for extracting an object from an ultrasound image includes:

[0062] S1. Obtain an ultrasound image containing the target object.

[0063] S2. Perform image segmentation on the ultrasound image to obtain a segmented image containing multiple feature regions.

[0064] S3. Connect several feature regions on the segmented image that meet the preset direction condition according to the direction information of each feature region to obtain a target image containing the complete target object.

[0065] S4. Extract the target object from the target image.

[0066] Specifically, in step S1, an ultrasound image containing the complete target object is obtained to facilitate the integrity of the final object extraction. The target object can be any target with bright features, such as a bone region, and the present invention places no restrictions on this.

[0067] In step S2, based on the above ultrasound image, image segmentation is performed on the ultrasound image to obtain a segmented image. The segmented image includes multiple feature regions. It should be noted that the feature regions can be regions corresponding to the target object, regions corresponding to non-target objects, or multiple sub-regions formed by the breakage of these regions during segmentation, etc. For example, for an ultrasound bone image, due to uneven image gray-scale, the same bone region may break into two or more feature regions after segmentation. At the same time, the bright regions on the ultrasound image that are not bones may also break into multiple feature regions during segmentation.

[0068] In step S3, based on the obtained segmented image above, the feature regions on the segmented image are connected according to the direction information of each feature region in the direction field of the segmented image. The direction information of the feature region refers to the directivity of the feature region in the direction field of the ultrasonic image. For example, the main direction of each feature region can be extracted, or the directions of the points (such as end points) within each feature region can be extracted, and the feature regions that meet the preset direction condition are connected. The preset direction condition can be, for example, when the direction difference between two feature regions is within the threshold range, then the two feature regions are connected, so as to finally obtain several connected feature regions. Among them, the several feature regions can be one, or two or more, as long as the direction information of the several feature regions meets the preset direction condition, they can be connected.

[0069] For example, taking the above ultrasonic bone image as an example, during the connection process, according to the direction information of each feature region, it is judged whether the main direction of each feature region is within the threshold range. It is considered that several feature regions within the threshold range are due to the lower brightness around the inside of the bone, resulting in their being segmented into several feature regions, that is, broken bones. Therefore, it is necessary to connect each broken feature region that meets the threshold range, so as to obtain the target image containing the complete bone feature region.

[0070] In step S4, on the connected target image above, the target object is extracted. Taking the above ultrasonic bone region as an example, the complete bone feature region is extracted on the target image, so as to obtain the target object. When the method of the present invention extracts the target from the ultrasonic image, the feature regions of the broken target object on the segmented image are connected, and the complete target object can be extracted, so that the target object can be recognized more completely and clearly in the ultrasonic image, making the subsequent measurement of various parameters more accurate.

[0071] In Figure 2 shows a method for extracting an object in an ultrasonic image according to some embodiments of the present invention. In an exemplary embodiment, for the sake of convenience of description, the ultrasonic image is taken as an ultrasonic fetal humerus / femur image. It should be noted that the method provided by the present invention is not limited to extracting the fetal humerus / femur, and any other applicable target object can use the present invention, such as ultrasonic images of bones in other positions, etc.

[0072] As Figure 2 shown, in some embodiments, the method of the present invention includes:

[0073] S10. Obtain an ultrasonic image containing a fetal humerus / femur.

[0074] In the medical diagnosis of the fetal humerus / femur, the length parameter of the humerus / femur is generally required. Therefore, in the acquisition of ultrasonic images, relatively complete bone images are generally used, so that the images contain the complete humerus / femur region.

[0075] S11. Perform filtering processing on the ultrasonic image.

[0076] Since there is a lot of noise in the ultrasonic image, in some embodiments, filtering processing is performed before segmenting the ultrasonic image. In an exemplary embodiment, anisotropic filtering can be used for the filtering processing. The main idea of the anisotropic filter can overcome the defects of the Gaussian blur filter, that is, it does not destroy the gray relative contrast information in the bone edge region and can retain the image edge while smoothing the image. This method is very suitable for the filtering processing of ultrasonic images.

[0077] Anisotropy regards the image as a heat field and each pixel as a heat flow. According to the relationship between the neighboring pixels and the current pixel, the degree of diffusion to the surrounding is determined. When the neighboring pixels are quite different from the current pixel, it means that the current pixel may be located on the boundary, so the heat flow is stopped from diffusing in this direction from the current pixel, and this boundary is retained; at the same time, if the difference is small, then the diffusion can continue.

[0078] The main iterative equation of anisotropy is as follows:

[0079]

[0080] Among them, x and y respectively represent the horizontal and vertical coordinates of the image, λ can take values in [0, 1 / 4], where 1 / 4 represents the average of the diffusion degrees in four directions. I represents the ultrasonic image. Since the above formula is an iterative formula, there is the current iteration number t. The four divergence formulas are to take the partial derivatives of the current pixel in four directions:

[0081]

[0082]

[0083]

[0084]

[0085] cN, cS, cE, cW represent the diffusion coefficients in four directions, and the diffusion coefficient at the boundary is relatively small. The formula for the diffusion coefficient c is as follows:

[0086]

[0087] Through the anisotropic filtering process, the ultrasound image can be smoother and the image noise can be removed to a certain extent. It should be noted that in this embodiment, the diffusion coefficient method is used, which is only used to explain this step and is not used to limit the present invention. Those skilled in the art can also use other anisotropic processing according to specific circumstances.

[0088] S20, performing binarization processing on the ultrasonic image to obtain a binary image containing multiple feature areas.

[0089] The ultrasonic image after filtering in step S11 is segmented. In some embodiments, the image segmentation adopts binarization processing. The binarization processing can convert the original detection of the target object into simple shape detection, feature detection, etc., thereby simplifying the calculation.

[0090] In an exemplary embodiment, the mean-shift method is used to segment the image. The goal of this segmentation method is to calculate the class label of each pixel. The class label depends on the cluster to which it belongs. For each cluster, it must have a class center. The center point gathers some points around it so that they can form a class, that is, those points with the same class center are in one class. Mean-shift uses the extreme point of its density function as the center point in the iterative process, called the model point. During the iterative process, the model point will gather toward the true center of the class to which it belongs.

[0091] The formula for the kernel density function is:

[0092]

[0093] K(x) can have various function forms, such as exponential kernel function:

[0094]

[0095] It should be noted that the mean-shift method given in this embodiment is only used as a preferred embodiment. Due to the differences in physical characteristics and parameter settings of machines and probes from different manufacturers, the segmentation method here can be selected according to the specific situation. The method used to extract the binary image can be flexibly selected according to the specific characteristics of the target image and the specific situation. For example, OTSU, maximum entropy threshold segmentation method, cluster segmentation method, level set method and other methods can be used to segment the image. Through this step, a binary image containing the fetal humerus / femur can be obtained.

[0096] S21. Based on the target object features, the segmented image is screened to obtain a screened segmented image.

[0097] In the above embodiments, due to the influence of speckle noise on the ultrasound image, there are often many interference regions in the binary image obtained after image segmentation. Moreover, due to the uneven gray-scale distribution on the ultrasound image, the characteristic regions of the binary image are often disconnected into two or more characteristic regions. Therefore, before connecting multiple characteristic regions, the binary image is screened first to remove the influence of some interference regions and simplify the subsequent calculations.

[0098] In an exemplary embodiment, a certain degree of screening is performed through the characteristic parameter of roundness, and its formula is:

[0099]

[0100] where L represents the perimeter of the connected region, and S represents the area of the region to be screened. Generally speaking, A = 1 corresponds to the case of a circle, while the bone region of the target object generally shows a slender feature. Therefore, in this embodiment, the regions with this parameter greater than the threshold T A are removed, and some interference regions can be screened out. In this embodiment, the threshold can be taken as T A = 0.75.

[0101] S30. Connect the characteristic regions on the segmented image according to the direction information of the characteristic regions to obtain a target image containing the complete target object.

[0102] For the sake of convenience of explanation, in an example, as Figure 3 shown, Figure 3 (a) is an ultrasound image containing the fetal humerus. The bright region in the middle of the image is the manifestation feature of the fetal humerus. It can be seen from the figure that there is a lot of noise on the ultrasound image, and at the same time, the gray-scale distribution in the middle of the humerus region is uneven. Therefore, on the binary image obtained after binary processing, the fetal humerus region is broken into two or more characteristic regions. At the same time, the non-humerus characteristic regions above also appear as multiple broken characteristic regions on the binary image, and there are also a lot of interference regions on the binary image. Figure 3 (b) is the binary image after feature screening in step S21. It can be seen from Figure 3 (b) that a lot of interference regions showing non-humerus characteristics are screened out on the screened binary image, but there are still non-humerus characteristic regions whose manifestations are relatively close to the humerus characteristics. At the same time, these non-humerus characteristic regions and the humerus characteristic regions all appear as being broken into multiple characteristic regions. Therefore, in this step, these broken characteristic regions are connected to form several complete and clearly bounded target regions.

[0103] S40. Extract the characteristic region corresponding to the target object from the connected several characteristic regions.

[0104] Based on the above-connected image, screening is performed to extract the fetal humerus / femur feature region. In an exemplary implementation, the screening method can select each feature region by constructing a cost function. The features used in this implementation are the major axis, average region brightness, roundness, etc. of each feature region. That is, the cost function of the i-th feature region is:

[0105] F(i) = c1f1(i) + c2f2(i) + … + c d f d (i) (9)

[0106] where d is the total number of features, c1, c2, …, c d are the weights of each feature parameter, and there are:

[0107]

[0108] In this implementation, Then, the feature region with the maximum cost function is the target object sought.

[0109] In Figure 2 , Figure 3 In some of the illustrated embodiments, taking the ultrasonic fetal humerus / femur image as an example, the ultrasonic image object extraction method of the present invention is described. In this method, by connecting the fractured fetal humerus / femur regions on the segmented image, a complete fetal humerus / femur target is obtained. In Figure 4 shows a schematic diagram of the method for connecting feature regions in some embodiments of the present invention.

[0110] As Figure 4 shown, connecting each feature region on the binary image includes:

[0111] S301. Refine each feature region on the binary image. Binary image refinement, also known as skeletonization, refers to reducing each feature region on the binary image to a unit pixel width to facilitate finding the region endpoints.

[0112] S302. Obtain the endpoints of the refined structure. Based on the above, the binary image includes feature regions of the fetal humerus / femur and non-fetal humerus / femur feature regions. Therefore, after binary image refinement, several curve segments are formed, and each curve segment has two endpoints. Obtain the endpoints of these line segments.

[0113] S303. Determine whether other endpoints are included in the preset region pointed to by any endpoint.

[0114] The purpose of this step is to confirm whether the positions of the endpoints to be connected are adjacent. When there are no other endpoints within the preset area pointed to by one endpoint, it can be considered that this endpoint is the endpoint of a certain feature area and no connection is required. When other endpoints are included within the preset area pointed to by one endpoint, proceed to step S304. In an exemplary implementation, the preset area pointed to by one endpoint can be the area with a radius of r around this endpoint, and r can take values from 3 to 8 pixels. Those skilled in the art should understand that the preset area pointed to by the endpoint can also be other shapes and threshold ranges, which will not be elaborated here.

[0115] S304. Extract another endpoint from the other endpoints that is not in the same feature area as one endpoint. The difference between the direction value of the other endpoint in the ultrasonic image direction field and the direction value of one endpoint in the ultrasonic image direction field is within a preset range. When other endpoints are detected within the preset area of one endpoint, further judge the other endpoint with a direction approximately the same as that of one endpoint, so as to confirm that one endpoint and the other endpoint need to be connected. When judging the endpoints, it can be judged based on the direction field of the image. Since the humerus / femur characteristics of the fetus are manifested as slender and straight shapes, the difference between the direction values of the two endpoints can be set within a relatively small range, such as 0 to 30°, or even a smaller range to improve the judgment accuracy.

[0116] S305. Perform pixel filling between one endpoint and the other endpoint to obtain a connection area. Based on the judgment of the above steps, obtain the two endpoints that need to be connected, and perform pixel filling between the two endpoints to obtain a connection line segment, that is, a connection area.

[0117] S306. Perform dilation processing on the connection area and superimpose the dilated connection area on the binary image. Dilate the connection line segment obtained in step S305 to obtain a dilated area, and superimpose the dilated area on the binary image in, for example, Figure 3 (b), and the connection of the fractured feature area can be completed.

[0118] Through the above implementation method, the ultrasonic image is binarized to obtain a binary image, and then the fractured feature areas on the binary image are connected to obtain a complete fetal humerus / femur target, and then a target object with a complete and relatively clear area is extracted from the ultrasonic image, which is convenient for subsequent measurement.

[0119] In the second aspect, the present invention also provides a method for measuring an object in an ultrasonic image. As Figure 5 shown, in some embodiments, the measurement method includes:

[0120] S50. Obtain a target object on the ultrasonic image, and the target object is obtained by using the object extraction method in any of the above embodiments of the ultrasonic image.

[0121] S60. Measure the parameters of the target object.

[0122] This method can be used to automatically measure the target object extracted in the above embodiments, avoiding the random errors and repetitive labor of manual measurement.

[0123] In an exemplary embodiment, taking the above ultrasonic fetal humerus / femur image as an example, the length of the fetal humerus / femur is measured. As Figure 6 shown, the measurement method includes:

[0124] S500. Obtain the fetal humerus / femur region by using the object extraction method in the ultrasonic image in the above embodiment.

[0125] S601. Perform linear fitting on the fetal humerus / femur region. The fitting method can adopt the least squares method, Hough line fitting, etc., and the present invention does not limit this.

[0126] In an exemplary embodiment, the least squares method is adopted for the fitting method. The least squares method calculates by minimizing the residuals of the fitted line or curve. Suppose there are n pairs of points to be fitted, and the fitted line equation is set as:

[0127] y = b0 + b1x (11)

[0128] Then the residual formula for line fitting is:

[0129]

[0130] Thus, the residual i The sum of squares is:

[0131]

[0132] By minimizing Q, that is, taking the extreme value is equivalent to finding its partial derivatives with respect to the parameters b0 and b1 and setting the partial derivatives equal to 0. At this time, substituting the obtained parameters b0 and b1 into the line equation (11), the line equation can be determined. If the curve fitting method needs to be adopted, it is a similar calculation idea. The difference is that there may be more parameters for the curve. By listing the curve expression equation, constructing the Q value in the way of formula (13), and respectively finding the parameters when taking the extreme value, the line equation where the line segment is located can be determined. The two intersection points of the line and the fetal humerus / femur region are the two end points of the line segment.

[0133] S602. Calculate the pixel unit length of the line segment. Calculate the pixel unit length of the two end points of the above line segment.

[0134] S603. Convert the above pixel unit length to the physical unit length, and the length parameter of the fetal humerus / femur can be known.

[0135] In a third aspect, the present invention provides an object extraction device in an ultrasonic image, as Figure 7 shown. The device may include:

[0136] An image acquisition module 10 for acquiring an ultrasonic image including a target object;

[0137] An image segmentation module 20 for performing image segmentation on the ultrasonic image to obtain a segmented image including a plurality of feature regions, and at least one of the plurality of feature regions corresponding to the target object;

[0138] A connection module 30 for connecting the feature regions on the segmented image according to the direction information of the feature regions to obtain a target image including a complete target object; and

[0139] An extraction module 40 for extracting the target object in the target image.

[0140] In a fourth aspect, the present invention provides an object measurement device in an ultrasonic image, as Figure 8 shown. The device may include:

[0141] An acquisition module 50 for acquiring a target object on an ultrasonic image, the target object being obtained by using the ultrasonic image object extraction method of the above embodiment; and

[0142] A measurement module 60 for measuring parameters of the target object.

[0143] In a fifth aspect, the present invention provides a medical device, which may be an ultrasonic device, such as an ultrasonic diagnostic instrument, and includes:

[0144] A processor; and

[0145] A memory communicably connected to the processor and storing computer-readable instructions executable by the processor. When the computer-readable instructions are executed, the processor executes the object extraction and / or measurement method in the ultrasonic image of the above embodiment.

[0146] In a sixth aspect, the present invention provides a storage medium storing computer instructions for causing a computer to execute the object extraction and / or measurement method in the ultrasonic image described above.

[0147] Specifically, Figure 9 shows a schematic structural diagram of a computer system 600 suitable for implementing the method or processor of the embodiments of the present invention. Through Figure 9 the system shown, the corresponding functions of the medical device and the storage medium are realized.

[0148] As Figure 9As shown, computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 602 or programs loaded from a storage section 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the system 600 are also stored. The CPU 601, ROM 602, and RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0149] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is mounted on the drive 610 as needed so that a computer program read therefrom can be installed into the storage section 608 as needed.

[0150] In particular, according to an embodiment of the present disclosure, the process described above with reference to Figure 1 can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product that includes a computer program tangibly embodied on a machine-readable medium, the computer program including program code for performing Figure 1 the method. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 609, and / or installed from the removable medium 611.

[0151] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that, in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combinations of blocks in the block diagram and / or flowchart, may be implemented by a dedicated hardware-based system that performs the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0152] Obviously, the above-described embodiments are merely examples for clear illustration and not limitations on the embodiments. For those of ordinary skill in the art, other different forms of changes or variations can be made based on the above description. It is not necessary and impossible to enumerate all the embodiments here. The obvious changes or variations derived therefrom are still within the protection scope of the present invention.

Claims

1. A method for extracting an object in an ultrasonic image, characterized in that, Comprising: Obtain an ultrasonic image including a target object; Perform image segmentation on the ultrasonic image to obtain a segmented image including a plurality of feature regions; According to the direction information of each of the feature regions, connect several of the feature regions on the segmented image that meet a preset direction condition to obtain a target image including the complete target object, where the preset direction condition is that when the direction difference between two feature regions is within a threshold range, connect the two feature regions; Extract the target object in the target image; Wherein, the performing image segmentation on the ultrasonic image to obtain a segmented image including a plurality of feature regions includes: Perform binarization processing on the ultrasonic image to obtain a binary image including a plurality of feature regions; The connecting several of the feature regions on the segmented image that meet the preset direction condition according to the direction information of each of the feature regions includes: Obtain the end points of the feature region; Judge whether there are other end points in a preset region pointed to by one end point, When there are other end points in the preset region pointed to by the one end point, extract, from the other end points, another end point that is not in the same feature region as the one end point, and the difference between the direction value of the another end point in the ultrasonic image direction field and the direction value of the one end point in the ultrasonic image direction field is within a preset range; Connect the one end point and the another end point.

2. The method for object extraction in an ultrasonic image according to claim 1, wherein The connecting the one end point and the another end point includes: Perform pixel filling between the one end point and the another end point to obtain a connection region.

3. The method for extracting an object in an ultrasonic image according to claim 2, wherein, The obtaining the end points of the feature region includes: Perform thinning processing on several of the feature regions on the binary image to obtain the end points of the thinned several of the feature regions; After the connecting the one end point and the another end point, further include: Perform dilation processing on the connection region, and superimpose the dilated connection region on the binary image.

4. The method for extracting an object in an ultrasonic image according to claim 1, wherein The performing image segmentation on the ultrasonic image includes: Perform filtering processing on the ultrasonic image, and perform image segmentation on the filtered ultrasonic image.

5. The method for extracting an object in an ultrasonic image according to claim 1, wherein Between the performing image segmentation on the ultrasonic image and connecting the feature regions on the segmented image according to the direction information of the feature regions, further include: Perform screening on the segmented image based on the features of the target object.

6. The method for extracting an object in an ultrasonic image according to claim 1, wherein The target object includes a fetal humerus and / or femur.

7. A method for measuring an object in an ultrasonic image, characterized in that, Comprising: Obtain a target object on an ultrasonic image, where the target object is obtained by using the method for extracting an object in an ultrasonic image according to any one of claims 1 to 6; Measure the parameters of the target object.

8. The method for measuring an object in an ultrasonic image according to claim 7, wherein The measuring the parameters of the target object includes: Perform linear fitting on the target object; Calculate the length of the line segment between the two intersection points of the line and the target object.

9. An object extraction device in an ultrasonic image, characterized in that, Comprising: An image acquisition module, configured to obtain an ultrasonic image including a target object; An image segmentation module, configured to perform image segmentation on the ultrasonic image to obtain a segmented image including a plurality of feature regions; A connection module, configured to connect a plurality of the feature regions on the segmented image that meet a preset direction condition according to the direction information of each of the feature regions, so as to obtain a target image including the complete target object, where the preset direction condition is that when the direction difference between two feature regions is within a threshold range, the two feature regions are connected; and an extraction module, configured to extract the target object from the target image; wherein, the performing image segmentation on the ultrasonic image to obtain a segmented image including a plurality of feature regions includes: performing binarization processing on the ultrasonic image to obtain a binary image including a plurality of feature regions; the connecting a plurality of the feature regions on the segmented image that meet the preset direction condition according to the direction information of each of the feature regions includes: obtaining endpoints of the feature regions; judging whether other endpoints are included in a preset region pointed to by one endpoint; when other endpoints are included in the preset region pointed to by the one endpoint, extracting, from the other endpoints, another endpoint that is not in the same feature region as the one endpoint, where the difference between the direction value of the another endpoint in the ultrasonic image direction field and the direction value of the one endpoint in the ultrasonic image direction field is within a preset range; connecting the one endpoint and the another endpoint.

10. An object measurement device in an ultrasonic image, characterized in that, including: an acquisition module, configured to acquire a target object on an ultrasonic image, where the target object is obtained by using the method for extracting an object in an ultrasonic image according to any one of claims 1 to 6; and a measurement module, configured to measure parameters of the target object.

11. A medical device, characterized in that, including: a processor; and a memory, communicably connected to the processor, storing computer-readable instructions that can be executed by the processor, and when the computer-readable instructions are executed, the processor executes the method for extracting an object in an ultrasonic image according to any one of claims 1 to 6.

Images