A needle tip position recognition method for thyroid fine needle aspiration ultrasound images

Abstract

The application relates to the technical field of image analysis, in particular to a needle tip position recognition method for thyroid fine needle puncture ultrasound images, which comprises the following steps: acquiring each frame of target ultrasound image in a current puncture process, and determining a local difference characteristic value corresponding to each pixel point; determining a feature state variation index corresponding to each pixel point; screening a target variation region from each frame of target ultrasound image, and performing edge detection on the target variation region; screening a target fine needle region from each frame of target ultrasound image; determining a fine needle motion state corresponding to each frame of target ultrasound image according to the motion change condition between the target fine needle regions in adjacent frames of target ultrasound image; and recognizing the needle tip position in each frame of target ultrasound image based on the fine needle motion state corresponding to each frame of target ultrasound image. The application realizes the recognition of the needle tip position by analyzing the target ultrasound image, and improves the accuracy of the needle tip position recognition.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image analysis, and particularly relates to a needle tip position recognition method for thyroid fine needle puncture ultrasound images. BACKGROUND

[0002] With the development of science and technology, image recognition technology is applied more and more widely, for example, it can be applied to the needle tip position recognition of thyroid fine needle puncture ultrasound images. At present, when an object is recognized, the method usually adopted is: the position of the object is recognized from the image according to the difference in gray value.

[0003] However, when the needle tip position of the thyroid fine needle puncture ultrasound image is recognized according to the difference in gray value, the following technical problems often exist:

[0004] The gray value of the needle tip in the ultrasound image is often similar to that of the calcification and fibrous septum and other high echo structures, so when the needle tip position of the thyroid fine needle puncture ultrasound image is recognized, if only the difference in gray value is considered, the needle tip pixel point may be misjudged, thereby causing poor accuracy of the needle tip position recognition. SUMMARY

[0005] In order to solve the technical problem of poor accuracy of the needle tip position recognition, the present application provides a needle tip position recognition method for thyroid fine needle puncture ultrasound images.

[0006] In a first aspect, the present application provides a needle tip position recognition method for thyroid fine needle puncture ultrasound images, which comprises:

[0007] acquiring each frame of target ultrasound image in the current puncture process, and determining a local difference feature value corresponding to each pixel point based on the gray value change in the preset neighborhood corresponding to each pixel point in each frame of target ultrasound image;

[0008] determining a feature state change index corresponding to each pixel point according to the difference between the local difference feature value corresponding to each pixel point and the local difference feature value corresponding to the corresponding pixel point in the previous frame of target ultrasound image;

[0009] screening a target change region from each frame of target ultrasound image according to the feature state change index corresponding to the pixel point in each frame of target ultrasound image, and performing edge detection on the target change region to obtain a target edge contour;

[0010] screening a target fine needle region from each frame of target ultrasound image according to the shape feature of all target edge contours in each frame of target ultrasound image;

[0011] Determine a needle motion state corresponding to each of the target ultrasound images according to motion changes between target needle regions in adjacent target ultrasound images;

[0012] Identify a needle tip position in each of the target ultrasound images based on the needle motion state corresponding to each of the target ultrasound images.

[0013] In combination with the first aspect, in a possible implementation manner, the determining of the local difference feature value corresponding to each of the pixel points in each of the target ultrasound images based on the gray scale changes between the pixel points in the preset neighborhood corresponding to each of the pixel points comprises:

[0014] Determine a local gray scale deviation factor corresponding to each of the pixel points in each of the target ultrasound images according to gray scale differences between each of the pixel points and each of the pixel points in the preset neighborhood corresponding to each of the pixel points.

[0015] Normalize a mean value of the local gray scale deviation factors corresponding to all the pixel points in the preset neighborhood corresponding to each of the pixel points in each of the target ultrasound images to obtain the local difference feature value corresponding to each of the pixel points in each of the target ultrasound images.

[0016] In combination with the first aspect, in a possible implementation manner, the determining of the local gray scale deviation factor corresponding to each of the pixel points in each of the target ultrasound images according to the gray scale differences between each of the pixel points and each of the pixel points in the preset neighborhood corresponding to each of the pixel points comprises:

[0017] Determine any one of the target ultrasound images as a marker ultrasound image, and determine any one of the pixel points in the marker ultrasound image as a marker pixel point;

[0018] Determine a difference absolute value between a gray scale value corresponding to the marker pixel point and a gray scale value corresponding to each of the pixel points in the preset neighborhood corresponding to the marker pixel point as a gray scale difference to obtain a gray scale difference set corresponding to the marker pixel point;

[0019] Determine a maximum gray scale difference in the gray scale difference set corresponding to the marker pixel point as a gray scale difference peak value corresponding to the marker pixel point;

[0020] Determine a mean value of all the gray scale differences in the gray scale difference set corresponding to the marker pixel point as a gray scale difference mean value corresponding to the marker pixel point;

[0021] Determine a product of the gray scale difference peak value and the gray scale difference mean value corresponding to the marker pixel point as a local gray scale deviation factor corresponding to the marker pixel point.

[0022] In a possible implementation manner of the first aspect, the feature state variation indicator corresponding to each pixel point is determined according to a difference between the local difference feature value corresponding to each pixel point and a local difference feature value corresponding to a pixel point in a previous frame of the target ultrasound image corresponding to the pixel point.

[0023] The absolute value of the difference between the local difference feature value corresponding to each pixel point in each frame of the target ultrasound image and the local difference feature value corresponding to the pixel point in the previous frame of the target ultrasound image corresponding to the pixel point is determined as an initial state variation factor corresponding to each pixel point in each frame of the target ultrasound image.

[0024] The feature state variation indicator corresponding to each pixel point in each frame of the target ultrasound image is determined according to the initial state variation factor corresponding to each pixel point in each frame of the target ultrasound image and the local gray scale deviation factor, and the initial state variation factor and the local gray scale deviation factor are positively correlated with the feature state variation indicator.

[0025] In a possible implementation manner of the first aspect, the target variation region is filtered out from each frame of the target ultrasound image according to the feature state variation indicator corresponding to each pixel point in each frame of the target ultrasound image, and the target variation region includes the following steps.

[0026] The pixel point with the feature state variation indicator greater than a preset variation threshold value is filtered out from each frame of the target ultrasound image as a variation pixel point.

[0027] The minimum circumscribed rectangle of a region formed by all the variation pixel points in each frame of the target ultrasound image is determined as the target variation region.

[0028] In a possible implementation manner of the first aspect, the target fine needle region is filtered out from each frame of the target ultrasound image according to the shape feature condition of all the target edge contours in each frame of the target ultrasound image, and the target fine needle region includes the following steps.

[0029] The local contour direction vector corresponding to each edge pixel point on each target edge contour is determined according to each edge pixel point and a neighboring edge pixel point of the edge pixel point.

[0030] The absolute value of the cosine similarity between the local contour direction vector corresponding to each edge pixel point on each target edge contour and the local contour direction vector corresponding to any one of the neighboring edge pixel points of the edge pixel point is determined as a direction change similarity corresponding to each edge pixel point on each target edge contour.

[0031] The fine needle shape possibility indicator corresponding to each target edge contour is determined according to a difference between the direction change similarities corresponding to all the neighboring edge pixel points on each target edge contour.

[0032] Screening the target edge contour corresponding to the maximum needle shape possibility index from each target ultrasound image as a needle candidate contour;

[0033] Performing region growing on the edge pixel point on the needle candidate contour in each target ultrasound image as an initial seed point to obtain a target sub-region;

[0034] Determining the union of the internal region of the needle candidate contour and the target sub-region to which the needle candidate contour belongs as a target needle region.

[0035] In combination with the first aspect, in a possible implementation manner, the determining, according to each edge pixel point on each target edge contour and its adjacent edge pixel points, of a local contour direction vector corresponding to each edge pixel point on each target edge contour comprises:

[0036] Determining any one target edge contour as a marker edge contour, and determining any one edge pixel point on the marker edge contour as a marker edge point;

[0037] Determining any one of the two edge pixel points adjacent to the marker edge point on the marker edge contour as a reference edge point;

[0038] Constructing a local contour direction vector corresponding to the marker edge point according to the marker edge point and the reference edge point, wherein a modulus of the local contour direction vector corresponding to the marker edge point is equal to a distance between the marker edge point and the reference edge point, and a direction of the local contour direction vector corresponding to the marker edge point is a direction in which the marker edge point points to the reference edge point.

[0039] In combination with the first aspect, in a possible implementation manner, the determining, according to a difference between direction change similarities corresponding to all adjacent edge pixel points on each target edge contour, of a needle shape possibility index corresponding to each target edge contour comprises:

[0040] Determining an absolute value of a difference between the direction change similarities corresponding to each adjacent edge pixel point on each target edge contour as a change similarity difference, to obtain a change similarity difference set corresponding to each target edge contour;

[0041] Determining a mean value of all change similarity differences in the change similarity difference set corresponding to each target edge contour as an overall similarity difference corresponding to each target edge contour;

[0042] Determining a needle shape possibility index corresponding to each target edge contour according to the overall similarity difference corresponding to each target edge contour, wherein the overall similarity difference and the needle shape possibility index are in a negative correlation relationship.

[0043] With reference to the first aspect, in a possible implementation manner, the determining of the needle motion state corresponding to each target ultrasound image according to motion changes between target needle regions in adjacent target ultrasound images comprises:

[0044] selecting, from the target needle region in each target ultrasound image, a pixel point corresponding to the largest feature state variation index as a needle tip candidate point;

[0045] selecting, from a preset neighborhood of the needle tip candidate point in each target ultrasound image, a pixel point with the largest gray scale difference with the needle tip candidate point as a reference point corresponding to the needle tip candidate point in each target ultrasound image;

[0046] constructing a reference direction vector corresponding to each target ultrasound image according to the needle tip candidate point and the reference point corresponding to the needle tip candidate point in each target ultrasound image, wherein a modulus of the reference direction vector is equal to a distance between the needle tip candidate point and the reference point corresponding to the needle tip candidate point, and a direction of the reference direction vector is a direction in which the needle tip candidate point points to the reference point corresponding to the needle tip candidate point;

[0047] if a cosine similarity between reference direction vectors corresponding to adjacent target ultrasound images is greater than a preset similarity threshold, determining that the needle motion states corresponding to the adjacent target ultrasound images are the same;

[0048] constructing a target ultrasound image group by using continuous target ultrasound images with the same needle motion state to obtain a target ultrasound image group sequence;

[0049] setting the needle motion states corresponding to all target ultrasound images in a target ultrasound image group with an odd sequence number in the target ultrasound image group sequence as a needle penetration state;

[0050] setting the needle motion states corresponding to all target ultrasound images in a target ultrasound image group with an even sequence number in the target ultrasound image group sequence as a needle withdrawal state.

[0051] With reference to the first aspect, in a possible implementation manner, the identifying of the needle tip position in each target ultrasound image based on the needle motion state corresponding to each target ultrasound image comprises:

[0052] if the needle motion state corresponding to the target ultrasound image is the needle penetration state, determining a position in which the needle tip candidate point is located in the target ultrasound image as the needle tip position in the target ultrasound image;

[0053] if the needle motion state corresponding to the target ultrasound image is the needle withdrawal state, determining a position in which a target needle region to which the needle tip candidate point belongs is located in the target ultrasound image as a position to which the needle tip belongs in the target ultrasound image.

[0054] In a second aspect, the present application provides a needle tip position recognition system for thyroid fine needle puncture ultrasound images, the system comprising:

[0055] The acquisition and determination module is configured to acquire each frame of target ultrasound image in the current puncture process, and determine a local difference feature value corresponding to each pixel point based on a gray scale change in a preset neighborhood corresponding to the pixel point in each frame of target ultrasound image.

[0056] The feature state variation index determination module is configured to determine a feature state variation index corresponding to each pixel point according to a difference between the local difference feature value corresponding to each pixel point and a local difference feature value corresponding to a corresponding pixel point in a previous frame of target ultrasound image.

[0057] The screening and edge detection module is configured to screen a target variation region from each frame of target ultrasound image according to the feature state variation index corresponding to the pixel point in each frame of target ultrasound image, and perform edge detection on the target variation region to obtain a target edge contour.

[0058] The region screening module is configured to screen a target fine needle region from each frame of target ultrasound image according to a shape feature of all target edge contours in each frame of target ultrasound image.

[0059] The fine needle motion state determination module is configured to determine a fine needle motion state corresponding to each frame of target ultrasound image according to a motion change between target fine needle regions in adjacent frames of target ultrasound image.

[0060] The needle tip position recognition module is configured to recognize a needle tip position in each frame of target ultrasound image based on the fine needle motion state corresponding to each frame of target ultrasound image.

[0061] In a third aspect, a server is provided, comprising a memory and a processor. The memory is configured to store executable program code, and the processor is configured to call and run the executable program code from the memory, so that the device executes the method in the first aspect or any possible implementation manner of the first aspect.

[0062] In a fourth aspect, a computer program product is provided, which comprises computer program code. When the computer program code runs on a computer, the computer program code causes the computer to execute the method in the first aspect or any possible implementation manner of the first aspect.

[0063] In a fifth aspect, a computer readable storage medium is provided, which stores computer program code. When the computer program code runs on a computer, the computer program code causes the computer to execute the method in the first aspect or any possible implementation manner of the first aspect.

[0064] The present application has the following beneficial effects:

[0065] The needle tip position recognition method of the thyroid fine needle puncture ultrasound image provided by the present application realizes the recognition of the needle tip position by analyzing the target ultrasound image, solves the technical problem of poor accuracy of needle tip position recognition, and improves the accuracy of needle tip position recognition. When recognizing the needle tip position, the present application comprehensively considers a plurality of factors related to needle tip position recognition, such as local difference characteristic values and characteristic state variation indexes, thereby more accurately recognizing the target fine needle region representing the fine needle to which the needle tip belongs, and comprehensively considering the motion change between the target fine needle regions in adjacent frame target ultrasound images, quantifying the fine needle motion state corresponding to each frame target ultrasound image, thereby realizing the recognition of the needle tip position and improving the accuracy of the needle tip position recognition to a certain extent. BRIEF DESCRIPTION OF DRAWINGS

[0066] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0067] Figure 1 The flow chart of the needle tip position recognition method of the thyroid fine needle puncture ultrasound image of the present application;

[0068] Figure 2 The composition structure schematic diagram of the needle tip position recognition system of the thyroid fine needle puncture ultrasound image of the present application;

[0069] Figure 3 The structure schematic diagram of the computer device of the present application. DETAILED DESCRIPTION

[0070] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific implementation, structure, features and effects of the technical solutions proposed according to the present application will be described in detail below in combination with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0071] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0072] REFERENCE Figure 1, shows the flow of some embodiments of a needle tip position recognition method of a thyroid fine needle aspiration ultrasound image according to the present application. The needle tip position recognition method of the thyroid fine needle aspiration ultrasound image comprises the following steps:

[0073] Step S1, obtain each frame of target ultrasound image in the current puncture process, and determine the local difference feature value corresponding to each pixel point based on the gray scale change in the preset neighborhood corresponding to each pixel point in each frame of target ultrasound image.

[0074] Wherein, the current puncture process can be the thyroid fine needle aspiration process being performed at the current time. Fine needle aspiration (FNA) is a diagnostic examination method commonly used to evaluate thyroid nodules or masses. This method uses a fine needle to extract a small amount of cell sample from the thyroid nodule or mass for microscopic examination to help the doctor determine the nature of the nodule. A complete thyroid fine needle aspiration process can be: the process of the doctor using an elongated needle to pass through the skin, enter the thyroid nodule, and completely pull out the needle under the guidance of ultrasound. The target ultrasound image can be an ultrasound image of the thyroid nodule collected during the thyroid fine needle aspiration process. The preset neighborhood can be a pre-set neighborhood. For example, the preset neighborhood can be an eight-neighborhood.

[0075] It should be noted that when guiding fine needle aspiration of the thyroid nodule of a patient, one or two medical staff often hold a linear array ultrasound probe to obtain ultrasound images of the thyroid nodule region while operating the puncture needle. During the puncture process of the patient, multiple guidance is often required to ensure accurate puncture of the thyroid nodule and sample collection. Therefore, when puncturing the thyroid nodule of the patient, the position of the needle tip needs to be located at all times to more accurately puncture the nodule and achieve sample collection.

[0076] As an example, the present step can include the following steps:

[0077] First, obtain each frame of target ultrasound image in the current puncture process.

[0078] For example, the linear array ultrasound probe can be used to collect ultrasound images of the thyroid nodule during the current puncture process, and each frame of ultrasound image collected during the current puncture process can be recorded as a target ultrasound image.

[0079] Wherein, the first frame of target ultrasound image can be an ultrasound image collected at the beginning of the current puncture process. At this time, the fine needle has not yet been used to pass through the skin when the first frame of target ultrasound image is collected, therefore, the fine needle region does not exist in the first frame of target ultrasound image, that is, the needle tip position recognition of the first frame of target ultrasound image is often not required.

[0080] It should be noted that the patient's thyroid nodule region is detected using a linear array probe, and an ultrasound image of the patient's thyroid nodule region can be obtained.

[0081] Secondly, according to the gray level difference between each pixel point in each target ultrasound image and each pixel point in the corresponding preset neighborhood, the local gray level deviation factor corresponding to each pixel point in each target ultrasound image is determined, which can include the following sub-steps:

[0082] The first sub-step is to determine any one frame of target ultrasound image as a marked ultrasound image, and any one pixel point in the above marked ultrasound image as a marked pixel point.

[0083] The second sub-step is to determine the absolute value of the difference between the gray value corresponding to the marked pixel point and the gray value corresponding to each pixel point in the corresponding preset neighborhood as the gray level difference, and obtain the gray level difference set corresponding to the marked pixel point.

[0084] The gray level difference set corresponding to the marked pixel point can include the absolute value of the difference between the gray value corresponding to the marked pixel point and the gray value corresponding to each pixel point in the corresponding preset neighborhood.

[0085] The third sub-step is to determine the maximum gray level difference in the gray level difference set corresponding to the marked pixel point as the gray level difference peak value corresponding to the marked pixel point.

[0086] The fourth sub-step is to determine the mean value of all gray level differences in the gray level difference set corresponding to the marked pixel point as the gray level difference mean value corresponding to the marked pixel point.

[0087] The fifth sub-step is to determine the product of the gray level difference peak value and the gray level difference mean value corresponding to the marked pixel point as the local gray level deviation factor corresponding to the marked pixel point.

[0088] For example, the formula for determining the local gray level deviation factor corresponding to the pixel point in the target ultrasound image can be:

[0089] δ i,j = MP i,j × μP i,j ; wherein δ i,j is the local gray level deviation factor corresponding to the jth pixel point in the ith target ultrasound image. i is the serial number of the target ultrasound image. j is the serial number of the pixel point in the ith target ultrasound image. MP i,j is the gray level difference peak value corresponding to the jth pixel point in the ith target ultrasound image, that is, the maximum gray level difference in the gray level difference set corresponding to the jth pixel point in the ith target ultrasound image. μP i,jis the mean of the gray scale difference corresponding to the jth pixel point in the ith target ultrasound image, that is, the mean of all gray scale differences in the gray scale difference set corresponding to the jth pixel point in the ith target ultrasound image.

[0090] It should be noted that during the puncture process, the gray scale of the area where the fine needle pierces into the patient often has a certain difference from the gray scale of the human tissue, and therefore the gray scale corresponding to the fine needle pixel point often has a certain difference from the gray scale in the neighborhood thereof. When the MP i,j is larger, it often indicates that the maximum gray scale difference between the jth pixel point and its neighborhood is larger, and often indicates that the gray scale difference between the jth pixel point and its neighborhood is larger. The MP i,j can be used as the weight of the μP i,j . When the μP i,j is larger, it often indicates that the overall gray scale difference between the jth pixel point and its neighborhood is larger. Therefore, the δ i,j can represent the gray scale distribution around the jth pixel point, and when the value thereof is larger, it often indicates that the gray scale difference between the jth pixel point and its neighborhood is larger, and often indicates that the jth pixel point is more likely to be a fine needle pixel point.

[0091] Thirdly, the mean of the local gray scale deviation factors corresponding to all pixel points in the preset neighborhood corresponding to each pixel point in each target ultrasound image is normalized to obtain a local difference feature value corresponding to each pixel point in each target ultrasound image.

[0092] For example, the formula corresponding to the local difference feature value corresponding to the pixel point in the target ultrasound image can be as follows:

[0093] wherein, is the local difference feature value corresponding to the jth pixel point in the ith target ultrasound image. i is the serial number of the target ultrasound image. j is the serial number of the pixel point in the ith target ultrasound image. norm() is a normalization function. n ij is the number of pixel points in the preset neighborhood corresponding to the jth pixel point in the ith target ultrasound image. a is the serial number of the pixel point in the preset neighborhood corresponding to the jth pixel point in the ith target ultrasound image. δ i,j is the local gray scale deviation factor corresponding to the jth pixel point in the ith target ultrasound image.

[0094] It should be noted that, can represent the gray scale change around the jth pixel point, and when the value thereof is larger, it often indicates that the degree of the gray scale change around the jth pixel point is relatively larger.

[0095] Step S2, determining the feature state variation index corresponding to each pixel point according to the difference between the local difference feature value corresponding to each pixel point and the local difference feature value corresponding to the corresponding pixel point in the previous frame target ultrasound image of each pixel point.

[0096] wherein any one pixel point is denoted as a temporary pixel point, and the corresponding pixel point of the temporary pixel point in the previous frame target ultrasound image of the temporary pixel point can be: the pixel point in the previous frame target ultrasound image of the target ultrasound image to which the temporary pixel point belongs, which has the same position as the temporary pixel point.

[0097] As an example, the present step can include the following steps:

[0098] Firstly, the absolute value of the difference between the local difference feature value corresponding to each pixel point in each frame target ultrasound image and the local difference feature value corresponding to the corresponding pixel point in the previous frame target ultrasound image of each pixel point is determined as the initial state variation factor corresponding to each pixel point in each frame target ultrasound image.

[0099] Secondly, the feature state variation index corresponding to each pixel point in each frame target ultrasound image is determined according to the initial state variation factor and the local gray scale deviation factor corresponding to each pixel point in each frame target ultrasound image.

[0100] Wherein, the initial state variation factor and the local gray scale deviation factor can both have a positive correlation with the feature state variation index.

[0101] For example, the formula for determining the feature state variation index corresponding to the pixel point can be:

[0102] wherein θ i,j is the feature state variation index corresponding to the jth pixel point in the ith frame target ultrasound image. i is the serial number of the target ultrasound image. j is the serial number of the pixel point in the ith frame target ultrasound image. δ i,j is the local gray scale deviation factor corresponding to the jth pixel point in the ith frame target ultrasound image. || is the absolute value function. is the local difference feature value corresponding to the jth pixel point in the ith frame target ultrasound image. is the local difference feature value corresponding to the jth pixel point in the i-1th frame target ultrasound image. The jth pixel point in the i-1th frame target ultrasound image is the corresponding pixel point in the previous frame target ultrasound image of the jth pixel point in the ith frame target ultrasound image. γ is a factor greater than 0 which is set in advance and is mainly used to prevent the denominator from being 0. For example, γ can be 0.001. is the initial state variation factor corresponding to the jth pixel point in the ith frame target ultrasound image.

[0103] It should be noted that during the puncture process, the gray scale of the area of the fine needle penetrating the patient often has certain differences with the gray scale of the human body tissue, therefore, the gray scale corresponding to the fine needle pixel point often has certain differences with the gray scale in the neighborhood thereof. Moreover, the fine needle traction often causes the gray scale between the pixel points at the same position in the adjacent frame ultrasound images to have deviations, therefore, the fine needle region can be identified by comparing the gray scale change differences between the pixel points at the same position in the adjacent frame ultrasound images. i,j The gray scale distribution condition around the jth pixel point can be represented, when the value is larger, it often indicates that the gray scale difference between the jth pixel point and its neighborhood is larger, and it often indicates that the jth pixel point is more likely to be a fine needle pixel point. When the value is larger, it often indicates that the surrounding gray scale change condition corresponding to the jth pixel point at the same position in the adjacent frame target ultrasound images is less similar, and it often indicates that the jth pixel point is more likely to be a fine needle pixel point. Therefore, when the value is larger, the jth pixel point in the ith frame target ultrasound image is more likely to be a fine needle pixel point. i,j When the value is larger, it often indicates that the jth pixel point in the ith frame target ultrasound image is more likely to be a fine needle pixel point.

[0104] Step S3, according to the feature state variation index corresponding to the pixel point in each frame target ultrasound image, a target variation region is screened out from each frame target ultrasound image, and edge detection is performed on the target variation region to obtain a target edge contour.

[0105] As an example, the present step can include the following steps:

[0106] Firstly, the pixel points with the feature state variation index greater than the preset variation threshold value in each frame target ultrasound image are screened out as variation pixel points.

[0107] The preset variation threshold value can be a minimum feature state variation index required when it is considered that the pixel point has variation. For example, the preset variation threshold value can be 0.3.

[0108] Secondly, the minimum circumscribed rectangle of the region composed of all the variation pixel points in each frame target ultrasound image is determined as the target variation region.

[0109] Thirdly, edge detection is performed on the target variation region to obtain a target edge contour.

[0110] For example, the target variation region can be subjected to edge detection by using a sobel bidirectional edge detection algorithm, and the edge contour obtained at this time is denoted as the target edge contour.

[0111] Step S4, according to the shape feature condition of all the target edge contours in each frame target ultrasound image, a target fine needle region is screened out from each frame target ultrasound image.

[0112] It should be noted that, since the shape of the fine needle region often has a certain uniqueness, analyzing the shape characteristics of the target edge contour can facilitate subsequent screening of the target fine needle region representing the fine needle.

[0113] As an example, the present step can include the following steps:

[0114] In a first step, determining the local contour direction vector corresponding to each edge pixel point on each target edge contour according to each edge pixel point and its adjacent edge pixel point on each target edge contour can include the following sub-steps:

[0115] In a first sub-step, any one target edge contour is determined as a marker edge contour, and any one edge pixel point on the marker edge contour is determined as a marker edge point.

[0116] In a second sub-step, any one of the two edge pixel points adjacent to the marker edge point on the marker edge contour is determined as a reference edge point.

[0117] It should be noted that there are usually two edge pixel points adjacent to the marker edge point on the marker edge contour, and the reference edge point is one of them.

[0118] In a third sub-step, the local contour direction vector corresponding to the marker edge point is constructed according to the marker edge point and the reference edge point.

[0119] The modulus of the local contour direction vector corresponding to the marker edge point can be equal to the distance between the marker edge point and the reference edge point. The direction of the local contour direction vector corresponding to the marker edge point can be the direction in which the marker edge point points to the reference edge point. The local contour direction vector corresponding to the marker edge point can represent the vector to which the line connecting the marker edge point and the adjacent edge pixel point belongs to a certain extent, which can also be referred to as the line connecting vector corresponding to the marker edge point.

[0120] In a second step, the absolute value of the cosine similarity between the local contour direction vectors corresponding to each edge pixel point and any one of its adjacent edge pixel points on each target edge contour is determined as the direction change similarity corresponding to each edge pixel point on each target edge contour.

[0121] For example, the absolute value of the cosine similarity between the local contour direction vector corresponding to the marker edge point and the local contour direction vector corresponding to the reference edge point can be determined as the direction change similarity corresponding to the marker edge point.

[0122] It should be noted that, since the fine needle often presents a linear shape, the included angle between the connecting vectors between the adjacent edge pixel points on the fine needle is often 0° or 180°. Therefore, the cosine similarity between the connecting vectors corresponding to the adjacent fine needle edge pixel points is often 1 or -1, so the absolute value of the cosine similarity between the connecting vectors corresponding to the adjacent fine needle edge pixel points is often 1. That is, the direction change similarity corresponding to the fine needle edge pixel points is often the same, often 1.

[0123] In the third step, the fine needle shape possibility index corresponding to each target edge contour can be determined according to the difference between the direction change similarities corresponding to all adjacent edge pixel points on each target edge contour, and can include the following sub-steps:

[0124] In the first sub-step, the absolute value of the difference between the direction change similarities corresponding to each adjacent edge pixel point on each target edge contour is determined as a change similarity difference, so as to obtain a change similarity difference set corresponding to each target edge contour.

[0125] In the first sub-step, the absolute value of the difference between the direction change similarities corresponding to each adjacent edge pixel point on each target edge contour is determined as a change similarity difference, so as to obtain a change similarity difference set corresponding to each target edge contour.

[0126] In the second sub-step, the mean value of all change similarity differences in the change similarity difference set corresponding to each target edge contour is determined as the overall similarity difference corresponding to each target edge contour.

[0127] In the third sub-step, the fine needle shape possibility index corresponding to each target edge contour is determined according to the overall similarity difference corresponding to each target edge contour.

[0128] In the third sub-step, the fine needle shape possibility index corresponding to each target edge contour is determined according to the overall similarity difference corresponding to each target edge contour.

[0129] For example, the formula for determining the fine needle shape possibility index corresponding to the target edge contour can be:

[0130] In the formula, w t is the fine needle shape possibility index corresponding to the tthtarget edge contour. t is the serial number of the target edge contour. exp() is the natural exponential function. m t is the number of edge pixel points on the tthtarget edge contour. b is the serial number of the edge pixel point on the tthtarget edge contour. || is the absolute value function. CS t,b is the direction change similarity corresponding to the bthedge pixel point on the tthtarget edge contour. CS t,b+1 is the direction change similarity corresponding to the b+1thedge pixel point on the tthtarget edge contour. is the overall similarity difference corresponding to the tth target edge contour. |CS t,b -CS t,b+1 | is the change similarity difference.

[0131] It should be noted that the direction change similarity corresponding to different fine needle edge pixel points is often the same. When |CS t,b -CS t,b+1 | is smaller, it often means that the direction change similarity corresponding to different edge pixel points on the tth target edge contour is more similar, it often means that the tth target edge contour is more likely to present a linear shape, and it often means that the tth target edge contour is more likely to be a fine needle contour. Therefore, when w t is larger, it often means that the tth target edge contour is more likely to be a fine needle contour.

[0132] Fourthly, the target edge contour with the largest possible fine needle shape indicator in each target ultrasound image is screened out as a fine needle candidate contour.

[0133] Fifthly, the edge pixel points on the fine needle candidate contour in each target ultrasound image are taken as initial seed points for region growing to obtain a target sub-region.

[0134] The target sub-region can be a region obtained after region growing.

[0135] For example, the edge pixel points on the fine needle candidate contour can be taken as initial seed points, and the pixel points in the target ultrasound image are regionally grown based on the gray values corresponding to the pixel points in the target ultrasound image. Each region obtained by region growing is recorded as a target sub-region.

[0136] It should be noted that further correcting the fine needle candidate contour by region growing can facilitate subsequent obtaining of a target fine needle region representing a real fine needle.

[0137] Sixthly, the union of the internal region of the fine needle candidate contour and the target sub-region to which the fine needle candidate contour belongs is determined as a target fine needle region.

[0138] The internal region of the fine needle candidate contour is the region surrounded by the fine needle candidate contour.

[0139] Step S5, according to the motion change between the target fine needle regions in adjacent frames of target ultrasound images, determine the fine needle motion state corresponding to each frame of target ultrasound images.

[0140] As an example, this step can include the following steps:

[0141] Firstly, the pixel point with the largest feature state variation indicator in the target fine needle region in each frame of target ultrasound image is screened out as a needle tip candidate point.

[0142] Secondly, a pixel point with the largest gray difference between the needle tip candidate point is selected from a preset neighborhood corresponding to the needle tip candidate point in each target ultrasound image as a reference point corresponding to the needle tip candidate point in each target ultrasound image.

[0143] The reference point can be a pixel point with the largest gray difference between the needle tip candidate point in a preset neighborhood corresponding to the needle tip candidate point.

[0144] Thirdly, a reference direction vector corresponding to each target ultrasound image is constructed according to the needle tip candidate point and the reference point corresponding to the needle tip candidate point in each target ultrasound image.

[0145] The modulus of the reference direction vector can be equal to the distance between the needle tip candidate point and the reference point corresponding to the needle tip candidate point. The direction of the reference direction vector can be the direction in which the needle tip candidate point points to the reference point corresponding to the needle tip candidate point.

[0146] Fourthly, if the cosine similarity between the reference direction vectors corresponding to adjacent target ultrasound images is greater than a preset similarity threshold, it is determined that the motion states of the fine needle corresponding to the adjacent target ultrasound images are the same.

[0147] The preset similarity threshold can be 0.

[0148] It should be noted that if the cosine similarity between the reference direction vectors corresponding to adjacent target ultrasound images is greater than 0, it is often indicated that the included angle between the reference direction vectors corresponding to the two target ultrasound images is less than 90°, it is often indicated that the moving direction of the needle tip candidate point in the two target ultrasound images can not have changed greatly, and it is often indicated that the motion state of the fine needle in the two target ultrasound images can not have changed.

[0149] Fifthly, if the cosine similarity between the reference direction vectors corresponding to adjacent target ultrasound images is less than or equal to the preset similarity threshold, it is determined that the motion states of the fine needle corresponding to the adjacent target ultrasound images are different.

[0150] Sixthly, the continuous target ultrasound images with the same motion state of the fine needle are constructed into a target ultrasound image group to obtain a target ultrasound image group sequence.

[0151] The target ultrasound image group often does not contain the first target ultrasound image.

[0152] For example, if 8 target ultrasound images are collected in the current puncture process, and the 8 target ultrasound images are the first image, the second image, the third image, the fourth image, the fifth image, the sixth image, the seventh image and the eighth image in turn; it is calculated that the fine needle motion state corresponding to the second image and the third image is the same; the fine needle motion state corresponding to the third image and the fourth image is the same; the fine needle motion state corresponding to the fourth image and the fifth image is different; the fine needle motion state corresponding to the fifth image and the sixth image is the same; the fine needle motion state corresponding to the sixth image and the seventh image is different; the fine needle motion state corresponding to the seventh image and the eighth image is the same, then 3 target ultrasound image groups can be obtained at this time, the 3 target ultrasound image groups are {second image, third image, fourth image}, {fifth image, sixth image} and {seventh image, eighth image} respectively, and the serial number of {second image, third image, fourth image} in the target ultrasound image group sequence is 1, which belongs to an odd serial number; the serial number of {fifth image, sixth image} in the target ultrasound image group sequence is 2, which belongs to an even serial number; the serial number of {seventh image, eighth image} in the target ultrasound image group sequence is 3, which belongs to an odd serial number.

[0153] In the seventh step, the fine needle motion state corresponding to all target ultrasound images in the target ultrasound image group with an odd serial number in the above target ultrasound image group sequence is set to a fine needle puncture state.

[0154] In the eighth step, the fine needle motion state corresponding to all target ultrasound images in the target ultrasound image group with an even serial number in the above target ultrasound image group sequence is set to a fine needle withdrawal state.

[0155] It should be noted that the fine needle is often inserted into the skin of the patient in the initial stage of the thyroid fine needle puncture process, that is, the needle tip in the ultrasound image collected in the initial stage of the thyroid fine needle puncture process is often in the puncture state. The target ultrasound images in the first target ultrasound image group often belong to the initial stage of the current puncture process, often indicating that the needle tip in each target ultrasound image in the first target ultrasound image group is in the puncture state, that is, the needle tip at this time is often inserted into a deeper position. The needle tip state of the second target ultrasound image group and the needle tip state of the first target ultrasound image group, since the needle tip of the first target ultrasound image group is in the puncture state, the needle tip of the second target ultrasound image group is often in the withdrawal state, the needle tip at this time is often withdrawn, that is, the position of the needle tip at the previous moment may be incorrect, the position of the needle tip is adjusted, and the needle tip is withdrawn to facilitate the subsequent insertion of the needle tip into the correct position. By analogy, the needle tip of the target ultrasound image group with an odd serial number is often in the puncture state, and the needle tip of the target ultrasound image group with an even serial number is often in the withdrawal state.

[0156] Step S6, identifying the needle tip position in each target ultrasound image based on the corresponding fine needle motion state of each target ultrasound image.

[0157] As an example, the present step can include the following steps:

[0158] Firstly, if the fine needle motion state corresponding to the target ultrasound image is the fine needle puncture state, the position where the needle tip candidate point is located in the target ultrasound image is determined as the needle tip position in the target ultrasound image.

[0159] It should be noted that in the process of needle tip puncture, the subcutaneous tissue close to the fine needle often changes due to friction on the surface of the fine needle, and the needle tip often needs to destroy the original subcutaneous tissue to facilitate the fine needle to enter the patient's body, so the change of the subcutaneous tissue at the needle tip position is more obvious. Therefore, compared with other positions of the fine needle, the feature state change at the needle tip position is often the largest, that is, the needle tip candidate point often represents the true needle tip in the process of needle tip puncture, and the position of the needle tip candidate point is the position of the needle tip at this time. The direction of the reference direction vector corresponding to the target ultrasound image is often the puncture direction of the needle tip in the target ultrasound image.

[0160] Secondly, if the fine needle motion state corresponding to the target ultrasound image is the fine needle extraction state, the position where the target fine needle region to which the needle tip candidate point belongs is located in the target ultrasound image is determined as the position to which the needle tip belongs in the target ultrasound image.

[0161] It should be noted that in the process of needle tip extraction, the degree of damage to the subcutaneous tissue by the needle tip is often less than the degree of damage caused in the process of needle tip puncture, and the feature state change near the needle tip is often similar at this time. Therefore, the needle tip candidate point at this time may not be the true needle tip, but a position near the needle tip, so the target fine needle region representing the fine needle to which the needle tip belongs can be marked at this time to facilitate the doctor to observe the target fine needle region and locate the needle tip position. The opposite direction of the direction of the reference direction vector corresponding to the target ultrasound image is often the extraction direction of the needle tip in the target ultrasound image.

[0162] Reference Figure 2 Based on the same inventive concept as the above method embodiments, the present application provides a needle tip position identification system for thyroid fine needle puncture ultrasound images, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. The above computer program is executed by the processor to realize the steps of a needle tip position identification method for thyroid fine needle puncture ultrasound images, which can specifically include:

[0163] The acquisition and determination module 201 is configured to acquire each frame of target ultrasound image in a current puncture process, and determine a local difference feature value corresponding to each pixel point based on a gray scale change in a preset neighborhood corresponding to the pixel point in each frame of target ultrasound image.

[0164] The feature state variation index determination module 202 is configured to determine a feature state variation index corresponding to each pixel point according to a difference between the local difference feature value corresponding to the pixel point and a local difference feature value corresponding to a corresponding pixel point in a previous frame of target ultrasound image.

[0165] The screening and edge detection module 203 is configured to screen a target variation region from each frame of target ultrasound image according to the feature state variation index corresponding to the pixel point in each frame of target ultrasound image, and perform edge detection on the target variation region to obtain a target edge contour.

[0166] The region screening module 204 is configured to screen a target fine needle region from each frame of target ultrasound image according to a shape feature of all target edge contours in each frame of target ultrasound image.

[0167] The fine needle motion state determination module 205 is configured to determine a fine needle motion state corresponding to each frame of target ultrasound image according to a motion change between target fine needle regions in adjacent frames of target ultrasound image.

[0168] The needle tip position recognition module 206 is configured to recognize a needle tip position in each frame of target ultrasound image based on the fine needle motion state corresponding to each frame of target ultrasound image.

[0169] Figure 3 is a structural schematic diagram of a computer device provided by an embodiment of the present application. As shown in the example, Figure 3 the computer device 300 includes a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302, wherein the processor 302 executes the computer program 303, so that the computer device can execute any one of the aforementioned thyroid fine needle puncture ultrasound image needle tip position recognition methods.

[0170] Based on the same inventive concept as the above method embodiment, the present application provides a server including a memory and a processor. The memory is configured to store executable program code, and the processor is configured to call and run the executable program code from the memory, so that the device executes any one of the aforementioned thyroid fine needle puncture ultrasound image needle tip position recognition methods.

[0171] Based on the same inventive concept as the above method embodiments, the present application provides a computer program product comprising computer program code which, when run on a computer, causes the computer to perform any of the above thyroid fine needle puncture ultrasound image needle tip position recognition methods.

[0172] Based on the same inventive concept as the above method embodiments, the present application provides a computer readable storage medium storing computer program code which, when run on a computer, causes the computer to perform any of the above thyroid fine needle puncture ultrasound image needle tip position recognition methods.

[0173] In summary, when performing needle tip position recognition, the present application comprehensively considers a plurality of factors related to needle tip position recognition, such as local difference characteristic values and characteristic state variation indicators, thereby accurately identifying the target fine needle region representing the fine needle to which the needle tip belongs, and comprehensively considering the motion change between target fine needle regions in adjacent frames of target ultrasound images, quantifying the fine needle motion state corresponding to each frame of target ultrasound images, thereby achieving needle tip position recognition and improving the accuracy of needle tip position recognition to a certain extent.

[0174] The above embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some technical features therein; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A needle tip position recognition method of a thyroid fine needle aspiration ultrasound image, characterized by, The method comprises the following steps: acquiring each frame of target ultrasound image in a current puncture process, and determining a local difference characteristic value corresponding to each pixel point in each frame of target ultrasound image based on a gray scale change condition in a preset neighborhood corresponding to each pixel point; determining a feature state variation index corresponding to each pixel point according to a difference between the local difference characteristic value corresponding to each pixel point and a local difference characteristic value corresponding to a corresponding pixel point in a previous frame of target ultrasound image; screening a target variation region from each frame of target ultrasound image according to the feature state variation index corresponding to the pixel point in each frame of target ultrasound image, and performing edge detection on the target variation region to obtain a target edge contour; screening a target fine needle region from each frame of target ultrasound image according to a shape feature condition of all target edge contours in each frame of target ultrasound image; determining a fine needle motion state corresponding to each frame of target ultrasound image according to a motion change condition between the target fine needle regions in adjacent frames of target ultrasound image; identifying a needle tip position in each frame of target ultrasound image based on the fine needle motion state corresponding to each frame of target ultrasound image.

2. The method of claim 1, wherein the method is characterized by: The method comprises the following steps: determining a local gray scale deviation factor corresponding to each pixel point in each frame of target ultrasound image according to a gray scale difference between each pixel point and each pixel point in a preset neighborhood corresponding to each pixel point; normalizing a mean value of the local gray scale deviation factors corresponding to all pixel points in the preset neighborhood corresponding to each pixel point in each frame of target ultrasound image to obtain a local difference characteristic value corresponding to each pixel point in each frame of target ultrasound image.

3. The method of claim 2, wherein the method further comprises: The method comprises the following steps: determining a local gray scale deviation factor corresponding to each pixel point in each frame of target ultrasound image according to a gray scale difference between each pixel point and each pixel point in a preset neighborhood corresponding to each pixel point; determining a gray scale difference between a gray scale value corresponding to the marker pixel point and a gray scale value corresponding to each pixel point in the preset neighborhood corresponding to the marker pixel point as a gray scale difference to obtain a gray scale difference set corresponding to the marker pixel point; determining a maximum gray scale difference in the gray scale difference set corresponding to the marker pixel point as a gray scale difference peak value corresponding to the marker pixel point; determining a mean value of all gray scale differences in the gray scale difference set corresponding to the marker pixel point as a gray scale difference mean value corresponding to the marker pixel point; determining a product of the gray scale difference peak value and the gray scale difference mean value corresponding to the marker pixel point as a local gray scale deviation factor corresponding to the marker pixel point.

4. The method of claim 2, wherein the method further comprises: The method comprises the following steps: An absolute value of a difference between the local difference feature value corresponding to each pixel point in each target ultrasound image and a local difference feature value corresponding to a pixel point in a previous target ultrasound image of each target ultrasound image is determined as an initial state variation factor corresponding to each pixel point in each target ultrasound image; A feature state variation index corresponding to each pixel point in each target ultrasound image is determined according to the initial state variation factor and the local gray scale deviation factor corresponding to each pixel point in each target ultrasound image, wherein the initial state variation factor and the local gray scale deviation factor are positively correlated with the feature state variation index.

5. The method of identifying the needle tip position in a thyroid fine needle aspiration ultrasound image of claim 1, wherein, The target variation region is filtered out from each target ultrasound image according to the feature state variation index corresponding to the pixel point in each target ultrasound image, and the target variation region includes: Pixels corresponding to the feature state variation index greater than a preset variation threshold value in each target ultrasound image are filtered out as variation pixels; A minimum circumscribed rectangle of a region formed by all variation pixels in each target ultrasound image is determined as the target variation region.

6. The method of identifying the needle tip position of a thyroid fine needle aspiration ultrasound image of claim 1, wherein, The target fine needle region is filtered out from each target ultrasound image according to a shape feature condition of all target edge contours in each target ultrasound image, and the target fine needle region includes: A local contour direction vector corresponding to each edge pixel point on each target edge contour is determined according to each edge pixel point and a neighboring edge pixel point on each target edge contour; An absolute value of a cosine similarity between the local contour direction vectors corresponding to each edge pixel point on each target edge contour and any one of the neighboring edge pixel points is determined as a direction change similarity corresponding to each edge pixel point on each target edge contour; A fine needle shape possibility index corresponding to each target edge contour is determined according to a difference between the direction change similarities corresponding to all neighboring edge pixel points on each target edge contour; A target edge contour corresponding to the maximum fine needle shape possibility index is filtered out from each target ultrasound image as a fine needle candidate contour; A target sub-region is obtained by region growing from the edge pixel points on the fine needle candidate contour in each target ultrasound image as initial seed points; A union of an internal region of the fine needle candidate contour and a target sub-region to which the fine needle candidate contour belongs is determined as the target fine needle region.

7. The method of identifying the needle tip position of a thyroid fine needle aspiration ultrasound image according to claim 6, wherein, The local contour direction vector corresponding to each edge pixel point on each target edge contour is determined according to each edge pixel point and a neighboring edge pixel point on each target edge contour, and the local contour direction vector includes: Any one target edge contour is determined as a marker edge contour, and any one edge pixel point on the marker edge contour is determined as a marker edge point; Any one of two edge pixel points adjacent to the marker edge point on the marker edge contour is determined as a reference edge point; According to the marked edge point and the reference edge point, a local contour direction vector corresponding to the marked edge point is constructed, wherein a modulus of the local contour direction vector corresponding to the marked edge point is equal to a distance between the marked edge point and the reference edge point; and a direction of the local contour direction vector corresponding to the marked edge point is a direction in which the marked edge point points to the reference edge point.

8. The method of claim 6, wherein the method further comprises: determining a position of the needle tip in the thyroid fine needle aspiration ultrasound image based on the determined position of the needle in the ultrasound image. The determining the fine needle shape possibility index corresponding to each target edge contour according to the difference between the direction change similarities corresponding to all adjacent edge pixel points on each target edge contour comprises: The absolute value of the difference between the direction change similarities corresponding to each adjacent edge pixel point on each target edge contour is determined as a change similarity difference, so as to obtain a change similarity difference set corresponding to each target edge contour; A mean value of all change similarity differences in the change similarity difference set corresponding to each target edge contour is determined as an overall similarity difference corresponding to each target edge contour; The fine needle shape possibility index corresponding to each target edge contour is determined according to the overall similarity difference corresponding to each target edge contour, wherein the overall similarity difference and the fine needle shape possibility index are in a negative correlation relationship.

9. The method of claim 1, wherein the method is a method of identifying a needle tip position of a thyroid fine needle aspiration ultrasound image, the method comprising: receiving a thyroid fine needle aspiration ultrasound image; identifying a needle tip position of the thyroid fine needle aspiration ultrasound image; and outputting the needle tip position of the thyroid fine needle aspiration ultrasound image. The determining the fine needle motion state corresponding to each target ultrasound image according to the motion change between the target fine needle regions in adjacent frame target ultrasound images comprises: A pixel point with a maximum feature state variation index in the target fine needle region in each target ultrasound image is selected as a needle tip candidate point; A pixel point with a maximum gray scale difference between the needle tip candidate point and the pixel point in a preset neighborhood of the needle tip candidate point in each target ultrasound image is selected as a reference point corresponding to the needle tip candidate point in each target ultrasound image; A reference direction vector corresponding to each target ultrasound image is constructed according to the needle tip candidate point and the reference point corresponding to the needle tip candidate point in each target ultrasound image, wherein a modulus of the reference direction vector is equal to a distance between the needle tip candidate point and the reference point corresponding to the needle tip candidate point, and a direction of the reference direction vector is a direction in which the needle tip candidate point points to the reference point corresponding to the needle tip candidate point; If a cosine similarity between the reference direction vectors corresponding to adjacent target ultrasound images is greater than a preset similarity threshold, it is determined that the fine needle motion states corresponding to the adjacent target ultrasound images are the same; Continuous frame target ultrasound images with the same fine needle motion state are constructed into a target ultrasound image group, so as to obtain a target ultrasound image group sequence; Fine needle motion states corresponding to all target ultrasound images in the target ultrasound image group with an odd sequence number in the target ultrasound image group sequence are set as a fine needle penetration state; Fine needle motion states corresponding to all target ultrasound images in the target ultrasound image group with an even sequence number in the target ultrasound image group sequence are set as a fine needle extraction state.

10. The method of identifying the needle tip position of a thyroid fine needle aspiration ultrasound image of claim 9, wherein, The identifying the needle tip position in each target ultrasound image based on the fine needle motion state corresponding to each target ultrasound image comprises: If the fine needle motion state corresponding to the target ultrasound image is the fine needle penetration state, a position of the needle tip candidate point in the target ultrasound image is determined as the needle tip position in the target ultrasound image. If the fine needle motion state corresponding to the target ultrasound image is the fine needle pulling-out state, a position where a target fine needle region to which the needle tip candidate point in the target ultrasound image belongs is determined as a position of the needle tip in the target ultrasound image.

Images