A method and system for intima-media measurement of a blood vessel ultrasound image

Abstract

The embodiment of the specification provides a method and system for measuring an intima-media of a vascular ultrasound image, the method comprising: acquiring a region of interest of an ultrasound image; determining an intima-media measurement type based on the region of interest through a preset classification algorithm; selecting a to-be-measured image from a current frame ultrasound image based on the region of interest; acquiring a target mask image of the current frame ultrasound image based on the to-be-measured image; and determining an intima-media thickness parameter of the current frame ultrasound image according to the target mask image of the current frame ultrasound image based on the intima-media measurement type.

Description

Technical Field

[0001] This specification relates to the field of medical image processing, and in particular to a method and system for measuring the intima-media layer in vascular ultrasound images. Background Technology

[0002] As lipids deposit on the inner walls of blood vessels, they promote the thickening and proliferation of the intima-media structure, narrowing the arteries and causing atherosclerosis. The carotid aorta is a common site of atherosclerotic lesions, and carotid intima-media thickness (IMT) is an important indicator of the degree of atherosclerosis. Examining the carotid arteries using ultrasound equipment and measuring parameters such as IMT is of great significance for the prevention of cardiovascular diseases.

[0003] Therefore, it is desirable to provide a method and system for measuring the intima-media layer in vascular ultrasound images to improve measurement efficiency and accuracy, while saving computing power and meeting the needs of real-time automatic IMT measurement. Summary of the Invention

[0004] This specification provides one or more embodiments of a method for measuring the intima-media layer of a vascular ultrasound image. The method includes: acquiring a region of interest (ROI) of the ultrasound image; determining the intima-media measurement type based on the ROI using a preset classification algorithm; extracting a measurement image from a current frame of the ultrasound image based on the ROI; acquiring a target mask image of the current frame of the ultrasound image based on the measurement image; and determining the intima-media thickness parameters of the current frame of the ultrasound image based on the intima-media measurement type and the target mask image.

[0005] This specification provides an intima-media measurement system for vascular ultrasound images through one or more embodiments. The system includes: at least one processor and at least one memory; the at least one memory is used to store computer instructions; the at least one processor is used to execute at least a portion of the computer instructions to implement the intima-media measurement method for vascular ultrasound images.

[0006] This specification provides one or more embodiments of a method for measuring the intima-media layer of real-time vascular ultrasound images. The method includes: acquiring a region of interest (ROI) from multiple ultrasound images; determining the intima-media measurement type based on the ROI using a preset classification algorithm; extracting a measurement image from the current ultrasound image based on the ROI; acquiring a target mask image of the current ultrasound image based on the measurement image; determining a target mask image of the next ultrasound image based on the target mask image of the current ultrasound image; and determining the intima-media thickness parameters of the multiple ultrasound images based on the intima-media measurement type and the target mask images of the multiple ultrasound images.

[0007] This specification provides an intima-media measurement device for vascular ultrasound images through one or more embodiments. The device includes at least one processing device and at least one memory. The at least one memory is used to store computer instructions. The at least one processing device is used to execute at least a portion of the computer instructions to implement the intima-media measurement method for vascular ultrasound images.

[0008] This specification provides a computer-readable storage medium that stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the intima-media measurement method of any of the above-described methods for vascular ultrasound images.

[0009] Some embodiments of this specification include at least the following beneficial effects: (1) In response to the problem that the operator needs to manually select the IMT measurement type, some embodiments of this specification propose an automatic IMT measurement type identification scheme, automatically extracts image feature data in the region of interest, and indicates whether the current IMT measurement type is anterior wall IMT, posterior wall IMT, or simultaneous measurement of anterior and posterior wall IMT; (2) For automatic IMT measurement under given sampling frame conditions, the technical solution proposed in some embodiments of this specification can realize automatic IMT extraction, avoids the complex numerical calculation process of traditional schemes, and improves the algorithm running efficiency; at the same time, the algorithm provides a solution with high robustness by analyzing the biological structural characteristics of the inner and middle membranes, and is suitable for automatic IMT extraction under more complex conditions; (3) In response to the real-time IMT extraction requirements of video data, some embodiments of this specification combine the dependence of the preceding and following frames of the image into the entire algorithm extraction process, realizes automatic IMT extraction under real-time animation, and based on the results of automatic extraction, can perform time-series statistical analysis and give the final IMT change trend analysis chart, so as to effectively avoid the situation where there is a large error between the IMT measurement item and the true value due to the poor selection of a single frozen image. Attached Figure Description

[0010] Figure 1 This is a schematic diagram illustrating an application scenario of the endometrial membrane measurement system according to some embodiments of this specification;

[0011] Figure 2 These are exemplary block diagrams of an endometrial membrane measurement system according to some embodiments of this specification;

[0012] Figure 3 This is an exemplary flowchart of an intima-media measurement method according to some embodiments of this specification;

[0013] Figure 4 This is an exemplary flowchart illustrating the determination of a target mask image according to some embodiments of this specification;

[0014] Figure 5 These are exemplary schematic diagrams illustrating intima-media measurement types according to some embodiments of this specification;

[0015] Figure 6 This is an exemplary flowchart illustrating the processing of the next frame of ultrasound image according to some embodiments of this specification;

[0016] Figure 7 This is an exemplary schematic diagram of real-time measurement of IMT according to some embodiments of this specification. Detailed Implementation

[0017] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this specification. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.

[0018] It should be understood that the terms “system,” “device,” “unit,” and / or “module” used herein are one way to distinguish different components, elements, parts, sections, or assemblies at different levels. However, if other terms can achieve the same purpose, they may be replaced by other expressions.

[0019] Flowcharts are used in this specification to illustrate the operations performed by the system according to embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0020] The intima-media structure of an artery is contained within two nearly parallel boundaries. IMT is the distance between the luminal-intima interface (LII) and the media-adventitia interface (MAI). A normal IMT is less than 1 mm; 1.0–1.2 mm indicates intimal thickening; 1.2–1.4 mm indicates plaque formation; and greater than 1.4 mm indicates carotid artery stenosis. During IMT measurement, it is necessary to adjust the appropriate probe depth, and image magnification may be required to observe the precise locations of the upper and lower boundaries.

[0021] Traditional IMT measurement methods rely heavily on manual measurement by physicians. This method is easily influenced by the physician's experience and subjective judgment, resulting in significant observer error. Furthermore, this purely manual measurement is time-consuming and inefficient. In recent years, researchers have proposed two main approaches to automated IMT measurement: semi-automatic and fully automatic IMT measurement. The semi-automatic IMT measurement scheme requires the physician to visually interpret the patient's carotid artery ultrasound image, use the ultrasound equipment, and employ a sampling frame tool to delineate the intima-media region to be measured. The IMT measurement type is then selected, and a background algorithm calculates various IMT parameters in real time, drawing the contours of the LII and MAI within the sampling frame tool. This significantly reduces the workload of the ultrasound physician and improves examination efficiency. However, this measurement scheme lacks the ability to automatically distinguish IMT measurement types, requiring physicians to manually select the current IMT measurement type in practical applications; or it only supports semi-automatic measurement of far-field IMT, and the corresponding semi-automatic IMT measurement algorithm cannot provide suitable measurement results when near-field IMT measurement is required. The fully automated IMT measurement scheme utilizes multiple algorithms to automatically measure and analyze the IMT within the measurement area. While this method reduces the number of steps required by the doctor, it has several drawbacks: Firstly, the algorithm not only needs to select a suitable location for IMT analysis but also needs to perform contour analysis of the IMT, significantly increasing the execution time and making it difficult to meet the requirements for real-time IMT measurement. Furthermore, the suitability of the selected measurement site still needs to be verified by the doctor. Secondly, due to the poor imaging quality of the anterior wall's intima-media layer, this type of automated measurement scheme often cannot achieve simultaneous measurement of the anterior and posterior fields or can only achieve a certain degree of measurement on the posterior wall. It requires high image quality, and because the sampling frame is autonomously selected by the algorithm, the doctor cannot perform IMT measurements again on the area of ​​interest.

[0022] Most current research mainly relies on methods such as gradient boundary detection, active contour models, dynamic programming, and Hough transform to automatically extract the upper and lower contours of IMT. Among them, the Snake model has the best extraction effect, but this model has computational stability issues and is difficult to meet the requirement of real-time extraction of the upper and lower contours of IMT.

[0023] Therefore, in some embodiments of this specification, for an input ultrasound video stream, the algorithm automatically identifies the IMT measurement type within a manually selected sampling frame and performs a complete IMT measurement process on the first frame of the ultrasound image. For other frames of ultrasound images, a partial IMT measurement process is performed based on the target mask image of the previous frame of the ultrasound image, avoiding repetitive operations, improving computational efficiency, and realizing real-time IMT extraction for each frame of the ultrasound video stream.

[0024] Figure 1 This is a schematic diagram illustrating the application scenarios of the endometrial membrane measurement system according to some embodiments of this specification.

[0025] In some embodiments, vascular ultrasound imaging systems for measuring the intima-media thickness (IMMT) can be applied in multiple fields. For example, in the medical field, IMT measurement is used to aid in the diagnosis and treatment of cardiovascular diseases such as atherosclerosis and coronary heart disease. Based on the measured IMT thickness, doctors can determine the degree of vascular stenosis, assess the condition, and develop treatment plans. As another example, in the research field, IMT measurement is used to study the progression of cardiovascular diseases and evaluate drug efficacy. Furthermore, arterial ultrasound can measure the IMT of various internal organs or blood vessels, such as the carotid artery, heart, blood vessels, liver, and gallbladder.

[0026] In some embodiments, such as Figure 1 As shown, the application scenario 100 of the endometrial membrane measurement system includes an ultrasonic device 110, a network 120, a terminal device 130, a processing device 140, and a storage device 150.

[0027] Ultrasound device 110 is capable of performing ultrasound imaging on a subject to be scanned. The subject can be biological or non-biological. For example, it can be a patient, a man-made object, etc. The subject can be examined in any position, such as supine, lateral, prone, semi-recumbent, or sitting. The subject can include specific parts, organs, tissues, and / or body parts of a patient. By way of example only, the subject can include the head, brain, neck, body, shoulder, arm, chest, heart, stomach, blood vessels, soft tissue, knee, foot, etc., or combinations thereof.

[0028] Ultrasound device 110 can reflect image information of the internal tissues of the object to be scanned, to assist doctors in disease diagnosis. In some embodiments, ultrasound device 110 can acquire ultrasound images by utilizing the physical properties of ultrasound waves and differences in the acoustic properties of the object. The ultrasound images can be displayed and / or recorded in the form of waveforms, curves, or images, and / or features related to the object. In some embodiments, ultrasound device 110 includes one or any combination of ultrasound pulse echo imaging device, ultrasound echo Doppler imaging device, ultrasound electronic endoscope, ultrasound Doppler blood flow analysis device, ultrasound human tissue measurement device, etc. In some embodiments, ultrasound device 110 can receive relevant information about the object to be scanned and / or imaging operation instructions sent from terminal device 130 or processing device 140 via network 120, and can send scan data or ultrasound images to processing device 140, storage device 150, or terminal device 130.

[0029] Network 120 refers to any suitable network that facilitates the exchange of information and / or data among one or more components in application scenario 100. In some embodiments, one or more other components of application scenario 100 (e.g., ultrasound device 110, terminal device 130, processing device 140, storage device 150, etc.) can exchange information and / or data with each other via network 120. For example, processing device 140 can acquire ultrasound images from ultrasound device 110 via network 120. As another example, processing device 140 can acquire user instructions from terminal device 130 via network 120.

[0030] In some embodiments, network 120 is any one or more of wired or wireless networks. For example, network 120 includes one or more combinations of cable networks, wired networks, fiber optic networks, telecommunications networks, local area networks (LANs), wireless local area networks (WLANs), metropolitan area networks (MANs), public switched telephone networks (PSTNs), Bluetooth™ networks, ZigBee™ networks, near field communication networks (NFCs), etc. Network connections between components can be achieved using one or more of the above methods. In some embodiments, network 120 can be a point-to-point, shared, centralized, or other topologies, or a combination of multiple topologies. In some embodiments, network 120 may include one or more network access points. For example, network 120 may include wired and / or wireless network access points, such as base stations and / or network switching points, through which one or more components of application scenario 100 can access network 120 for data and / or information exchange.

[0031] Terminal device 130 is capable of communicating and / or connecting with ultrasound device 110, processing device 140, and / or storage device 150. For example, terminal device 130 can send one or more control commands to ultrasound device 110 via network 120 to control ultrasound device 110 to scan the object to be scanned according to the commands. In some embodiments, terminal device 130 includes one or any combination of mobile device 130-1, tablet computer 130-2, laptop computer 130-3, or other devices with input and / or output functions. In some embodiments, terminal device 130 is capable of remotely operating ultrasound device 110. In some embodiments, terminal device 130 is capable of operating ultrasound device 110 via wireless connection. In some embodiments, terminal device 130 is capable of receiving information and / or instructions input by a user and sending the received information and / or instructions to ultrasound device 110 or processing device 140 via network 120. In some embodiments, terminal device 130 is capable of receiving data and / or information from processing device 140. In some embodiments, terminal device 130 may be part of processing device 140. In some embodiments, the terminal device 130 may be integrated with the processing device 140 as an operating console for the ultrasound device 110. In some embodiments, the terminal device 130 may be omitted in application scenario 100.

[0032] Processing device 140 is capable of processing data and / or information obtained from ultrasound device 110, terminal device 130, and / or storage device 150. For example, processing device 140 can acquire a current frame ultrasound image from ultrasound device 110 and determine the inner and middle membrane boundaries of the current frame ultrasound image. In some embodiments, processing device 140 can send the processing results to terminal device 130. For example, processing device 140 can send the acquired current frame ultrasound image, inner and middle membrane boundaries, inner and middle membrane thickness, etc., to terminal device 130 and display them on one or more display devices in terminal device 130. In some embodiments, processing device 140 can be a single server or a server group. The server group can be centralized or distributed. In some embodiments, processing device 140 can be local or remote. For example, processing device 140 can access information and / or data stored in ultrasound device 110, terminal device 130, and / or storage device 150 via network 120. For example, processing device 140 can be directly connected to ultrasound device 110, terminal device 130, and / or storage device 150 to access their stored information and / or data. In some embodiments, processing device 140 can be executed on a cloud platform. In some embodiments, processing device 140 is part of ultrasound device 110 or terminal device 130.

[0033] Storage device 150 may store data, instructions, and / or other information. In some embodiments, storage device 150 may store data obtained from terminal device 130 and / or processing device 140. In some embodiments, storage device 150 may store data and / or instructions executed or used by processing device 140 for performing the exemplary methods described in this application. In some embodiments, storage device 150 may include one or more combinations of mass storage, removable storage, volatile read-write storage, read-only storage (ROM), etc. In some embodiments, storage device 150 may be implemented on a cloud platform.

[0034] In some embodiments, storage device 150 can connect to network 120 to communicate with one or more other components in application scenario 100 (e.g., processing device 140, terminal device 130, etc.). One or more components in application scenario 100 can access data or instructions stored in storage device 150 via network 120. In some embodiments, storage device 150 can directly connect to or communicate with one or more other components in application scenario 100 (e.g., processing device 140, terminal device 130, etc.). In some embodiments, storage device 150 may be part of processing device 140.

[0035] It is important to note that the application scenario 100 for the endometrial membrane measurement system is provided for illustrative purposes only and is not intended to limit the scope of this specification. Various changes and modifications can be made by those skilled in the art based on the description in this specification. For example, application scenario 100 may also include a database, information source, etc. Furthermore, application scenario 100 may be implemented on other devices to achieve similar or different functions. However, these changes and modifications will not depart from the scope of this specification.

[0036] Figure 2 This is a schematic diagram of the structure of an endometrial membrane measurement system 200 according to some embodiments of this specification.

[0037] like Figure 2 As shown, the inner membrane measurement system 200 includes an acquisition module 210, a classification module 220, an extraction module 230, a masking module 240, and a determination module 250. In some embodiments, the acquisition module 210, classification module 220, extraction module 230, masking module 240, and determination module 250 can be implemented by the processing device 140.

[0038] In some embodiments, the acquisition module 210 is used to acquire the region of interest in the ultrasound image.

[0039] In some embodiments, the classification module 220 is used to determine the intima measurement type based on the region of interest using a preset classification algorithm.

[0040] In some embodiments, the classification module 220 is further configured to: extract image features based on the region of interest to obtain image feature data; and classify the intima-media measurement type based on the image feature data to determine the intima-media measurement type.

[0041] In some embodiments, the classification module 220 is further configured to: extract the ultrasound image region corresponding to the region of interest, optimize the brightness distribution of the ultrasound image region, and determine a first image; perform multi-scale image reconstruction based on the first image to determine a second image; and extract image features based on the second image to determine image feature data.

[0042] In some embodiments, the extraction module 230 is used to extract the image to be measured from the current frame ultrasound image based on the region of interest. In some embodiments, the extraction module 230 is used to perform image feature optimization on the current frame ultrasound image based on the region of interest to obtain the image to be measured.

[0043] In some embodiments, the extraction module 230 is further configured to: expand the region of interest outward, and extract the image to be measured from the current frame ultrasound image according to the expanded region. In some embodiments, the extraction module 230 is further configured to: perform grayscale optimization and scale optimization on the image region corresponding to the expanded region in the ultrasound image to obtain the image to be measured.

[0044] In some embodiments, the mask module 240 is used to acquire a target mask image of the current frame ultrasound image based on the image to be measured.

[0045] In some embodiments, the mask module 240 is further configured to: extract gradients based on the image to be measured to obtain a gradient image; determine the estimation result of the inner and middle membrane boundaries of the gradient image based on the gradient image; obtain an initial mask image of the current frame ultrasound image based on the estimation result of the inner and middle membrane boundaries; and obtain a target mask image of the current frame ultrasound image by iteratively calculating using a preset contour optimization algorithm based on the initial mask image of the current frame ultrasound image.

[0046] In some embodiments, the mask module 240 is further configured to: extract image gradients based on the image to be measured to obtain a positive gradient image; extract phase gradients based on the image to be measured to obtain a phase gradient image; and determine a gradient image based on the positive gradient image and the phase gradient image.

[0047] In some embodiments, the mask module 240 is further configured to: determine the upper and lower boundary positions of each column of pixels in the gradient image based on the estimation results of the inner and middle membrane boundaries; and perform inner and middle membrane boundary undulation analysis based on the upper and lower boundary positions to determine the initial mask image of the current frame ultrasound image.

[0048] In some embodiments, the determining module 250 is used to determine the intima-media thickness parameters of the current frame ultrasound image based on the intima-media measurement type and the target mask image of the current frame ultrasound image.

[0049] In some embodiments, the determining module 250 is further configured to: generate an inner-middle membrane boundary based on a target mask image of the current frame ultrasound image and the region of interest; determine a first contour line of the outer membrane boundary and a second contour line of the inner membrane boundary based on the inner-middle membrane measurement type and the inner-middle membrane boundary; and determine the inner-middle membrane thickness parameter corresponding to the inner-middle membrane measurement type based on the first contour line and the second contour line.

[0050] For more information on the acquisition module 210, classification module 220, extraction module 230, masking module 240, and determination module 250, please refer to [link to documentation]. Figures 3-5 And its related descriptions.

[0051] In some embodiments, the intima-media measurement system 200 for arterial ultrasound images further includes a real-time measurement module ( Figure 2 (Not shown). In some embodiments, the real-time measurement module acquires the next frame of ultrasound image of the current frame; determines the target mask image of the next frame of ultrasound image based on the target mask image of the current frame of ultrasound image and the next frame of ultrasound image; and determines the intima-media thickness parameters of the next frame of ultrasound image based on the intima-media measurement type and the target mask image of the next frame of ultrasound image. In some embodiments, the real-time measurement module is further configured to: extract a first image region corresponding to the region of interest from the next frame of ultrasound image; determine an initial mask image of the next frame of ultrasound image based on the target mask image of the current frame of ultrasound image and the first image region; and determine the target mask image of the next frame of ultrasound image based on the initial mask image of the next frame of ultrasound image. For more details about the real-time measurement module, see [link to documentation]. Figure 6 And its related descriptions.

[0052] It should be noted that the above description of the inner membrane measurement system 200 and its modules is for convenience only and should not be construed as limiting this specification to the scope of the embodiments described. It is understood that those skilled in the art, after understanding the principle of the system, may arbitrarily combine the various modules or construct subsystems connected to other modules without departing from this principle. In some embodiments, the acquisition module 210, classification module 220, extraction module 230, mask module 240, and determination module 250 may be different modules within a single system, or a single module may implement the functions of two or more of the aforementioned modules. For example, the modules may share a storage module, or each module may have its own separate storage module. Such variations are all within the scope of this specification.

[0053] Figure 3 This is an exemplary flowchart of a method for measuring the intima-media layer of an arterial ultrasound image, according to some embodiments of this specification. In some embodiments, process 300 may be executed by the processing device 140 of an intima-media layer measurement system based on an arterial ultrasound image. Figure 3 As shown, process 300 includes the following steps.

[0054] Step 310: Acquire the region of interest in the ultrasound image. In some embodiments, step 310 may be performed by the acquisition module 210 or the processing device 140.

[0055] An ultrasound image includes at least one frame. In some embodiments, the ultrasound image may be a real-time image, i.e., an image obtained in real time by scanning with an ultrasound device. In some embodiments, the ultrasound image may also be a historical image stored in a storage device.

[0056] A region of interest (ROI) is a specific area in an ultrasound image that requires processing. For example, the ROI could be the region containing the inner membrane structure in an ultrasound image. In some embodiments, the ROI can be a region selected within a sampling frame. The sampling frame marks the ROI to distinguish it from other parts of the ultrasound image. A sampling frame is a graphical structure used to perform the selection task in an ultrasound image. A sampling frame can be a closed graphical structure with shapes such as rectangles, circles, or ellipses.

[0057] In some embodiments, the processing device can obtain the region of interest (ROI) of an ultrasound image in various ways. For example, the processing device can obtain the ROI of an ultrasound image through manual input. For instance, a user can interact with the user interface of a terminal device to select the region containing the carotid intima-media of an object as the ROI by drawing a sampling box. In some embodiments, the user is an operator of the arterial ultrasound intima-media measurement system 200, such as a doctor.

[0058] In some embodiments, the processing device can obtain parameter information (such as location, shape, area, size, etc.) of the region of interest based on the sampling frame.

[0059] Step 320: Based on the region of interest, determine the intima-media measurement type using a preset classification algorithm. In some embodiments, step 320 may be performed by the classification module 220 or the processing device 140.

[0060] Intima-media measurement type refers to the type of intima being measured. Intima-media measurement types can include anterior wall IMT, posterior wall IMT, and anterior-posterior wall IMT. Anterior wall IMT refers to measuring the thickness of the anterior wall of the intima-media, which is the intima-media structure closest to the ultrasound probe. Posterior wall IMT refers to measuring the thickness of the posterior wall of the intima-media, which is the intima-media structure furthest from the probe. Anterior-posterior wall IMT refers to simultaneously measuring the thickness of both the anterior and posterior walls of the intima-media.

[0061] In some embodiments, the processing device can determine the endothelial measurement type based on the region of interest (ROI) using various methods, including machine learning models, image matching, and feature algorithms. For example, the processing device can perform image matching in a database based on the ROI to obtain reference image data with high similarity, and determine the ROI measurement type corresponding to the reference image data as the ROI measurement type corresponding to the current ROI. The database can include preset correspondences between different reference image data and different ROI measurement types. The database can be determined based on prior knowledge or historical data.

[0062] In some embodiments, the preset classification algorithm includes: extracting image features based on the region of interest to obtain image feature data; and classifying the intima-media measurement type based on the image feature data to determine the intima-media measurement type.

[0063] Image feature data is data used to describe the features of a region of interest. In some embodiments, image feature data includes geometric features, color features, texture features, shape features, histogram features, etc., of the region of interest. In some embodiments, image feature data includes grayscale distribution features of the region of interest. Grayscale distribution features can be used to represent the number of pixels with each grayscale value in the region of interest, and / or the distribution of grayscale values ​​of each pixel in the region of interest.

[0064] In some embodiments, the processing device can extract image features in various ways. For example, the processing device can encode the region of interest to obtain encoded data as image feature data. Image encoding methods include, but are not limited to, compression coding, transform coding, and predictive coding. For example, the processing device can extract image features from the region of interest using image feature extraction algorithms. Exemplary image feature extraction algorithms include, but are not limited to, SIFT feature extraction, HOG feature extraction, LBP feature extraction, and Haar feature extraction. This specification does not specifically limit the method of image feature extraction in the embodiments; any operation well-known to those skilled in the art can be used.

[0065] In some embodiments, the processing device may extract the ultrasound image region corresponding to the region of interest, optimize the brightness distribution of the ultrasound image region, and determine a first image; perform multi-scale image reconstruction based on the first image to determine a second image; and extract image features based on the second image to determine image feature data.

[0066] An ultrasound image region refers to a portion of the ultrasound image located within the sampling frame.

[0067] In some embodiments, the ultrasound image region may be an image region comprising at least one or any combination of the carotid artery, the carotid artery wall, the intima, media and adventitia of the vessel wall and the external structures of the vessel.

[0068] The first image refers to the image obtained after optimizing the brightness and darkness distribution of the ultrasound image region.

[0069] Brightness and darkness distribution optimization refers to methods that adjust the exposure, contrast, and color of ultrasound image regions. By optimizing brightness and darkness distribution, the grayscale differences of blood flow and blood vessel walls in ultrasound image regions can be enhanced.

[0070] In some embodiments, the processing device can optimize the brightness and darkness distribution in various ways. For example, the processing device can optimize the brightness and darkness distribution of the ultrasound image region by histogram equalization, percentage stretching, etc., to obtain a first image.

[0071] The second image refers to the image obtained after multi-scale image reconstruction of the first image.

[0072] Multi-scale image reconstruction is a method of reconstructing images through multi-scale transformations. These multi-scale transformations include wavelet transforms, image interpolation, and image reconstruction.

[0073] In some embodiments, the processing device can filter out the detail information of the first image through multi-scale image reconstruction to obtain a second image that retains grayscale transition features. The detail information includes at least one of the color features, texture features, and shape features of the first image. Grayscale transition features are features related to the grayscale differences between blood flow and the vessel wall.

[0074] In some embodiments, the processing device can perform multi-scale image reconstruction in various ways. For example, the processing device can filter detailed information of the first image using an image filter. The image filter may include a Gaussian filter, a mean filter, or other blur filters.

[0075] In some embodiments, the processing device can alternately and repeatedly apply multiple different multi-scale image reconstruction methods to a first image to obtain a second image. These multi-scale image reconstruction methods include image pyramids, Laplacian pyramids, etc. Image resampling operators can include nearest neighbor interpolation, bilinear interpolation, bicubic interpolation, etc., or any combination thereof. Alternating repetition refers to processing multiple different multi-scale image reconstruction methods in a preset order. The preset order can be manually predetermined.

[0076] In some embodiments, before performing multi-scale image reconstruction on the first image, the classification module 220 may adjust the size of the first image to obtain a first image of a preset size; and perform multi-scale image reconstruction on the first image of the preset size to obtain a second image. The preset size may be a manually preset value, a system default value, etc.

[0077] In some embodiments of this specification, by optimizing the light and dark distribution, the light and dark details of the ultrasound image area can be improved and the gray-scale difference between blood flow and blood vessel wall in the ultrasound image area can be enhanced; by performing multi-scale image reconstruction on the first image, unnecessary details can be filtered out, more noteworthy content can be retained, and the resolution and clarity of the second image can be guaranteed.

[0078] In some embodiments, the processing device can classify the intima-media measurement type based on image feature data using various methods to determine the intima-media measurement type. In some embodiments, the processing device can classify the intima-media measurement type based on the grayscale distribution characteristics of the region of interest. In some embodiments, the processing device can determine the relative position of the blood vessel within the region of interest based on the grayscale distribution characteristics of the region of interest; and determine the intima-media measurement type based on the relative position of the blood vessel. Since the grayscale values ​​of pixels inside the blood vessel are lower than those outside the blood vessel, the processing device can determine the relative position of the blood vessel based on the position of the portion with a grayscale value lower than a preset grayscale threshold within the region of interest. The preset grayscale threshold can be a system default value, an empirical value, a manually preset value, or any combination thereof, and can be set according to actual needs; this specification does not impose any restrictions on it.

[0079] In some embodiments, the relative position of the blood vessel includes any one of the following: the lower part, the middle part, and the upper part of the region of interest (ROI). The upper part of the RIO refers to the area that partially or completely includes the upper edge of the RIO, where the upper edge is the side edge closest to the ultrasound probe. The lower part of the RIO refers to the area that partially or completely includes the lower edge of the RIO, where the lower edge is the side edge closest to the ultrasound probe. The middle part of the RIO refers to the area other than the upper and lower parts. When the relative position of the blood vessel is in the middle of the RIO, the grayscale value in the middle of the RIO is lower than the grayscale values ​​in the upper and lower parts. When the relative position of the blood vessel is in the upper part of the RIO, the grayscale value in the upper part of the RIO is lower than the grayscale value in the lower part.

[0080] In some embodiments, the processing device may determine the intima-media measurement type as anterior wall IMT in response to the vessel being located below the region of interest; determine the intima-media measurement type as posterior wall IMT in response to the vessel being located above the region of interest; and determine the intima-media measurement type as anterior or posterior wall IMT in response to the vessel being located in the middle of the region of interest. This is merely an example. Figure 5 As shown, in region of interest 510 of Figure A, the gray value of the upper region is lower than that of the lower region, indicating that the relative position of the blood vessel in this figure is located in the upper part of region of interest 510. Therefore, the intima-media measurement type of region of interest 510 can be determined to be posterior wall IMT. In region of interest 520 of Figure B, the gray value of the upper region is higher than that of the lower region, indicating that the relative position of the blood vessel in this figure is located in the lower part of region of interest 520. Therefore, the intima-media measurement type of region of interest 520 can be determined to be anterior wall IMT. Similarly, the intima-media measurement type of region of interest 530 in Figure C can be determined to be posterior wall IMT, the intima-media measurement type of region of interest 540 in Figure D can be determined to be anterior and posterior wall IMT, the intima-media measurement type of region of interest 550 in Figure E can be determined to be anterior wall IMT, and the intima-media measurement type of region of interest 560 in Figure F can be determined to be anterior wall IMT.

[0081] In some embodiments, the preset classification algorithm may further include: processing the region of interest using a classification model to determine the endometrial measurement type. In some embodiments, the classification model is a machine learning model, such as a support vector machine classifier, decision tree, or any combination thereof. In some embodiments, the input to the classification model includes the region of interest and / or an ultrasound image containing the region of interest, and the output is the endometrial measurement type. In some embodiments, the classification model can be trained using a large number of labeled training samples through various feasible methods, such as updating parameters based on gradient descent. In some embodiments, the training samples include the sample region of interest and / or an ultrasound image containing the sample region of interest, labeled as the endometrial measurement type of the sample region of interest. The training samples can be determined based on historical data, and the labels can be determined by the system or manually based on historical data. For example, the endometrial measurement type can be labeled by the system or manually using the methods described in the foregoing embodiments.

[0082] In some embodiments of this specification, classifying the type of endometrial measurement using image feature data can improve classification accuracy. Compared to manual classification, this increases classification efficiency while saving human resources. Furthermore, utilizing classification models can further enhance both the accuracy and efficiency of classification.

[0083] In some embodiments, when the processing device determines that the intima-media measurement type of the region of interest cannot be determined, it may trigger an instruction to reacquire ultrasound images and / or re-determine the region of interest, and terminate the current process.

[0084] Step 330: Extract the image to be measured from the current frame ultrasound image based on the region of interest. In some embodiments, step 330 may be performed by extraction module 230.

[0085] The current frame ultrasound image refers to the ultrasound image for which intima-media measurement is required at the current moment. For example, the current frame ultrasound image could be a frame selected from historical ultrasound images. Alternatively, it could be a frame of ultrasound image acquired in real-time by the ultrasound equipment at the current moment.

[0086] The image to be measured refers to an image containing the intima region selected by the user.

[0087] In some embodiments, the processing device can extract the image to be measured from the current frame ultrasound image in various ways based on the region of interest. For example, the processing device can directly use the region of interest as the image to be measured.

[0088] In some embodiments, the processing device may adjust the region of interest and extract the image to be measured from the current frame ultrasound image based on the adjusted region.

[0089] The adjusted region refers to the region obtained after adjusting the sampling frame of the region of interest. In some embodiments, the adjustment process may include, but is not limited to, one or more of the following: expansion, reduction, and movement.

[0090] In some embodiments, the processing device may expand the region of interest and extract the image to be measured from the current frame of ultrasound image based on the expanded region. The expanded region refers to the area obtained after expanding the sampling frame of the region of interest.

[0091] In some embodiments, the sampling frame can be expanded outward according to a preset ratio through magnification. For example, when the sampling frame is rectangular, the long side and / or the wide side of the rectangular sampling frame can be expanded outward according to a preset ratio. As another example, when the sampling frame is circular, the diameter of the circular sampling frame can be expanded outward according to a preset ratio. The preset ratio can be a system preset value, a system default value, etc., and the preset ratio is greater than 1. The outward expansion method can differ for sampling frames of different shapes, and can be set according to actual needs.

[0092] In some embodiments, the processing device may determine the selected area of ​​the expanded sampling frame as the extended region, and crop the current frame ultrasound image based on the extended region to obtain the image to be measured.

[0093] In some embodiments of this specification, by expanding the region of interest outward, a larger measurement area can be obtained than that selected manually, effectively avoiding the occurrence of unclear boundaries between the inner and middle membranes within the sampling frame, and thus improving the clarity of the inner and middle membrane boundaries.

[0094] In some embodiments, the processing device can perform grayscale optimization and multi-scale optimization on the image region corresponding to the adjusted region in the ultrasound image to obtain the image to be measured.

[0095] The image region corresponding to the adjusted region refers to the image region extracted from the current frame of the ultrasound image based on the adjusted region.

[0096] In some embodiments, the processing device can perform grayscale optimization and multi-scale optimization on the image region corresponding to the adjusted region in the ultrasound image in various ways. For example, the processing device can perform grayscale optimization on the image region corresponding to the adjusted region through histogram equalization, percentage stretching, etc. As another example, the processing device can perform multi-scale optimization on the image region corresponding to the adjusted region through algorithms such as USM sharpening enhancement, Laplace sharpening, and frequency domain sharpening.

[0097] It should be noted that single-scale sharpening schemes have limited sharpening effects. If the parameters are not set properly, over-sharpening can easily occur, increasing image noise and affecting subsequent identification of the inner and middle membrane boundaries. Using a multi-scale sharpening scheme with multiple scales will affect the algorithm's execution efficiency.

[0098] In some embodiments, the processing device can perform multi-scale optimization on the image region corresponding to the adjusted region using the Laplacian pyramid image processing method to obtain the image to be measured. For example, the processing device can process the image region corresponding to the adjusted region using the Laplacian pyramid transform to obtain image edge information; and then, through the inverse Laplacian pyramid transform, superimpose the extracted image edge information onto the image region corresponding to the adjusted region to obtain the image to be measured.

[0099] In some embodiments of this specification, by performing grayscale optimization on the image region corresponding to the adjusted region, the brightness distribution of the image region corresponding to the adjusted region can be optimized. By performing multi-scale optimization on the image region corresponding to the adjusted region, the sharpness features of the image can be further improved, and the sharpness information of boundary regions such as LII and MAI in the image can be increased, thereby improving the image clarity. By using the Laplacian pyramid image processing scheme and adopting the idea of ​​multi-level pyramids to perform image sharpening processing, the execution efficiency of the algorithm can be optimized, the image clarity can be improved, and it is beneficial to subsequently determine the clear inner and middle membrane boundaries of the current frame ultrasound image.

[0100] Step 340: Based on the image to be measured, obtain the target mask image of the current frame ultrasound image. In some embodiments, step 340 may be performed by the mask module 240.

[0101] A target mask image is an image obtained by partially obscuring a portion of an ultrasound image. In some embodiments, the target mask image may be an image that obscures non-critical content in the ultrasound image, retaining only the critical content. The critical content may include the intima-media boundary. In some embodiments, the critical content may also include vascular portions. The target mask image may be in various forms, such as a binary image or a grayscale image.

[0102] In some embodiments, the target mask image can be a binary image generated based on the intima-media boundary of the image to be measured, representing the contour information of the intima-media boundary of the image to be measured. In some embodiments, the target mask image is a binary image with the same size as the image to be measured, and can be represented by a matrix, which includes two values: 0 and 1. Here, 0 represents other tissue regions, and 1 represents non-uniform tissue regions (i.e., the contour information of the intima-media boundary). In some embodiments, the target mask image can also be in the form of a multi-valued image (where 0 represents the background, and different values ​​represent different types of vascular membranes).

[0103] In some embodiments, the processing device can acquire the target mask image of the current frame ultrasound image based on the image to be measured through various methods. For example, the mask module 240 can use an image segmentation algorithm to segment the image to be measured to obtain the target mask image of the current frame ultrasound image. Image segmentation algorithms include, but are not limited to, thresholding and region growing methods. Alternatively, the mask module 240 can also use a segmentation model to segment the image to be measured to obtain the target mask image of the current frame ultrasound image. The segmentation model can be a trained machine learning model, such as a deep learning model based on the VB-Net network.

[0104] In some embodiments, the processing device may perform gradient extraction on the image to be measured to obtain a gradient image, determine the estimated boundary of the inner and middle membranes based on the gradient image, and then determine the target mask image. For further explanation on determining the target mask image, see [link to documentation]. Figure 4 And its related descriptions.

[0105] Step 350: Based on the intima-media measurement type, determine the intima-media thickness parameters of the current frame ultrasound image according to the target mask image of the current frame ultrasound image. In some embodiments, step 350 may be performed by the determination module 250.

[0106] Inner membrane thickness parameters refer to parameters related to the thickness of the inner membrane boundary. For example, inner membrane thickness parameters may include the maximum, minimum, mean, and standard deviation of the inner membrane boundary thickness.

[0107] The intima-media boundary refers to the dividing line between the arterial intima and the arterial media in the image to be measured.

[0108] In some embodiments, the inner-middle membrane boundary includes the luminal-intima interface (LII) and the media-adventitia interface (MAI).

[0109] In some embodiments, the processing device can determine the inner-middle membrane boundary of the current frame ultrasound image in various ways based on the inner-middle membrane measurement type and the target mask image of the current frame ultrasound image. For example, the processing device can determine the inner-middle membrane boundary in the target mask image of the current frame ultrasound image based on the grayscale distribution characteristics of different inner-middle membrane measurement types (e.g., anterior wall IMT, posterior wall IMT, and anterior and posterior wall IMT). For example, when the intima-media measurement type is anterior wall IMT, the processing device can analyze each column of the target mask image. Starting from the first pixel of each column, it searches from top to bottom until the gray value of the pixel in that column is 255, records the information of that pixel, and uses it as the boundary point of the intima-media. The boundary points of the intima-media and intima of all columns are determined as the boundaries of the intima-media of the current frame of the ultrasound image. Then, it searches from bottom to top until the gray value of the pixel in that column is less than 255 or is 0, records the information of that pixel, and uses it as the boundary point of the luminal endothelial. The boundary points of the luminal endothelial of all columns are determined as the boundaries of the luminal endothelial of the current frame of the ultrasound image. As another example, when the intima-media measurement type is posterior wall IMT, the processing device can analyze each column of the target mask image. Starting from the first pixel of each column, it searches from top to bottom until the gray value of a pixel in that column is not 255, records the information of that pixel, and uses it as the boundary point of the luminal endothelial membrane. The boundary points of the luminal endothelial membranes in all columns are then used to determine the luminal endothelial membrane boundaries of the current frame of the ultrasound image. Then, it searches from bottom to top until the gray value of a pixel in that column is 255, records the information of that pixel, and uses it as the boundary point of the intima-media and epithelial membranes. The boundary points of the intima-media and epithelial membranes in all columns are then used to determine the intima-media and epithelial membrane boundaries of the current frame of the ultrasound image. For another example, the processing device can perform preliminary contour extraction of LII and MAI based on the Canny operator, Hough transform, etc., and then input iterative optimization of the Snake model to output the final LII and MAI contours. The embodiments in this specification do not have special limitations on the method of contour boundary extraction; operations familiar to those skilled in the art can be used.

[0110] In some embodiments, the processing device may also generate the inner-middle membrane boundary based on the target mask image and region of interest of the current frame ultrasound image; determine the first contour line of the outer membrane boundary and the second contour line of the inner membrane boundary based on the inner-middle membrane measurement type and the inner-middle membrane boundary; and determine the inner-middle membrane thickness parameter corresponding to the inner-middle membrane measurement type based on the first contour line and the second contour line.

[0111] The first profile line is used to reflect the shape of the outer membrane boundary (MAI). The second profile line is used to reflect the shape of the inner membrane boundary (LII).

[0112] In some embodiments, the processing device can connect the boundary points of the inner, middle, and outer membranes of the target mask image to obtain a first contour line in the target mask image; and connect the boundary points of the inner and middle membranes to obtain a second contour line in the target mask image. For example... Figure 5 As shown, in Figure A, the intima-media measurement type of region 510 is posterior wall IMT, 511 is the second contour line formed by connecting the boundary points of the luminal intima, and 512 is the first contour line formed by connecting the boundary points of the medial and lateral membranes; in Figure B, the intima-media measurement type of region 520 is anterior wall IMT, 521 is the first contour line formed by connecting the boundary points of the medial and lateral membranes, and 522 is the second contour line formed by connecting the boundary points of the luminal intima; and so on. Therefore, in Figure C, 531 is determined to be the second contour line formed by connecting the boundary points of the luminal intima, and 532 is the second contour line formed by connecting the boundary points of the medial and lateral membranes. First contour line; In Figure D, 541 is the first contour line formed by connecting the boundary points of the middle and outer membranes, 542 is the second contour line formed by connecting the boundary points of the inner membrane, 543 is the second contour line formed by connecting the boundary points of the inner membrane, and 544 is the first contour line formed by connecting the boundary points of the middle and outer membranes; In Figure E, 551 is the first contour line formed by connecting the boundary points of the middle and outer membranes, and 552 is the second contour line formed by connecting the boundary points of the inner membrane; In Figure F, 561 is the first contour line formed by connecting the boundary points of the middle and outer membranes, and 562 is the second contour line formed by connecting the boundary points of the inner membrane.

[0113] In some embodiments, the processing device can obtain the first contour line and the second contour line in the target mask image based on a boundary tracking algorithm. In some embodiments, the processing device can crop the target mask image based on the position, size, and shape of the region of interest, perform logical operations on the region of interest in the current frame ultrasound image based on the cropped target mask image, obtain the first contour line and the second contour line within the region of interest, and send the logically processed current frame ultrasound image to the display interface of the terminal device for display. The logical operation can be a pixel-by-pixel AND operation.

[0114] In some embodiments, the processing device can determine the IMT thickness corresponding to the IMT measurement type in various ways based on the first and second contour lines. For example, the processing device can calculate the distance between pixels on the first contour line and pixels on the second contour line in each column based on the first and second contour lines of the current frame ultrasound image, taking the maximum distance as the maximum IMT thickness, the minimum distance as the minimum IMT thickness, and the average distance of all columns as the average IMT thickness. Alternatively, the processing device can take the standard deviation of the distances of all columns as the standard deviation of the IMT. Another example is that the processing device can take the number of pixels between the endpoints of the first contour line as the length of the IMT / outer membrane boundary within the sampling frame.

[0115] In some embodiments of this specification, by drawing a first contour line and a second contour line, the boundary between the inner and outer membranes, the shape and position of the inner and outer membranes can be displayed intuitively, improving the user experience, making it easier for users to select measurement locations of interest, and improving the flexibility and convenience of measurement.

[0116] In some embodiments, the processing device may perform the intima-media measurement method for arterial ultrasound images as shown in some embodiments of this specification on each frame of ultrasound image acquired by the ultrasound device. In some embodiments, the processing device may perform the intima-media measurement method for arterial ultrasound images as shown in some embodiments of this specification sequentially on each frame of ultrasound image acquired by the ultrasound device in chronological order. In some embodiments, after determining the intima-media boundary and intima-media thickness of the current frame of ultrasound image, the processing device may perform intima-media measurement on the next frame of ultrasound image in the manner described in process 600. See further details. Figure 6 And its related descriptions.

[0117] In some embodiments of this specification, determining the intima-media measurement type through the region of interest (IMT) saves time associated with manual judgment, eliminates the influence of operator bias, improves accuracy, and facilitates subsequent determination of the IMT boundary. By determining the target mask image, the IMT can be precisely segmented, improving the accuracy of IMT measurement. The target mask image of the current frame of ultrasound image provides precise contour information of the IMT boundary, which is beneficial for determining the target mask image for the next frame, improving both computational accuracy and efficiency. For ultrasound images, due to the high frame rate requirements, using a target mask image significantly improves efficiency.

[0118] Figure 4This is an exemplary flowchart illustrating the determination of a target mask image according to some embodiments of this specification. In some embodiments, process 400 may be performed by the processing device 140 or mask module 240 of an intima-media measurement system based on arterial ultrasound images. Figure 4 As shown, process 400 includes the following steps.

[0119] Step 410: Extract gradients based on the image to be measured to obtain a gradient image.

[0120] Gradient images are used to describe the differences or changes between individual pixels in an ultrasound image. For example, a gradient image can be a matrix composed of the image gradients of each pixel in an ultrasound image. The image gradient represents the trend of pixel change in different directions. In some embodiments, for each pixel, the image gradient of the image to be measured at that pixel can be determined based on the pixel value and the pixel values ​​of its neighboring pixels. Neighboring pixels refer to pixels located within the vicinity of a given pixel. Neighborhoods include 4-neighborhoods, 8-neighborhoods, diagonal neighborhoods, etc. Accordingly, neighboring pixels include pixels within the 4-neighborhood, 8-neighborhood, or diagonal neighborhood of a given pixel.

[0121] In some embodiments, the processing device can perform gradient extraction in various ways. For example, the processing device can perform convolution operations on the image to be measured based on operators such as Sobel, Prewitt, and Laplace to obtain a gradient image.

[0122] In some embodiments, the image gradient of each pixel can be represented by a vector (B1, B2, ...), where vector element B1 represents the gradient of pixel B in the horizontal direction and vector element B2 represents the gradient of pixel B in the vertical direction.

[0123] In some embodiments, the processing device performs image gradient extraction based on the image to be measured to obtain a positive gradient image; performs phase gradient extraction based on the image to be measured to obtain a phase gradient image; and determines a gradient image based on the positive gradient image and the phase gradient image.

[0124] A positive gradient image is a gradient image determined based on the positive gradient of the image to be measured. A positive gradient means that the pixel value of a pixel increases along the gradient direction, i.e., the brightness or intensity of the pixel value increases along the gradient direction. In some embodiments, the gradient direction for image gradient extraction can be the ultrasound detection direction or the opposite direction of the ultrasound detection direction. In some embodiments, the gradient direction for image gradient extraction can be the Y direction of the sampling frame or the opposite direction of the Y direction. The Y direction of the sampling frame can be the direction of the frame line that is perpendicular or approximately perpendicular to the blood flow direction (e.g., the angle is greater than 45 degrees and less than 90 degrees), and this direction is close to the ultrasound probe.

[0125] In some embodiments, the processing device can perform image gradient extraction in various ways. For example, the processing device can calculate the pixel value change of each pixel in the image to be measured in the ultrasonic detection direction or the Y direction to obtain an initial gradient image, traverse each pixel in the initial gradient image, and remove pixels with negative gradient directions to obtain a positive gradient image.

[0126] In some embodiments, the processing device can perform convolution operations on the image to be measured using the Prewitt operator in the ultrasonic detection direction or the Y direction, and the Sobel operator in the vertical direction, to obtain an initial gradient image.

[0127] A phase gradient image is a gradient image determined based on the rate of phase change of the image to be measured.

[0128] In some embodiments, the processing device can extract the phase gradient in various ways. For example, the processing device can calculate the frequency matrix and homography matrix of the image to be measured, perform a Fourier transform on the frequency matrix and homography matrix, perform multi-scale transform processing, and then filter using a LogGabor filter. The filtered result is processed using an inverse Fourier transform, and the local energy is calculated using a Hilbert transform. After processing with a Rayleigh distribution model, a symmetrical energy map is obtained, which is the phase gradient image.

[0129] In some embodiments, the processing device can determine the gradient image based on the forward gradient image and the phase gradient image in various ways. For example, the processing device can multiply each pixel of the forward gradient image and the phase gradient image to obtain the gradient image.

[0130] In some embodiments of this specification, the edge features of the image to be measured can be effectively preserved through the phase gradient image; the frequency characteristics of the extracted image to be measured are not easily affected by changes in image contrast, and relatively complete image features can be obtained under complex conditions (e.g., changes in brightness); by multiplying the corresponding pixels of the phase gradient image and the positive gradient image to obtain the gradient image, the common boundary information in the phase gradient image (frequency domain information) and the positive gradient image (image information) can be preserved, the interference of other boundaries can be suppressed, and it is beneficial to the subsequent estimation of the inner and middle membrane boundaries.

[0131] Step 410: Based on the gradient image, determine the estimation result of the inner membrane boundary of the gradient image.

[0132] The estimation result of the inner and middle membrane boundary refers to the preliminary determination of the inner and middle membrane boundary in the gradient image, i.e., the estimated inner and middle membrane boundary. The estimation result of the inner and middle membrane boundary may include the estimation result of LII and / or MAI.

[0133] In some embodiments, the processing device determines the estimation result of the inner and middle membrane boundaries of the gradient image through various methods based on the gradient image. In some embodiments, the processing device can analyze each column or each row of the gradient image to obtain the gradient distribution of each column or each row, and use the gradient value that meets the preset condition as the target threshold of that column or row; perform binarization processing on the gradient of the pixels in that column or row based on the target threshold, and obtain the binary image corresponding to the gradient image based on the result of the binarization processing of each column or each row; determine the estimation result of the inner and middle membrane boundaries in the binary image corresponding to the gradient image. Here, the gradient distribution refers to the distribution of gradient values ​​of different pixels in a certain column.

[0134] Preset conditions refer to the evaluation conditions for selecting the target threshold. For example, a preset condition is that the occurrence frequency of a certain gradient value reaches a preset number. Another preset condition is that the occurrence percentage of a certain gradient value is greater than or equal to a percentage threshold (e.g., 90%). The preset number of occurrences and the percentage threshold can be determined based on experiments or experience.

[0135] Generally, the gradient value is larger at the boundary of the inner and middle membranes, so the binarization result of pixels with gradient values ​​greater than or equal to the target threshold can be set to a maximum gray value (e.g., 255); the gradient value is smaller at non-inner and middle membrane boundaries, so the binarization result of pixels with gradient values ​​less than the target threshold can be set to a minimum gray value (e.g., 0).

[0136] It should be noted that due to noise interference and other factors, pixels not located at the inner-middle membrane boundary may result in a grayscale maxima in their binarization. In some embodiments, the processing device can perform connected component analysis on the binary image corresponding to the gradient image; based on the results of the connected component analysis, the binary image corresponding to the gradient image is updated or corrected to determine the target binary image of the gradient image. Accordingly, the processing device can determine the estimated inner-middle membrane boundary from the target binary image corresponding to the gradient image.

[0137] A connected component, also known as a connected region, refers to a region in a binary image corresponding to a gradient image that consists of pixels with the same gray value and adjacent positions. In some embodiments, a connected component can be a region in a binary image corresponding to a gradient image that consists of multiple pixels with a maximum gray value and adjacent positions.

[0138] In some embodiments, the processing device may employ existing connected component extraction techniques to extract connected components in the binary image corresponding to the gradient image. For example, the processing device may use a pixel-by-pixel comparison algorithm to extract connected components in the binary image corresponding to the gradient image.

[0139] In some embodiments, connected component analysis includes: classifying connected components based on their area; retaining a connected component if its area is greater than or equal to a first area threshold; removing a connected component if its area is less than a second area threshold (the second area threshold is less than the first area threshold); and performing a narrowing analysis on the connected component if its area is less than the first area threshold but greater than or equal to the second area threshold.

[0140] In some embodiments, preserving the connected component may mean preserving the original grayscale values ​​of all pixels in the connected component. Removing the connected component may mean updating the original grayscale values ​​(e.g., grayscale maxima) of all pixels in the connected component to their opposite values ​​(e.g., grayscale minima).

[0141] In some embodiments, the processing device may perform a narrow length analysis based on the characteristics of the inner membrane boundary. The characteristics of the inner membrane boundary are features reflecting the shape and size of the inner membrane boundary. For example, the characteristics of the inner membrane boundary include its length, width, and / or aspect ratio.

[0142] In some embodiments, narrow length analysis includes: analyzing whether the length of the connected components meets the length requirement, retaining the connected components that meet the length requirement, and removing the connected components that do not meet the length requirement. The length requirement may be that the length of the connected component is between a set upper and lower limit, where the lower limit can be determined based on experience or historical data, and the upper limit can be determined based on the size of the sampling frame. In some embodiments, narrow length analysis includes: analyzing whether the width of the connected components meets the width requirement, retaining the connected components that meet the width requirement, and removing the connected components that do not meet the width requirement. The width requirement is similar to the length requirement and will not be described again here. In some embodiments, narrow length analysis includes: analyzing whether the aspect ratio of the connected components meets the ratio requirement, retaining the connected components that meet the ratio requirement, and removing the connected components that do not meet the ratio requirement. The ratio requirement is that the aspect ratio of the connected components is between a set upper and lower limit, where the upper and lower limits can be determined based on experience or historical data.

[0143] In some embodiments, the processing device may determine a target binary image of the gradient image based on the results of connected component analysis. The target binary image can be used to determine the estimated boundary of the inner and middle membranes.

[0144] In some embodiments, the processing device may determine an initial binary image of the gradient image based on the results of connected component analysis; and perform filtering analysis based on the initial binary image to determine a target binary image of the gradient image. In some embodiments, the filtering analysis includes: determining the ratio of the area of ​​the sampling frame to the area of ​​each connected component in the initial binary image; removing connected components in the initial binary image whose area ratio is less than a preset ratio threshold; retaining connected components in the initial binary image whose area ratio is greater than or equal to the preset ratio threshold; and obtaining the target binary image of the gradient image.

[0145] In some embodiments, the processing device can connect adjacent pixels corresponding to gray-level maxima in the target binary image to obtain an estimation result of the inner and middle membrane boundary. In some embodiments, the processing device can also perform interpolation processing on the target binary image (e.g., changing the gray-level minima between two adjacent connected components to gray-level maxima), and connect adjacent pixels corresponding to the gray-level maxima to obtain an estimation result of the inner and middle membrane boundary. Through interpolation processing, connected components belonging to the same inner and middle membrane boundary but located in different positions can be connected to form a "large connected region," improving the accuracy of the determined inner and middle membrane boundary estimation result.

[0146] Step 430: Based on the estimation results of the inner membrane boundary, obtain the initial mask image of the current frame ultrasound image.

[0147] The initial mask image refers to the mask image that is initially determined and used to determine the target mask image.

[0148] In some embodiments, the processing device can obtain an initial mask image of the current frame ultrasound image based on the estimation result of the inner and middle membrane boundaries through various methods. For example, the processing device can multiply the pixels corresponding to the estimation result of the inner and middle membrane boundaries with the corresponding pixels of the image to be measured to obtain the initial mask image. In the initial mask image, the pixel values ​​of the inner and middle membrane boundary portion remain unchanged, while the pixel values ​​of the remaining areas are 0.

[0149] In some embodiments, the processing device can perform smoothing processing based on the estimated inner-middle membrane boundary. This involves multiplying the pixels corresponding to the smoothed inner-middle membrane boundary with the corresponding pixels of the image to be measured to obtain an initial mask image. The smoothing processing includes denoising, noise reduction, and filtering of the estimated inner-middle membrane boundary. Smoothing processing can make the estimated inner-middle membrane boundary smoother, reducing the impact of noise and oscillations.

[0150] In some embodiments, the processing device can smooth the estimated inner and middle membrane boundaries in various ways. For example, the processing device can smooth the estimation results of the inner and middle membrane boundaries using methods such as weighted average, exponential average, and SG filtering.

[0151] In some embodiments, the processing device can determine the upper and lower boundary positions of each column of pixels in the gradient image based on the estimation results of the inner and middle membrane boundaries; and perform inner and middle membrane boundary undulation analysis based on the upper and lower boundary positions to determine the initial mask image of the current frame ultrasound image.

[0152] The upper and lower boundary positions refer to the positions of LII and MAI within each column of pixels in the gradient image. In some embodiments, the upper and lower boundary positions include the upper boundary position and the lower boundary position. The upper boundary position refers to the position of a pixel within a column relative to the upper boundary. The lower boundary position refers to the position of a pixel within a column relative to the lower boundary.

[0153] The boundary type varies depending on the type of intima-media measurement, with the upper or lower boundary positions differing. The boundary type is either the intima boundary or the epicardium boundary. For example, when the intima-media measurement type is anterior wall IMT, the upper boundary position is the LII (Limited Intima Inlet) and the lower boundary position is the MAI (Mass Intima Inlet). When the intima-media measurement type is posterior wall IMT, the upper boundary position is the MAI and the lower boundary position is the LII.

[0154] In some embodiments, the processing device can analyze the gradient values ​​of each column of pixels in the gradient image to determine the upper and lower boundary positions of each column of pixels. For example, the processing device can analyze each column of the gradient image, starting from the first pixel of each column, searching from top to bottom or from bottom to top until the gradient of the pixels in that column reaches its peak, recording the position information of the two pixels, and finally determining the upper and lower boundary positions based on the intima-media measurement type.

[0155] In actual ultrasound examination scenarios, the boundary of the intima-media is continuous in ultrasound images without significant boundary undulations. Therefore, it is necessary to perform intima-media boundary undulation analysis to meet the needs of practical applications.

[0156] In some embodiments, the processing device can perform intima-media boundary undulation analysis in various ways to determine the initial mask image of the current frame ultrasound image. For example, the processing device can perform statistical analysis on the upper and lower boundary positions of each column of pixels in the gradient image to obtain the change amplitude data of the upper or lower boundary position respectively; for the upper or lower boundary positions with a change amplitude greater than an amplitude threshold, smoothing processing is performed based on morphological transformation. The amplitude threshold can be a manually preset value or a system default value.

[0157] Variation amplitude data is used to reflect changes in the boundary positions of different columns. For example, variation amplitude data can include a sequence of variation amplitudes of the upper / lower boundary positions of different columns. Variation amplitude refers to the amount of change of a boundary position relative to a specified object. A specified object refers to boundary positions with a specified adjacency relationship or the average of all boundary positions of the same type. A specified adjacency relationship refers to the adjacency relationship between any boundary position and at least one other boundary position. For example, a specified adjacency relationship refers to a boundary position directly adjacent to its left side, such as a boundary position directly on its left.

[0158] In some embodiments, the processing device may perform morphological transformations based on morphological operators. Morphological operators include, but are not limited to, erosion, dilation, opening, and closing operations. The boundary of the inner membrane is generally relatively flat and smooth, but due to the influence of noise, small burrs may exist near the surface of the inner membrane in the initial mask image. Therefore, it is necessary to process the initial mask image based on morphological operators, such as erosion followed by dilation, to eliminate small burrs and smooth the boundary of larger objects without significantly changing their area.

[0159] It should be noted that gradient extraction and inner membrane boundary estimation can yield a mask image suitable for general scenarios. However, when the gradient information is not significant, leading to an underestimation of the inner membrane boundary, or when excessive interference noise causes an overestimation, or when there are holes or horizontal depressions in the gradient image, the accuracy of the inner membrane boundary estimation may differ significantly from the true value, resulting in low system stability and robustness.

[0160] In some embodiments of this specification, by performing undulation analysis of the intima-media boundary, the situation where the boundary suddenly increases is suppressed, and the situation where the boundary suddenly decreases is mitigated, so as to ensure the normal transition of the boundary. By using morphological operators to optimize and adjust the situation where the boundary fluctuation is large, higher quality and more realistic estimation results of the intima-media boundary can be obtained, avoiding the handling of other complex situations (such as noise, plaque, etc.) and helping to improve the robustness of the system.

[0161] Step 440: Based on the initial mask image of the current frame ultrasound image, the target mask image of the current frame ultrasound image is obtained by iterative calculation using a preset contour optimization algorithm.

[0162] In some embodiments, the preset contour optimization algorithm may include active contour models. In some embodiments, the preset contour optimization algorithm may include geodesic active contours (GAC). In some embodiments, the preset contour optimization algorithm may include a morphological snake algorithm.

[0163] In some embodiments, the processing device can obtain the target mask image of the current frame ultrasound image based on the initial mask image of the current frame ultrasound image through various methods. For example, the processing device can input the initial mask image into a morphological Snake algorithm for iterative optimization to obtain the target mask image of the current frame ultrasound image.

[0164] In some embodiments, the processing device may determine the number of iterations of the morphological Snake algorithm based on the width of the initial mask image. Different widths of the initial mask image correspond to different numbers of iterations.

[0165] In some embodiments of this specification, the morphological Snake algorithm is used to perform contour iterative optimization on the initial mask image. Compared with the traditional numerical Snake algorithm, it has higher computational efficiency and more robust calculation results. The mask image optimized by the morphological Snake algorithm can better fit the inner and middle membrane boundaries, resulting in a higher quality and more realistic target mask image.

[0166] Figure 6 This is an exemplary flowchart illustrating the processing of a next frame of ultrasound image according to some embodiments of this specification. In some embodiments, process 600 may be executed by a processing device based on an intima-media measurement system for arterial ultrasound images. Figure 6 As shown, process 600 includes the following steps:

[0167] Step 610: Obtain the next frame of ultrasound image from the current frame of ultrasound image.

[0168] The next frame of ultrasound image refers to any ultrasound image acquired after the current frame. For example, the next frame could be an ultrasound image acquired by the ultrasound device at the moment immediately following the current moment. A preset time interval exists between the current moment and the next time interval; this preset time can be a system preset value, a system default value, etc.

[0169] In some embodiments, the processing device can periodically acquire ultrasound images collected by the ultrasound device and store them in a storage device. After performing an intima-media measurement on the current frame of the ultrasound image, the processing device can read the next frame of the ultrasound image from the storage device according to the image acquisition time. In some embodiments, after performing an intima-media measurement on the current frame of the ultrasound image, the processing device can control the ultrasound device to acquire the next frame of the ultrasound image in real time and send it to the processing device for processing.

[0170] Step 620: Based on the target mask image of the current frame ultrasound image and the next frame ultrasound image, determine the target mask image of the next frame ultrasound image.

[0171] In some embodiments, the processing device can determine the target mask image of the next frame of ultrasound image based on the target mask image of the current frame of ultrasound image through various methods such as image merging (overlay) and registration. For example, the processing device can determine the initial mask image of the next frame of ultrasound image based on the target mask image of the current frame of ultrasound image, and according to the initial mask image of the next frame of ultrasound image, for example, the processing device can overlay the target mask image of the current frame of ultrasound image onto the region of interest of the next frame of ultrasound image to obtain the initial mask image of the next frame of ultrasound image, and then determine the target mask image of the next frame of ultrasound image through the methods described in the previous embodiments.

[0172] In some embodiments, the processing device may extract a first image region corresponding to the region of interest from the next frame of ultrasound image; determine an initial mask image for the next frame of ultrasound image based on the target mask image and the first image region of the current frame of ultrasound image; and determine a target mask image for the next frame of ultrasound image based on the initial mask image of the next frame of ultrasound image.

[0173] The first image region refers to the image region in the next frame of ultrasound image that corresponds to the region of interest. In some embodiments, the processing device can extract the first image region corresponding to the region of interest from the next frame of ultrasound image based on the previously determined sampling frame.

[0174] In some embodiments, the processing device may perform appropriate preprocessing on the target mask image of the current frame of ultrasound image, such as morphological dilation operation, morphological opening and closing operation, etc., to obtain the initial mask image of the next frame of ultrasound image.

[0175] In some embodiments, the processing device may, based on the initial mask image of the next frame of ultrasound image, through... Figure 4 The target mask image for the next ultrasound image is determined in a similar manner to the initial mask image of the current frame ultrasound image. (See further details.) Figure 4 And its related descriptions.

[0176] In some embodiments of this specification, by extracting a first image region corresponding to the region of interest in the next frame of ultrasound image, registration between the region of interest and the first image region can be achieved, which helps to quickly and accurately obtain the inner and middle membrane boundaries in the first image region based on the target mask image of the current frame of ultrasound image. In some embodiments, the processing device can also determine the target mask image of the next frame of ultrasound image in various ways based on the target mask image of the current frame of ultrasound image and the target mask images of historical frames of ultrasound image. For example, the processing device can fuse the target mask image of the current frame of ultrasound image and the target mask images of historical frames of ultrasound image (e.g., weighted fusion) to determine the target mask image of the next frame of ultrasound image. The historical frames of ultrasound image can be one or more frames of ultrasound image that have undergone inner and middle membrane measurements before the current frame of ultrasound image.

[0177] Step 630: Determine the boundary of the intima-media region in the next frame of ultrasound image based on the intima-media measurement type and the target mask image of the next frame of ultrasound image.

[0178] In some embodiments, the processing device, based on the intima-media measurement type and the target mask image of the next frame of ultrasound image, compares with... Figure 1 The method for determining the endothelial-medial boundary of the next ultrasound image is similar to the method used to determine the endothelial-medial boundary of the current frame ultrasound image based on the endothelial-medial measurement type and the target mask image of the current frame ultrasound image. See further explanation. Figure 1 And its related descriptions.

[0179] In some embodiments of this specification, under the condition of real-time acquisition by the ultrasound equipment, the changes in the inner membrane structure and position of the inner membrane structure in the ultrasound images of consecutive frames are within a certain range. Therefore, by using the target mask image and the first image region of the current frame ultrasound image to determine the initial mask image of the next frame ultrasound image, the calculation process of determining the inner membrane measurement type and determining the gradient image in subsequent ultrasound images can be reduced, thereby improving the calculation efficiency and realizing the real-time measurement of the inner membrane thickness of ultrasound image video.

[0180] It should be noted that the above descriptions of processes 300, 400, and 600 are for illustrative purposes only and do not limit the scope of this specification. Those skilled in the art can make various modifications and changes to processes 300, 400, and 600 under the guidance of this specification. However, such modifications and changes remain within the scope of this specification.

[0181] Figure 7 This is an exemplary schematic diagram of real-time measurement according to some embodiments of this specification.

[0182] In some embodiments, the processing device may acquire regions of interest (ROIs) of multiple ultrasound images; determine the intima-media measurement type based on the ROIs; extract the image to be measured from the current ultrasound image based on the ROIs; acquire the target mask image of the current ultrasound image based on the image to be measured; determine the target mask image of the next ultrasound image based on the target mask image of the current ultrasound image; and determine the intima-media boundaries of the multiple ultrasound images based on the intima-media measurement type and the target mask images of the multiple ultrasound images.

[0183] In some embodiments, the multiple ultrasound images can be multiple consecutive ultrasound images in an ultrasound video stream. The ultrasound video stream can be acquired in real time by an ultrasound device.

[0184] In some embodiments, during real-time ultrasound scanning, the ultrasound probe remains fixed, and the region of interest (ROI) of multiple ultrasound images is consistent. After the physician determines the ROI of the first ultrasound image using a sampling frame, the processing device can determine the ROI of subsequently acquired ultrasound images through methods such as registration. For further explanation on determining the ROI, see [link to documentation]. Figure 3 And its related descriptions.

[0185] The intima-media boundary in multi-frame ultrasound images refers to the intima-media boundary corresponding to each frame of the multi-frame ultrasound images.

[0186] In some embodiments, when the processing device acquires the first frame of ultrasound image, it can use the first frame of ultrasound image as the current frame of ultrasound image, and then... Figures 2-6 The method described in some embodiments determines the target mask image and the inner membrane measurement results (including the inner membrane boundary and / or inner membrane thickness) of the first frame ultrasound image, and then determines the inner membrane measurement results of the next frame ultrasound image based on the target mask image of the first frame ultrasound image.

[0187] See Figure 7The processing device can determine the intima-media measurement type 730 of the first frame ultrasound image 711 and the region of interest 720, and extract the image to be measured 740 from the first frame ultrasound image 711; obtain the initial mask image 751 of the first frame ultrasound image based on the image to be measured 740; obtain the target mask image 761 of the first frame ultrasound image based on the initial mask image 751; and determine the intima-media measurement result 771 of the first frame ultrasound image based on the target mask image 761, the region of interest 720 and / or the intima-media measurement type 730. After determining the target mask image 761 of the first frame of ultrasound image or after acquiring the next frame of ultrasound image 712, the processing device can determine the target mask image 762 of the next frame of ultrasound image based on the target mask image 761 of the first frame of ultrasound image, and determine the inner and middle membrane measurement result 772 of the next frame of ultrasound image based on the target mask image 762 of the next frame of ultrasound image, the region of interest 720 and / or the inner and middle membrane measurement type 730. Similarly, the processing device can determine the target mask image 763 of the (n-1)th frame of ultrasound image, and the endothelial-medial measurement result 773 of the (n-1)th frame of ultrasound image; and after determining the target mask image 763 of the (n-1)th frame of ultrasound image or acquiring the nth frame of ultrasound image 713, it determines the target mask image 764 of the nth frame of ultrasound image based on the target mask image 763 of the (n-1)th frame of ultrasound image, and determines the endothelial-medial measurement result 774 of the nth frame of ultrasound image based on the target mask image 764 of the nth frame of ultrasound image, the region of interest 720, and / or the endothelial-medial measurement type 730. For more related explanations, see [link to related documentation]. Figures 3-6 And its related descriptions.

[0188] In some embodiments, the processing device can perform time-series analysis on the intima-media measurement results of multiple ultrasound images during real-time intima-media measurement to determine an IMT time-series analysis report. In some embodiments, the IMT time-series analysis report may include the intima-media measurement results corresponding to multiple ultrasound images. In some embodiments, the IMT time-series analysis report may include a trend analysis graph of intima-media thickness variation. The trend analysis graph refers to a graph showing the variation trend of the measured intima-media thickness relative to time. Figure 7 As shown, the processing device can perform time-series analysis on the target mask image 761 of the first frame of ultrasound image, the intima-media measurement result 772 of the next frame of ultrasound image, ..., the intima-media measurement result 773 of the (n-1)th frame of ultrasound image, and the intima-media measurement result 774 of the nth frame of ultrasound image, and determine the IMT time-series analysis report 780.

[0189] In some embodiments of this specification, by determining the target mask image of the first frame of ultrasound image, and based on the target mask image and the inner membrane measurement results of the first frame of ultrasound image, the target mask image and the inner membrane measurement results of each subsequent frame of ultrasound image are determined in real time. This improves the efficiency of determining the inner membrane boundary of each subsequent frame of ultrasound image, and facilitates the real-time measurement of the inner membrane thickness in ultrasound image video. By performing time-series analysis on the inner membrane measurement results of multiple frames of images, it is possible to effectively avoid situations where there is a large error between the IMT measurement item and the true value due to poor selection of a single frame image.

[0190] One or more embodiments of this specification also provide an intima-media measurement device for arterial ultrasound images, including a processing device for performing the intima-media measurement method for arterial ultrasound images as described in any of the above embodiments.

[0191] One or more embodiments of this specification also provide a computer-readable storage medium that stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer runs the intima-media measurement method of arterial ultrasound images as described in any of the above embodiments.

[0192] The basic concepts have been described above. Obviously, for those skilled in the art, the detailed disclosure above is merely illustrative and does not constitute a limitation of this specification. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this specification. Such modifications, improvements, and corrections are suggested in this specification and therefore remain within the spirit and scope of the exemplary embodiments described herein.

[0193] Furthermore, this specification uses specific terms to describe embodiments thereof. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic associated with at least one embodiment of this specification. Therefore, it should be emphasized and noted that references to "an embodiment," "one embodiment," or "an alternative embodiment" in different locations throughout this specification do not necessarily refer to the same embodiment. Moreover, certain features, structures, or characteristics in one or more embodiments of this specification can be appropriately combined.

[0194] Furthermore, unless expressly stated in the claims, the order of processing elements and sequences, the use of numbers and letters, or other names described in this specification are not intended to limit the order of the processes and methods described herein. Although various examples have been discussed in the foregoing disclosure of some embodiments of the invention that are currently considered useful, it should be understood that such details are for illustrative purposes only, and the appended claims are not limited to the disclosed embodiments; rather, the claims are intended to cover all modifications and equivalent combinations that conform to the spirit and scope of the embodiments described herein. For example, while the system components described above can be implemented using hardware devices, they can also be implemented solely using software solutions, such as installing the described system on existing servers or mobile devices.

[0195] Similarly, it should be noted that, in order to simplify the description disclosed herein and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of embodiments in this specification may sometimes combine multiple features into a single embodiment, drawing, or description thereof. However, this method of disclosure does not imply that the subject matter of this specification requires more features than those mentioned in the claims. In fact, the embodiments contain fewer features than all the features of a single embodiment disclosed above.

[0196] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. A method for measuring the intima-media layer in vascular ultrasound images, characterized in that, The method includes: Obtain the region of interest in the ultrasound image; Based on the region of interest, the type of measurement for the inner membrane is determined by a preset classification algorithm; Based on the region of interest, the image to be measured is extracted from the current frame of ultrasound image; Based on the image to be measured, obtain the target mask image of the current frame ultrasound image; Based on the intima-media measurement type, the intima-media thickness parameters of the current frame ultrasound image are determined according to the target mask image of the current frame ultrasound image. Wherein, obtaining the target mask image of the current frame ultrasound image based on the image to be measured includes: Gradient extraction is performed on the image to be measured to obtain a gradient image; Based on the gradient image, the estimated result of the inner and middle membrane boundary of the gradient image is determined; Based on the estimation results of the inner and middle membrane boundaries, the initial mask image of the current frame ultrasound image is obtained; Based on the initial mask image of the current frame ultrasound image, the target mask image of the current frame ultrasound image is obtained by iterative calculation using a preset contour optimization algorithm.

2. The method as described in claim 1, characterized in that, The step of extracting the image to be measured from the current frame of ultrasound image based on the region of interest includes: The region of interest is adjusted, and the image to be measured is extracted from the current frame ultrasound image based on the adjusted region.

3. The method as described in claim 1, characterized in that, The step of extracting gradients from the image to be measured to obtain a gradient image includes: Image gradient extraction is performed based on the image to be measured to obtain a positive gradient image; the positive gradient image is a gradient image determined based on the positive gradient of the image to be measured, and the positive gradient refers to the increase in brightness or intensity of the pixel value of a pixel along the gradient direction; Phase gradient extraction is performed on the image to be measured to obtain a phase gradient image; The gradient image is determined based on the positive gradient image and the phase gradient image.

4. The method as described in claim 1, characterized in that, The preset contour optimization algorithm includes the morphological snake algorithm.

5. The method as described in claim 1, characterized in that, The process of obtaining the initial mask image of the current frame ultrasound image based on the estimation result of the intima-media boundary includes: Based on the estimation results of the inner and middle membrane boundaries, the upper and lower boundary positions of each column of pixels in the gradient image are determined. Based on the upper and lower boundary positions, an intima-media boundary undulation analysis is performed to determine the initial mask image of the current frame ultrasound image; The analysis of the undulation of the inner and middle membrane boundaries based on the upper and lower boundary positions includes: Statistical analysis is performed on the upper and lower boundary positions of each column of pixels in the gradient image to obtain the change amplitude data of the upper or lower boundary positions. For the upper or lower boundary positions where the change amplitude exceeds the amplitude threshold, smoothing is performed based on morphological transformation.

6. The method as described in claim 1, characterized in that, The determination of the intima-media thickness parameters of the current frame ultrasound image based on the intima-media measurement type and the target mask image of the current frame ultrasound image includes: Based on the target mask image of the current frame ultrasound image and the region of interest, the inner and middle membrane boundaries are generated; Based on the inner and middle membrane measurement type and the inner and middle membrane boundary, a first contour line of the middle and outer membrane boundary and a second contour line of the lumen inner membrane boundary are determined. Based on the first contour line and the second contour line, the inner membrane thickness parameter corresponding to the inner membrane measurement type is determined.

7. The method as described in claim 1, characterized in that, The method further includes: Obtain the next frame of ultrasound image from the current frame of ultrasound image; Determining the target mask image of the next frame ultrasound image based on the target mask image of the current frame ultrasound image and the next frame ultrasound image includes: Extract the first image region corresponding to the region of interest from the next frame of ultrasound image; Based on the target mask image of the current frame ultrasound image and the first image region, determine the initial mask image of the next frame ultrasound image; Based on the initial mask image of the next frame ultrasound image, determine the target mask image of the next frame ultrasound image; Based on the intima-media measurement type and the target mask image of the next frame ultrasound image, the intima-media thickness parameters of the next frame ultrasound image are determined.

8. A computer-readable storage medium storing computer instructions, wherein when a computer reads the computer instructions in the storage medium, the computer executes the intima-media measurement method of vascular ultrasound images as described in any one of claims 1-7.

9. A method for measuring the intima-media layer in real-time vascular ultrasound images, the method comprising: Obtain the region of interest from multiple ultrasound images; Based on the region of interest, the type of measurement for the inner membrane is determined by a preset classification algorithm; Based on the region of interest, image feature optimization is performed on the current frame ultrasound image to obtain the image to be measured. Based on the image to be measured, obtain the target mask image of the current frame ultrasound image; Based on the target mask image of the current frame ultrasound image, determine the target mask image of the next frame ultrasound image; Based on the intima-media measurement type and the target mask image of the multi-frame ultrasound images, the intima-media thickness parameters of the multi-frame ultrasound images are determined. Wherein, obtaining the target mask image of the current frame ultrasound image based on the image to be measured includes: Gradient extraction is performed on the image to be measured to obtain a gradient image; Based on the gradient image, the estimated result of the inner and middle membrane boundary of the gradient image is determined; Based on the estimation results of the inner and middle membrane boundaries, the initial mask image of the current frame ultrasound image is obtained; Based on the initial mask image of the current frame ultrasound image, the target mask image of the current frame ultrasound image is obtained by iterative calculation using a preset contour optimization algorithm.

Images