WO2024093911A1 - 一种超声成像方法及超声设备 - Google Patents

一种超声成像方法及超声设备 Download PDF

Info

Publication number
WO2024093911A1
WO2024093911A1 PCT/CN2023/127760 CN2023127760W WO2024093911A1 WO 2024093911 A1 WO2024093911 A1 WO 2024093911A1 CN 2023127760 W CN2023127760 W CN 2023127760W WO 2024093911 A1 WO2024093911 A1 WO 2024093911A1
Authority
WO
WIPO (PCT)
Prior art keywords
region
organ
dimensional data
ovarian
transparency
Prior art date
Application number
PCT/CN2023/127760
Other languages
English (en)
French (fr)
Inventor
陈文卉
刘超越
邹耀贤
林穆清
Original Assignee
深圳迈瑞生物医疗电子股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳迈瑞生物医疗电子股份有限公司 filed Critical 深圳迈瑞生物医疗电子股份有限公司
Publication of WO2024093911A1 publication Critical patent/WO2024093911A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T15/00—3D [Three Dimensional] image rendering
    • G06T15/005—General purpose rendering architectures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
    • G06T17/20—Finite element generation, e.g. wire-frame surface description, tesselation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T19/00—Manipulating 3D models or images for computer graphics
    • G06T19/20—Editing of 3D images, e.g. changing shapes or colours, aligning objects or positioning parts
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10132—Ultrasound image

Definitions

  • the present invention relates to the technical field of medical devices, and more specifically, to an ultrasonic imaging method and an ultrasonic device.
  • organs or tissue structures in the human body are wrapped or blocked, such as ovaries and follicles, uterus and endometrium, etc.
  • the blocked organs cannot be seen. Therefore, transparent rendering of the outer organs can more clearly show the shape structure of the blocked organs, as well as the positional relationship between the two organs, so as to better assist doctors in clinical diagnosis.
  • the three-dimensional mixed rendering of ovaries and follicles can assist doctors in ovarian receptivity analysis
  • the three-dimensional mixed rendering of cervix and uterine body can enable doctors to judge the degree of uterine curvature based on the positional relationship between the rendered cervix and uterine body.
  • a fixed transparency coefficient is set for organs that need to be displayed transparently. This makes the organ rendering lack of three-dimensionality and unable to reflect the transformation of the organ at different depths or angles. The visual experience is poor, and the positional relationship between the organs presented is not clear enough, which may affect the results of the doctor's clinical diagnosis.
  • the first aspect of the present invention provides an ultrasonic imaging method, comprising:
  • the ovarian region corresponding to the ovary and the plurality of follicles corresponding to the ovarian region ... Perform positioning detection on multiple follicle areas to obtain positioning detection results;
  • the three-dimensional data is rendered according to the determined transparency coefficient to obtain a rendered image corresponding to the ovarian region and the multiple follicle regions, wherein in the rendered image, different parts of the ovarian region present different transparencies.
  • a second aspect of the present invention provides an ultrasonic imaging method, comprising:
  • the three-dimensional data is rendered according to the determined transparency coefficient to obtain rendered images of the first organ region and the second organ region, wherein in the rendered images, different parts of the first organ region present different transparencies.
  • a third aspect of the present invention provides an ultrasonic device, comprising:
  • a transmitting and receiving circuit which is configured to control the ultrasonic probe to transmit ultrasonic waves to a target tissue of the object under test, and to control the ultrasonic probe to receive an echo signal of the ultrasonic waves;
  • a memory for storing computer executable instructions
  • a processor configured to, when executing the computer executable instructions, obtain three-dimensional data according to the echo signal, and execute the ultrasound imaging method of any one of the above embodiments to generate a rendered image
  • a display is configured to display the rendered image.
  • the transparency coefficients corresponding to different parts of the external organ can be determined, so that different parts of the external organ present different transparencies, increasing the three-dimensional sense of the external organ. Therefore, a rendering effect with a clear position relationship can be presented to the user, so that the user can clearly see the position, size and inclusion relationship between the organs, etc., thereby enhancing the visual experience and helping to improve work efficiency and accuracy.
  • FIG1 shows a schematic block diagram of an ultrasound device according to an embodiment of the present invention
  • FIG2 shows a schematic flow chart of an ultrasound imaging method according to an embodiment of the present invention
  • 3a-3c are schematic diagrams showing rendered images according to an embodiment of the present invention.
  • FIG4 shows another schematic flow chart of an ultrasound imaging method according to an embodiment of the present invention.
  • FIG. 5 shows a schematic diagram of a computing device according to an embodiment of the present invention.
  • the present invention provides an ultrasonic imaging method and an ultrasonic device.
  • the transparency coefficients corresponding to different parts of the external organ can be determined, so that different parts of the external organ present different transparencies, thereby increasing the three-dimensional sense of the external organ. Therefore, a rendering effect with a clear position relationship can be presented to the user, so that the user can clearly see the position, size and inclusion relationship between the organs, etc., thereby improving the visual experience and helping to increase the accuracy of diagnostic analysis.
  • Fig. 1 shows a schematic block diagram of an ultrasound device according to an embodiment of the present invention.
  • the ultrasound device 100 includes an ultrasound probe 110, a transmitting and receiving circuit 111, a processor 112, a memory 113, and a display 114. Furthermore, the ultrasound device 100 may also include a beamforming circuit and a transmitting/receiving selection switch.
  • the ultrasound probe 110 generally includes an array of multiple array elements. Each time an ultrasound wave is emitted, all or part of the array elements of the ultrasound probe 110 participate in the emission of the ultrasound wave. At this time, each array element or part of the array elements participating in the ultrasound emission is excited by the emission pulse and emits ultrasound waves respectively. The ultrasound waves emitted by these array elements are superimposed during the propagation process to form a synthetic ultrasound beam emitted to the area where the region of interest of the object under test is located.
  • the region of interest may be an ovarian region, a uterine region, etc.
  • the transmitting and receiving circuit 111 can be coupled to the ultrasonic probe 110 through a transmitting and receiving selection switch.
  • the transmitting and receiving selection switch can also be called a transmitting and receiving controller, which can include a transmitting controller and a receiving controller.
  • the transmitting controller is used to stimulate the ultrasonic probe 110 to transmit ultrasonic waves to the area of interest of the object under test via the transmitting circuit;
  • the receiving controller is used to receive the ultrasonic echo returned from the area of interest of the object under test via the ultrasonic probe 110 via the receiving circuit, thereby obtaining the ultrasonic echo signal.
  • the transmitting and receiving circuit 111 sends the echo signal to the beamforming circuit.
  • the beamforming circuit performs focusing delay, weighting and channel summing on the electrical signal, and then sends the processed ultrasound echo data to the processor 112.
  • the processor 112 may be implemented by software, hardware, firmware or any combination thereof, and may use circuits, single or multiple application specific integrated circuits (ASICs), single or multiple general purpose integrated circuits, single or multiple microprocessors, single or multiple programmable logic devices, or any combination of the aforementioned circuits or devices, or other suitable circuits or devices, so that the processor 112 can perform the corresponding steps of the methods in various embodiments of the present specification.
  • the processor 112 can control other components in the ultrasound device 100 to perform desired functions.
  • the processor 112 processes the echo signal of the ultrasonic wave it receives to obtain a three-dimensional ultrasonic image of the region of interest of the object under test.
  • the ultrasonic probe 110 transmits or receives ultrasonic waves in a series of scanning planes, and the processor 112 integrates them according to their three-dimensional spatial relationship to achieve the scanning of the region of interest of the object under test in three-dimensional space and the reconstruction of the three-dimensional image.
  • the processor 114 performs some or all of the image post-processing steps such as denoising, smoothing, and enhancement to obtain a three-dimensional ultrasonic image of the region of interest of the object under test.
  • the obtained three-dimensional ultrasonic image can be stored in the memory 113, or displayed on the display 114, or transmitted to other storage devices for storage via wired or wireless communication lines.
  • the memory 113 is used to store instructions executed by the processor, store echo signals of received ultrasound waves, store ultrasound images, etc.
  • the memory may be a flash memory card, a solid-state memory, a hard disk, etc. It may be a volatile memory and/or a non-volatile memory, a removable memory and/or a non-removable memory, etc.
  • the display 114 is communicatively coupled to the processor 112.
  • the display 112 may be a touch screen, a liquid crystal display, or the like.
  • the display 112 is shown as being part of the ultrasound device 100, in other embodiments, the display 112 may also be an independent display device such as a liquid crystal display, a television, or the like that is independent of the ultrasound device 100; or, the display 112 may also be a display screen of an electronic device such as a smart phone or a tablet computer, or the like.
  • the number of displays 112 may be one or more.
  • the display 112 may include a main screen and a touch screen, wherein the main screen is mainly used to display ultrasound images, and the touch screen is mainly used for human-computer interaction.
  • the display 114 can display the ultrasound image obtained by the processor 112.
  • the display 114 can also provide a graphical interface for human-computer interaction to the user while displaying the ultrasound image.
  • One or more controlled objects are arranged on the graphical interface.
  • the user uses the human-computer interaction device to input operation instructions to control these controlled objects, thereby performing corresponding control operations.
  • an icon is displayed on the graphical interface, and the icon can be operated by the human-computer interaction device to perform a specific function.
  • the ultrasound device 100 may further include other human-computer interaction devices in addition to the display 114, which are communicatively coupled to the processor 112.
  • the processor 112 may be connected to the human-computer interaction device via an external input/output port, and the external input/output port may be a wireless communication module, or a wired communication module, or a combination of the two.
  • the external input/output port may also be implemented based on USB, bus protocols such as CAN, and/or wired network protocols, etc.
  • the human-computer interaction device may include an input device for detecting user input information, which may be, for example, a control instruction for ultrasonic transmission/reception timing, an operation input instruction for drawing points, lines or frames on an ultrasonic image, or other instruction types.
  • the input device may include a keyboard, a mouse, a scroll wheel, a trackball, a mobile input device (such as a mobile device with a touch screen, a mobile phone, etc.), a multi-function knob, etc., or a combination of multiple thereof.
  • the human-computer interaction device may also include an output device such as a printer.
  • the components included in the ultrasound device 100 shown in FIG1 are only exemplary, and the ultrasound device 100 may include more or fewer components, and the present invention is not limited thereto.
  • FIG2 shows a schematic flow chart of an ultrasound imaging method according to an embodiment of the present invention.
  • the ultrasound imaging method 200 in FIG2 can be executed by the processor 112 of the ultrasound device 100 in FIG1 to obtain a rendered image and display it via the display 114.
  • the ultrasound imaging method 100 can be executed by a processor of any other computing device to obtain a rendered image and display it via a display coupled to the processor.
  • the transparency coefficients corresponding to different parts of the outer ovarian region are determined, so that different parts of the outer ovarian region present different transparencies, thereby increasing the three-dimensional sense of the ovarian region and presenting a rendering effect with clear positional relationships to the user, so that the user can clearly see the position, size, and inclusion relationship between the ovaries and follicles, thereby enhancing the visual experience and helping to improve work efficiency and accuracy.
  • an ultrasound imaging method 200 includes the following steps:
  • step 201 the ovary of the subject to be tested and the plurality of follicles contained in the ovary are obtained.
  • the acquired three-dimensional data shall include the complete ovarian and follicle structure of the object under test.
  • the three-dimensional data may be three-dimensional raw data or three-dimensional image data.
  • the user can use the probe of the ultrasonic device to scan the ovarian tissue of the object to be measured and collect three-dimensional data or four-dimensional data.
  • the ultrasonic probe of the ultrasonic device transmits ultrasonic waves to the ovarian tissue of the object to be measured.
  • the ovarian tissue may include ovaries and multiple follicles wrapped by the ovaries.
  • the ultrasonic wave needs to be transmitted to the inside of the object to be measured.
  • the ultrasonic probe may be placed at a position corresponding to the ovarian tissue on the surface of the object to be measured.
  • the scanning area of the ultrasonic probe may cover the ovarian tissue by intracavitary ultrasonic scanning, and then the ultrasonic probe is used to transmit the ultrasonic wave, thereby realizing the transmission value of the ultrasonic wave to the inside of the ovary of the object to be measured.
  • the object to be measured may be an object including ovarian tissue, such as a human organ or a human tissue structure. Then, the ultrasonic probe receives the echo signal of the ultrasonic wave returned from the ovarian tissue of the object to be measured, and performs beam synthesis, three-dimensional reconstruction and other processing on the echo signal to obtain a three-dimensional ultrasonic image or a four-dimensional ultrasonic image.
  • pre-stored three-dimensional data or four-dimensional data can be obtained from a memory.
  • the user can use the probe of an ultrasound device to scan the ovarian tissue of the subject, obtain the three-dimensional data or four-dimensional data, and store it in the memory or send it to other computing devices for storage.
  • image rendering is required later, the three-dimensional data or four-dimensional data is obtained from the memory by other computing devices.
  • three-dimensional data at a certain moment needs to be selected from the four-dimensional data as the three-dimensional data to be rendered.
  • the three-dimensional data can be selected manually or automatically.
  • the user can view the four-dimensional data by moving an input device such as a mouse or a trackball, and select the three-dimensional data containing the complete ovarian and follicle structure as the three-dimensional data to be rendered.
  • the automatic three-dimensional data selection method can automatically identify the four-dimensional data by applying a machine learning or deep learning algorithm, and select the three-dimensional data to be rendered.
  • the three-dimensional data to be rendered can be selected from the four-dimensional data according to the target recognition algorithm.
  • the target recognition algorithm is implemented by directly extracting features from the three-dimensional data at each moment, and then scoring and classifying the data.
  • the main steps of the algorithm are: 1) Building a database: The database includes a large number of three-dimensional data sets and their corresponding calibration results. The calibration result is whether it is standard three-dimensional data to be rendered (such as whether it contains a complete organ structure). 2) Identification and positioning steps of the three-dimensional data to be rendered.
  • the target recognition algorithm may include target recognition algorithms based on traditional machine learning. Recognition algorithm and target recognition algorithm based on deep learning.
  • the target recognition algorithm based on traditional machine learning extracts features from three-dimensional data, such as local context information, texture information, Harr features, etc., and then integrates the relevant features into the cascaded classifier, such as support vector machine (SVM), Adaboost, random forest, etc., to discriminate and classify the features. After traversing the three-dimensional data at all times in the four-dimensional data, the three-dimensional data with the highest score is selected as the three-dimensional data to be rendered.
  • SVM support vector machine
  • Adaboost Adaboost
  • random forest random forest
  • the target recognition algorithm based on deep learning is implemented through a neural network architecture and belongs to a classification network. Its main structure is a stack of convolutional layers, activation layers, pooling layers, and fully connected layers.
  • the shallow convolutional layer extracts relevant features from the three-dimensional data, and then the extracted features are linearly combined in the fully connected layer, and finally the probability of the current image is output. After traversing the three-dimensional data at all times in the four-dimensional data, the three-dimensional data with the highest probability is selected as the three-dimensional data to be rendered.
  • Common classification networks include 3D FCN, etc.
  • three-dimensional data to be rendered can be selected from four-dimensional data according to a target detection algorithm.
  • the ovarian region and follicle region contained in the three-dimensional data at each moment are detected by the target detection algorithm, and the three-dimensional data to be rendered are selected according to the number and volume of the detected ovarian region and follicle region.
  • the main steps of the algorithm are: 1) Building a database: The database includes a large number of three-dimensional data sets and their corresponding calibration results. The calibration result is a bounding box of the ovarian region and follicle region of the three-dimensional data. 2) Detection and positioning steps of standard three-dimensional data to be rendered.
  • the target detection algorithm can include a target detection algorithm based on traditional machine learning and a target detection algorithm based on deep learning.
  • the target detection algorithm based on traditional machine learning is mainly divided into the following steps: 1) First, the region is selected by moving sliding windows of different scales and different aspect ratios; 2) Relevant features (such as Harr features, HOG features, etc.) are extracted based on the image blocks in the region; 3) The extracted features are sent to a classifier (such as SVM, Adaboost, etc.) for classification to determine the ovarian region or follicle region.
  • a classifier such as SVM, Adaboost, etc.
  • the target detection algorithms based on deep learning are divided into target detection methods using candidate regions and deep learning classification, regression methods based on deep learning, and target detection methods based on point clouds.
  • the target detection method using candidate regions and deep learning classification extracts candidate regions and performs deep learning-based classification on the corresponding regions, such as transforming the convolution kernels of R-CNN (Selective Search + CNN + SVM), SPP-Net (ROI Pooling), Fast R-CNN (Selective Search + CNN + ROI), Faster R-CNN (RPN + CNN + ROI), R-FCN, etc. into three Dimension for classification.
  • R-CNN Selective Search + CNN + SVM
  • SPP-Net ROI Pooling
  • R-CNN Selective Search + CNN + ROI
  • R-CNN Faster R-CNN
  • R-FCN etc.
  • the main steps of the deep learning-based regression method are to first divide the image into S ⁇ S ⁇ S grids, and then let each grid be responsible for detecting objects whose centers fall on this grid (such as the ovarian area and the follicle area). Finally, the network outputs the coordinates of each position of the object (such as the center point coordinates, length and width, etc.) and the probability of the category to which the object belongs. This can be achieved by transforming the convolution kernel of the YOLO series (YOLO V1, YOLO V2, YOLO V3) into three-dimensional, SSD, DenseBox and other algorithms.
  • Point cloud-based object detection methods are divided into two types: point-based and voxel-based.
  • the point-based detection method is a method that directly applies a deep learning model to point cloud data.
  • the input is an n ⁇ 3 point cloud tensor.
  • After the network extracts features from the point cloud data, it modifies the features in the standard coordinates to obtain the final detection result.
  • An example of the network is Point-RCNN.
  • the voxel-based method divides the point cloud into 3D voxels, which are processed by a three-dimensional CNN, and then multiple detection heads are used to explore the position to improve detection performance.
  • point-based and voxel-based detection methods can be combined to combine the three-dimensional voxel convolutional neural network (CNN) and the point network-based set abstraction technology to accurately estimate the position of three-dimensional objects (such as PV-RCNN, point-voxel RCNN).
  • CNN three-dimensional voxel convolutional neural network
  • point network-based set abstraction technology to accurately estimate the position of three-dimensional objects (such as PV-RCNN, point-voxel RCNN).
  • the location of the ovarian region, the location and number of the follicle region and other information can be obtained.
  • the number of follicle regions can also be used as a selection criterion.
  • the selected 3D data is the 3D data to be rendered.
  • the three-dimensional data to be rendered can be selected from the four-dimensional data according to the target segmentation algorithm.
  • the ovarian region and the follicle region contained in the three-dimensional data at each moment are segmented respectively by the target segmentation algorithm, and the three-dimensional data to be rendered are selected according to the number and volume of the segmented ovarian region and follicle region.
  • the target segmentation algorithm can be directly used to segment the ovarian region and the follicle region contained in the three-dimensional data, or the three-dimensional data can be split into several two-dimensional data, for example, radial sampling or parallel sampling is performed with the center of the ovarian region as the axis, and after segmenting the two-dimensional data, several segmentation results are fused to obtain the final three-dimensional segmentation result.
  • Traditional segmentation algorithms include level set-based segmentation algorithms, random walkers, graph cuts, snakes, etc.
  • Object segmentation algorithms can also include traditional machine learning-based object segmentation algorithms and deep learning-based object segmentation algorithms.
  • the main steps of the target segmentation algorithm based on traditional machine learning are: 1) Build a database:
  • the database includes a large number of ultrasound data sets and their corresponding calibration results (in the case of direct segmentation of three-dimensional data, it is a three-dimensional data set, and in the case of segmentation of multiple two-dimensional data included in the three-dimensional data, it is a two-dimensional data set).
  • the calibration result is a mask of the ovarian region and the follicle region of the ultrasound data, that is, the segmentation result.
  • the ultrasound data is divided into multiple data blocks (in the case of direct segmentation of three-dimensional data, it is an image block of S ⁇ S ⁇ S size, and in the case of segmentation of two-dimensional data included in the three-dimensional data, it is an image block of S ⁇ S size), and then the data blocks are subjected to feature extraction.
  • the feature extraction methods include traditional PCA, LDA, Harr features, texture features, etc., and deep neural networks (such as Overfeat networks) can also be used to extract features.
  • the extracted features are classified using cascaded classifiers, such as KNN, SVM, random forests, and other discriminators, so as to determine whether the current data block is the ovarian region or the follicle region, and the classification result is used as the center point marking result of the current data block, and finally the segmentation result of the entire ultrasound data is obtained.
  • cascaded classifiers such as KNN, SVM, random forests, and other discriminators
  • the main steps of the target segmentation algorithm based on deep learning are: 1) Build a database:
  • the database includes a large number of ultrasound data sets and their corresponding calibration results (in the case of direct segmentation of three-dimensional data, it is a three-dimensional data set, and in the case of segmentation of multiple two-dimensional data included in the three-dimensional data, it is a two-dimensional data set).
  • the calibration result is the mask of the ovarian area and follicle area of the ultrasound data, that is, the segmentation result.
  • Segmentation of the ovarian area and follicle area that is, an end-to-end semantic segmentation algorithm.
  • the input can be an image or a three-dimensional point cloud (in the case of direct segmentation of three-dimensional data).
  • an output image with the same size as the input image is obtained by stacking convolutional layers, pooling layers, upsampling or deconvolution layers, etc.
  • the output image directly segments the required target organ area.
  • This method is a supervised learning.
  • Common two-dimensional networks include FCN, U-Net, Mask R-CNN, etc.
  • three-dimensional segmentation networks include 3D U-Net, 3D FCN, Medical-Net, etc.
  • point cloud is used as input, a. first represent the data as a set of point clouds, expressed as an n ⁇ 3 tensor, where n represents the number of point clouds, b. align the input data through a learned transformation matrix to ensure the invariance of the model to feature space transformation, c.
  • the 3D data is split into several 2D data and the 2D data is segmented, the 2D segmentation results need to be fused to obtain the final 3D segmentation result.
  • the three-dimensional segmentation result can be obtained by combining them, for example, directly adding them together, or taking the maximum value or average value of the pixels as the segmentation result.
  • the three-dimensional segmentation result can also be obtained by interpolation (applicable to radial sampling method).
  • the location of the ovarian region, the location and number of the follicle region and other information in the three-dimensional data can be obtained.
  • the number of follicles can also be used as a selection criterion at the same time.
  • the selected three-dimensional data is the three-dimensional data to be rendered.
  • step 202 the ovarian region corresponding to the ovary and the multiple follicle regions corresponding to the multiple follicles are positioned in the three-dimensional data to obtain positioning detection results.
  • the positioning detection results include the type (such as ovary and follicle) and position of the organ region.
  • the ovarian region and the follicle region can be positioned and detected according to the target detection algorithm.
  • the target detection algorithm detects the ovarian region and the follicle region contained in the three-dimensional data at each moment, thereby obtaining the type and location information of the organ region.
  • the main steps of the algorithm are: 1) Building a database: The database includes a large number of three-dimensional data sets and their corresponding calibration results. The calibration result is the bounding box of the ovarian region and the follicle region. 2) Detection and positioning steps of the ovarian region and the follicle region.
  • the target detection algorithm can be divided into a target detection algorithm based on traditional machine learning, a target detection algorithm based on deep learning, and a target detection algorithm based on point cloud.
  • the specific steps of the algorithm are similar to the target detection algorithm used in step 201, which are described in detail above and will not be repeated here.
  • the ovarian region and the follicle region can be positioned and detected according to the target segmentation algorithm. Similar to the target segmentation algorithm used in step 201, during segmentation, the target segmentation algorithm can be directly used to segment the ovarian region and the follicle region contained in the three-dimensional data, or the three-dimensional data can be split into several two-dimensional data, for example, radial sampling or parallel sampling is performed with the center of the ovarian region as the axis, and after segmenting the two-dimensional section, several segmentation results are fused to obtain the final three-dimensional segmentation result.
  • Traditional segmentation algorithms include segmentation algorithms based on level sets (Level Set), random walks (Random Walker), graph cuts (Graph Cut), Snake, etc.
  • the target segmentation algorithm can also include a target segmentation algorithm based on traditional machine learning and a target segmentation algorithm based on deep learning.
  • the specific steps of the two-dimensional or three-dimensional target segmentation algorithm are similar to the target segmentation algorithm used in step 201, which are described in detail above and will not be repeated here.
  • the user can manually trace the ovarian region and the follicle region.
  • a user traces the ovarian region and the follicle region in a three-dimensional ultrasound image by moving an input device such as a mouse or a trackball.
  • the processor receives the user's tracing operation, segments the ovarian region and the follicle region in the three-dimensional ultrasound image, and obtains a segmentation result.
  • the ovarian region and a part of the multiple follicle regions can be located and detected in the three-dimensional data according to the target detection algorithm, and the ovarian region and another part of the multiple follicle regions can be located and detected in the three-dimensional data according to the target segmentation algorithm.
  • the ovarian region and the multiple follicle regions can be located and detected according to different location detection algorithms. For example, the ovarian region is located and detected in the three-dimensional data according to the target detection algorithm, and the multiple follicle regions are located and detected in the three-dimensional data according to the target segmentation algorithm.
  • the ovarian region and a part of the follicle regions are located and detected in the three-dimensional data according to the target detection algorithm, and the other follicle regions are located and detected in the three-dimensional data according to the target segmentation algorithm.
  • the ovarian region is located and detected in the three-dimensional data according to the target segmentation algorithm, and the multiple follicle regions are located and detected in the three-dimensional data according to the target detection algorithm.
  • step 203 based on the positioning detection result, the transparency coefficients corresponding to different parts of the ovarian region are determined.
  • the entire ovarian region is rendered transparently, and each part has a different transparency coefficient.
  • the corresponding transparency coefficient can be adaptively determined for each triangular facet (rendering method using face rendering) or each pixel (rendering method using volume rendering) of the entire ovarian region, and the transparency coefficient determination rule is pre-set. For example, the more vertical the normal line of the surface of the ovarian region and the line of sight are, the smaller the transparency coefficient is, and the more opaque it is. For another example, the more vertical the angle between the normal line of the surface of the ovarian region and a fixed direction is, the smaller the transparency coefficient is, and the more opaque it is. For another example, the angle between the normal line of the surface of the ovarian region and different directions is calculated to determine the transparency coefficient.
  • the front portion of the ovarian region under the user's perspective presents a variety of different transparencies, while other portions of the ovarian region under the user's perspective present the same transparency or opacity.
  • the front portion of the ovarian region under the user's perspective can be automatically detected, and a corresponding transparency coefficient can be determined for each triangle or each pixel corresponding to the front portion of the ovarian region.
  • method 200 may also include: determining the front portion of the ovarian region under the user's perspective based on the positioning detection result, and determining corresponding transparency coefficients for the front portion and other portions, respectively.
  • the spatial depth information of the ovarian region or the normal direction of the ovarian contour can be obtained based on the positioning detection result.
  • the spatial depth information of the ovarian region or the normal direction of the contour for example, can be obtained.
  • the fixed transparency coefficient can be obtained from the model vertex coordinate information used to construct the ovarian region. Afterwards, based on the spatial depth information or the normal direction of the contour, the front part of the ovarian region under the user's perspective is determined, and corresponding transparency coefficients are determined for the front part and other parts respectively.
  • the fixed transparency coefficient can be, for example, 0 (i.e., opaque), or can be less than the transparency coefficient of the front part (i.e., the transparency is less than the transparency of the front part).
  • the fixed transparency coefficient can be pre-set or input by the user.
  • the target part of the ovarian region under the user's perspective presents a variety of different transparencies, while the other parts of the ovarian region under the user's perspective present the same transparency or opacity.
  • the target part can be determined first, and then the corresponding transparency coefficient is determined for each triangular facet or each pixel corresponding to the target part.
  • the target part can be automatically determined (such as according to the user's perspective), or can be selected by the user, and the processor determines it in response to the user's selection operation.
  • method 200 can also include: determining the target part to be transparently rendered in the ovarian region, and determining the corresponding transparency coefficients for the target part and other parts respectively.
  • the fixed transparency coefficient can be, for example, 0 (i.e., opaque), or can be less than the transparency coefficient of the target part (i.e., the transparency is less than the transparency of the front part).
  • the fixed transparency coefficient can be pre-set or input by the user.
  • the following describes the process of adaptively determining the transparency factor for the entire ovarian region or a portion thereof.
  • the image is rendered by face rendering, and model vertex coordinate information of the ovarian region can be extracted from the three-dimensional data based on the positioning detection result.
  • the model vertex coordinate information is used to form a plurality of triangular facets for constructing a mesh model of the ovarian region.
  • the transparency coefficient corresponding to at least some of the plurality of triangular facets is determined. Specifically, if the transparency coefficient is adaptively determined for the entire ovarian region, the corresponding transparency coefficient is determined for all triangular facets of the mesh model of the ovarian region. If the transparency coefficient is adaptively determined only for the front part or the target part of the ovarian region, the corresponding transparency coefficient is determined for those triangular facets in the mesh model of the ovarian region that correspond to the front part or the target part.
  • the normal vector of the triangle can be calculated based on the coordinate information of the three vertices of the triangle, and then the angle between the normal vector and the preset reference direction can be calculated. Finally, the transparency coefficient corresponding to the triangle is determined according to the size of the angle.
  • the direction vector between the three vertices of the triangular face and the centroid of the model can be calculated based on the coordinate information of the three vertices of the triangular face, and then the angle between the direction vector and the preset reference direction can be calculated. Finally, the corresponding position of the triangular face can be determined based on the size of the angle. Transparency factor.
  • angles between the three vertices and a preset reference direction are calculated based on the coordinate information of the three vertices of the triangular face, and then the transparency coefficient corresponding to the triangular face is determined based on the size of the angle.
  • the same preset reference direction may be used, such as the sight line direction or a fixed direction.
  • different preset reference directions may be used for different triangular facets.
  • the triangular facets may be grouped, and a preset reference direction may be used for each group of triangular facets.
  • the calculation rule of the transparency coefficient may be set according to the preset reference direction. For example, when the preset reference direction is the sight line direction, the calculation rule of the transparency coefficient may be set such that the larger the angle, the smaller the transparency coefficient.
  • a transparency coefficient corresponding to at least some of the multiple pixels used to render the ovarian region may be determined. Specifically, if the transparency coefficient is adaptively determined for the entire ovarian region, the corresponding transparency coefficient is determined for all pixels in the ovarian region. If the transparency coefficient is adaptively determined only for the front part or the target part of the ovarian region, the corresponding transparency coefficient is determined for those pixels corresponding to the front part or the target part.
  • the boundary contour of the ovarian region can be segmented based on the positioning detection result, and then the transparency coefficient corresponding to at least some of the pixels in the boundary contour of the ovarian region is determined according to the algorithm corresponding to the volume rendering.
  • a ray tracing algorithm is used.
  • volume rendering multiple rays passing through the three-dimensional data are emitted based on the line of sight direction, and each ray is progressively advanced at a fixed step length.
  • the three-dimensional data on the ray path is sampled, and the opacity of each sampling point is determined according to the gray value of each sampling point.
  • the opacity of each sampling point on each ray path is accumulated to obtain the cumulative opacity.
  • the cumulative opacity on each ray path is mapped to a transparency coefficient, and the transparency coefficient is mapped to a pixel of the two-dimensional image.
  • the transparency coefficients of the pixels corresponding to all ray paths are obtained, and a rendered image with different transparencies can be obtained.
  • a mapping relationship between grayscale value and transparency coefficient may be preset, grayscale values corresponding to multiple pixels within the boundary contour of the ovarian region may be extracted from the three-dimensional data, and then the transparency coefficient corresponding to each pixel may be determined based on the mapping relationship.
  • step 204 the three-dimensional data is rendered according to the determined transparency coefficient.
  • the rendered image corresponding to the ovarian region and the multiple follicle regions is obtained.
  • different parts of the ovarian region present different transparency.
  • the entire ovarian region is rendered transparently, and different parts have different transparency.
  • Figure 3a shows a schematic diagram. As can be seen from Figure 3a, the more perpendicular the part of the entire ovarian region to the line of sight (such as the edge part), the lower the transparency, that is, the more opaque, and the smaller the angle with the user's line of sight (such as the part near the middle of the ovarian region), the higher the transparency, that is, the more transparent.
  • the front part of the ovarian region in the user's perspective presents different transparency, while the other parts of the ovarian region in the user's perspective present the same transparency or opacity. In this way, the amount of calculation of the transparency coefficient can be reduced without affecting the user's observation, and the rendering process speed can be accelerated.
  • the target portion of the ovarian region under the user's perspective presents different transparencies, while other portions of the ovarian region under the user's perspective present the same transparencies or opacities.
  • FIG. 3b shows a schematic diagram.
  • the upper right portion of the ovarian region presents a variety of different transparencies, wherein the portion that is more vertical to the user's line of sight (such as the edge portion) has a lower transparency, i.e., the more opaque, and the portion that is smaller in angle with the user's line of sight (such as the portion close to the middle of the ovarian region) has a higher transparency, i.e., the more transparent.
  • FIG. 3c shows another schematic diagram. It can be seen in FIG. 3c that, compared to FIG. 3b, the transparent portion of the right side of the ovarian region is slightly larger than that in FIG. 3b.
  • the ovarian region and the multiple follicle regions may be rendered in any color.
  • the rendering colors of the ovarian region and the multiple follicle regions may be different.
  • the rendering color of each follicle region may also be different to distinguish follicle regions of different sizes and positions.
  • method 200 may also include: determining the rendering colors of the ovarian region and the multiple follicle regions. Thereafter, the three-dimensional data is rendered according to the determined rendering colors. In this way, the position and size relationship of the ovarian region and each follicle region can be presented to the user more clearly, thereby further improving the user's visual experience.
  • the target follicle region in addition to the ovarian region, can also be transparent Rendering and determining the corresponding transparency coefficient.
  • the target follicle region may be a follicle region that blocks other follicle regions, or may be a follicle region selected as needed.
  • the target follicle region may be determined automatically, or may be selected by the user, and the processor determines it in response to the user's selection operation. In the rendered image, different parts of the target follicle region present different transparency.
  • method 200 may also include: based on the positioning detection result, determining the target follicle region that blocks one or more other follicle regions from the user's perspective in multiple follicle regions, and then determining the transparency coefficients corresponding to different parts of the target follicle region.
  • the spatial depth information or the normal direction of the contour of multiple follicles may be obtained based on the positioning detection result.
  • the spatial depth information or the normal direction of the contour of the follicle region may be obtained from the model vertex coordinate information used to construct the follicle region, for example.
  • the target follicle region that blocks one or more other follicle regions from the user's perspective is determined in multiple follicle regions.
  • the transparency coefficients corresponding to different parts of the target follicle region are determined. Similar to the ovarian region, the transparency coefficient can also be adaptively determined for the entire target follicle region or the target portion. For example, only the overlapping portion (i.e., the blocked portion) of the target follicle region and other follicle regions can be transparently rendered, and the transparency coefficient can be adaptively determined, while other portions are not transparently rendered.
  • the corresponding transparency coefficient can be calculated for each triangular facet or pixel corresponding to the entire target follicle region or the target portion. The process of calculating the transparency coefficient is the same as described above and will not be repeated here.
  • FIG4 shows a schematic flow chart of an ultrasound imaging method according to an embodiment of the present invention.
  • the ultrasound imaging method 300 in FIG4 can be executed by the processor 112 of the ultrasound device 100 in FIG1 to obtain a rendered image and display it via the display 114.
  • the ultrasound imaging method 300 can be executed by a processor of any other computing device to obtain a rendered image and display it via a display coupled to the processor.
  • the ultrasound imaging method 300 includes the following steps:
  • step 301 the three-dimensional data to be rendered of the first organ and the second organ of the object to be measured are obtained, wherein the first organ at least partially wraps the second organ.
  • the first organ and the second organ can be any organ or tissue structure with a wrapping relationship.
  • the first organ is an ovary and the second organ is an ovarian follicle;
  • the first organ is a uterine body and the second organ is an endometrium;
  • the first organ is a fetal cranial brain and the second organ is an internal structure of the cranial brain (such as a cerebellum, a cavum septum pellucidum, a thalamus and/or a lateral ventricle, etc.);
  • the first organ is a fetal abdomen and the second organ is an internal structure of the abdomen (such as a gastric bubble, a fetal heart, a spine, a kidney and/or a blood vessel, etc.);
  • the first organ is a liver and the second organ is an intrahepatic blood vessel;
  • the first organ is a heart and the second organ is an internal structure of the heart (such as an internal chamber and/or a blood vessel, etc.
  • the acquired three-dimensional data should contain the complete first organ and second organ structures of the object to be measured.
  • the three-dimensional data can be three-dimensional raw data or three-dimensional image data.
  • the user may directly collect three-dimensional data or four-dimensional data through the probe of the ultrasound device, or may obtain pre-stored three-dimensional data or four-dimensional data from a memory. If four-dimensional data is obtained, three-dimensional data at a moment may be selected from the four-dimensional data manually or automatically.
  • the three-dimensional data may be the three-dimensional ultrasound data with the best quality among the four-dimensional data.
  • the three-dimensional data can be selected according to the target recognition algorithm, the target detection algorithm or the target segmentation algorithm.
  • the target recognition algorithm can be used to extract features from the three-dimensional data at each moment, and then the data is scored and classified. After traversing the three-dimensional data at all moments in the four-dimensional data, the three-dimensional data with the highest score is selected as the three-dimensional data to be rendered.
  • the target detection algorithm can be used to detect the ovarian region and follicle region contained in the three-dimensional data at each moment, and the three-dimensional data to be rendered can be selected according to the number and volume of the detected ovarian region and follicle region.
  • the ovarian region and follicle region contained in the three-dimensional data at each moment are segmented respectively by the target segmentation algorithm, and the three-dimensional data to be rendered is selected according to the number and volume of the segmented ovarian region and follicle region.
  • the specific selection process and algorithm are similar to those described in relation to FIG. 2, and will not be repeated here.
  • a first organ region corresponding to the first organ and a second organ region corresponding to the second organ are positioned in the three-dimensional data to obtain a positioning detection result.
  • the positioning detection result includes the type (such as the first organ and the second organ) and the position of the organ region.
  • the first organ region and the second organ region may be positioned and detected according to a target detection algorithm or a target segmentation algorithm.
  • the first organ region and the second organ region contained in the three-dimensional data at each moment may be detected by a target detection algorithm, thereby obtaining the type and position information of the organ region.
  • the first organ region and the second organ region contained in the three-dimensional data may be segmented by a target segmentation algorithm (direct segmentation or segmentation of the two-dimensional section and then fusion), thereby obtaining the type and position information of the organ region.
  • the user may manually trace the first organ region and the second organ region, and the processor receives the user's tracing operation, and segments the first organ region and the second organ region in the three-dimensional data to obtain a segmentation result.
  • one of the first organ region and the second organ region may be positioned and detected in the three-dimensional data according to a target detection algorithm, and another of the first organ region and the second organ region may be positioned and detected in the three-dimensional data according to a target segmentation algorithm.
  • the first organ region on the outside is positioned and detected in the three-dimensional data according to a target detection algorithm
  • the second organ region on the inside is positioned and detected in the three-dimensional data according to a target segmentation algorithm.
  • the specific positioning detection process and algorithm are similar to those described in relation to FIG. 2, and will not be repeated here.
  • the transparency coefficients corresponding to different parts of the first organ region are determined.
  • the first organ region can be automatically determined as an organ wrapped outside, that is, an organ that needs to be transparently rendered, and the transparency coefficients corresponding to different parts thereof are determined.
  • the user can select the first organ region, and the processor responds to the user's selection operation, and regards the first organ region as an organ that needs to be transparently rendered, and then determines the transparency coefficients corresponding to different parts thereof.
  • the entire first organ region is rendered transparently, and each part has a different transparency coefficient.
  • the corresponding transparency coefficient can be adaptively determined for each triangular facet (rendering method using face rendering) or each pixel (rendering method using volume rendering) of the entire first organ region, and the transparency coefficient determination rule is pre-set. For example, the more perpendicular the normal line of the surface of the first organ region and the line of sight are, the smaller the transparency coefficient is, and the more opaque it is. For another example, the more perpendicular the angle between the normal line of the surface of the first organ region and a fixed direction is, the smaller the transparency coefficient is, and the more opaque it is. For another example, the angle between the normal line of the surface of the first organ region and different directions is calculated to determine the transparency coefficient.
  • the front portion of the first organ region in the user's perspective presents a plurality of different transparencies, while the other portions of the first organ region in the user's perspective present the same transparencies.
  • the front part of the first organ region under the user's perspective can be automatically detected, and a corresponding transparency coefficient is determined for each triangular facet or each pixel of the front part of the first organ region.
  • the method 300 may also include: based on the positioning detection result, determining the front part of the first organ region under the user's perspective, and determining corresponding transparency coefficients for the front part and other parts respectively. Specifically, the spatial depth information or the normal direction of the contour of the first organ region can be obtained based on the positioning detection result.
  • the spatial depth information or the normal direction of the contour of the first organ region can be obtained, for example, from the vertex coordinate information of the model used to construct the first organ region. Afterwards, based on the spatial depth information or the normal direction of the contour, the front part of the first organ region under the user's perspective is determined, and corresponding transparency coefficients are determined for the front part and other parts respectively.
  • the fixed transparency coefficient can be, for example, 0 (i.e., opaque), or can be less than the transparency coefficient of the front part (i.e., the transparency is less than the transparency of the front part).
  • the fixed transparency coefficient can be pre-set or input by the user.
  • the target part of the first organ region under the user's perspective presents a variety of different transparencies, while the other parts of the first organ region under the user's perspective present the same transparency or opacity.
  • the target part of the first organ region to be rendered transparently can be determined first, and then the corresponding transparency coefficient is determined for each triangular facet or each pixel corresponding to the target part of the first organ region.
  • the target part can be automatically determined (such as the part that wraps the second organ under the user's perspective), or it can be selected by the user, and the processor determines it in response to the user's selection operation.
  • method 200 can also include: determining the target part to be rendered transparently in the first organ region, and determining the corresponding transparency coefficients for the target part and other parts respectively.
  • the fixed transparency coefficient can be, for example, 0 (i.e., opaque), or can be less than the transparency coefficient of the target part (i.e., the transparency is less than the transparency of the front part).
  • the fixed transparency coefficient can be pre-set or input by the user.
  • the following describes a process of adaptively determining the transparency coefficient for the entire first organ region or a portion thereof.
  • model vertex coordinate information of the first organ region can be extracted from the three-dimensional data based on the positioning detection result.
  • the model vertex coordinate information is used to form a plurality of triangular facets for constructing a mesh model of the first organ region.
  • the perspective view corresponding to at least some of the plurality of triangular facets is determined. Specifically, if the transparency coefficient is adaptively determined for the entire first organ region, corresponding transparency coefficients are determined for all triangular facets of the mesh model of the first organ region. If the transparency coefficient is adaptively determined only for the front part or the target part of the first organ region, corresponding transparency coefficients are determined for those triangular facets in the mesh model of the first organ region that correspond to the front part or the target part.
  • the normal vector of the triangle can be calculated based on the coordinate information of the three vertices of the triangle, and then the angle between the normal vector and the preset reference direction can be calculated. Finally, the transparency coefficient corresponding to the triangle is determined according to the size of the angle.
  • the direction vector between the three vertices of the triangle and the center of gravity of the model can be calculated based on the coordinate information of the three vertices of the triangle, and then the angle between the direction vector and the preset reference direction can be calculated. Finally, the transparency coefficient corresponding to the triangle is determined according to the size of the angle.
  • angles between the three vertices and a preset reference direction are calculated based on the coordinate information of the three vertices of the triangular face, and then the transparency coefficient corresponding to the triangular face is determined based on the size of the angle.
  • the same preset reference direction may be used, such as the sight line direction or a fixed direction.
  • different preset reference directions may be used for different triangular facets.
  • the triangular facets may be grouped, and a preset reference direction may be used for each group of triangular facets.
  • the calculation rule of the transparency coefficient may be set according to the preset reference direction. For example, when the preset reference direction is the sight line direction, the calculation rule of the transparency coefficient may be set such that the larger the angle, the smaller the transparency coefficient.
  • a transparency coefficient corresponding to at least some of the multiple pixels used to draw the first organ region may be determined. Specifically, if the transparency coefficient is adaptively determined for the entire first organ region, the corresponding transparency coefficient is determined for all pixels in the first organ region. If the transparency coefficient is adaptively determined only for the front part or the target part of the first organ region, the corresponding transparency coefficient is determined for those pixels corresponding to the front part or the target part.
  • the boundary contour of the first organ region can be segmented based on the positioning detection result, and then the transparency coefficient corresponding to at least some of the pixels in the boundary contour of the first organ region can be determined according to the algorithm corresponding to the volume rendering.
  • a ray tracing algorithm is used.
  • volume rendering multiple light rays passing through three-dimensional data are emitted based on the line of sight, and each ray advances at a fixed step size to sample the three-dimensional data on the light path.
  • the opacity of each sampling point is determined according to the grayscale value of each sampling point, and the opacity of each sampling point on each light path is accumulated to obtain the cumulative opacity.
  • the cumulative opacity on each light path is mapped to a transparency coefficient, and the transparency coefficient is mapped to a pixel of the two-dimensional image. In this way, the transparency coefficients of the pixels corresponding to all light paths are obtained, and rendered images with different transparencies can be obtained.
  • a mapping relationship between grayscale values and transparency coefficients may be preset, grayscale values corresponding to multiple pixels within the boundary contour of the first organ region may be extracted from the three-dimensional data, and then the transparency coefficient corresponding to each pixel may be determined based on the mapping relationship.
  • step 304 the three-dimensional data is rendered according to the determined transparency coefficient to obtain a rendered image of the first organ region and the second organ region.
  • the rendered image different parts of the first organ region present different transparency.
  • the entire first organ region can be rendered transparently, and each part can be rendered with a different degree of transparency.
  • the front part of the first organ region can be rendered with a different degree of transparency from the user's perspective, while the other parts of the first organ region can be rendered with the same degree of transparency or opacity from the user's perspective.
  • the target part of the first organ region can be rendered with a different degree of transparency from the user's perspective, while the other parts of the ovarian region can be rendered with the same degree of transparency or opacity from the user's perspective.
  • FIG5 shows a schematic diagram of a computing device according to an embodiment of the present invention.
  • the computing device 500 includes a processor (e.g., a central processing unit (CPU)) 501 and a memory 502 coupled to the processor 501.
  • the memory 502 is used to store computer executable instructions, and when the computer executable instructions are executed, the processor 501 executes the method in the above embodiment.
  • the processor 501 and the memory 502 are connected to each other through a bus, and an input/output (I/O) interface is also connected to the bus.
  • I/O input/output
  • the computing device 500 may also include a plurality of components (not shown in FIG5) connected to the I/O interface, including but not limited to: an input unit, such as a keyboard, a mouse, etc.; an output unit, such as various types of displays, speakers, etc.; a storage unit, such as a disk, an optical disk, etc.; and a communication unit, such as a network card, a modem, a wireless communication transceiver, etc.
  • the communication unit allows the computing device 500 to exchange information/data with other devices through a computer network such as the Internet and/or various telecommunication networks.
  • the above method can be implemented by a computer-readable storage medium.
  • the machine-readable storage medium carries computer-readable program instructions for executing various embodiments of the present invention.
  • the computer-readable storage medium can be a tangible device that can hold and store instructions used by an instruction execution device.
  • the computer-readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the above.
  • Non-exhaustive examples of computer-readable storage media include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a protruding structure in a groove on which instructions are stored, and any suitable combination of the above.
  • RAM random access memory
  • ROM read-only memory
  • EPROM or flash memory erasable programmable read-only memory
  • SRAM static random access memory
  • CD-ROM compact disk read-only memory
  • DVD digital versatile disk
  • memory stick a floppy disk
  • mechanical encoding device such as a punch card or a protruding structure in a groove on which instructions are stored, and any suitable combination of the above.
  • the computer-readable storage medium used here is not interpreted as a transient signal itself, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagated by a waveguide or other transmission medium (for example, a light pulse through an optical fiber cable), or an electrical signal transmitted by a wire.
  • the present invention provides a computer-readable storage medium having computer-executable instructions stored thereon, wherein the computer-executable instructions are used to execute the methods in various embodiments of the present invention.
  • various example embodiments of the present invention may be implemented in hardware or dedicated circuits, software, firmware, logic, or any combination thereof. Certain aspects may be implemented in hardware, while other aspects may be implemented in firmware or software that may be executed by a controller, microprocessor, or other computing device.
  • firmware or software that may be executed by a controller, microprocessor, or other computing device.
  • Computer-readable program instructions or computer program products for executing the various embodiments of the present invention can also be stored in the cloud. When needed, users can access the computer-readable program instructions for executing an embodiment of the present invention stored in the cloud through mobile Internet, fixed network or other networks, thereby implementing the technical solutions disclosed in accordance with the various embodiments of the present invention.
  • the disclosed devices and methods can be implemented in other ways.
  • the device embodiments described above are only illustrative.
  • the division of the units is only a logical function division.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Graphics (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Geometry (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Health & Medical Sciences (AREA)
  • Architecture (AREA)
  • Computer Hardware Design (AREA)
  • General Engineering & Computer Science (AREA)
  • Ultra Sonic Daignosis Equipment (AREA)

Abstract

本发明提供了一种超声成像方法。该方法包括：获取被测对象的卵巢和卵巢包裹的多个卵泡的待渲染的三维数据；在三维数据中对卵巢对应的卵巢区域和多个卵泡对应的多个卵泡区域进行定位检测，以得到定位检测结果；基于定位检测结果，确定卵巢区域的不同部分对应的透明系数；以及根据所确定的透明系数对三维数据进行渲染，以得到卵巢区域和多个卵泡区域对应的渲染图像。在渲染图像中，卵巢区域的不同部分呈现不同的透明度。该方法使外侧的卵巢区域的不同部分呈现不同的透明度，增加了卵巢区域的立体感，呈现给用户位置关系清晰的渲染效果，使用户能够清楚地看到卵巢和卵泡之间的位置、大小和包含关系，提升了视觉体验感，有助于提高工作效率和准确性。

Description

一种超声成像方法及超声设备 技术领域
本发明涉及医疗器械的技术领域，更具体地，涉及一种超声成像方法及超声设备。
背景技术
人体中的很多器官或者组织结构存在包裹或者遮挡关系，如卵巢和卵泡、子宫和子宫内膜等等，此时，将不同的器官结构进行三维渲染显示时，就会看不到被遮挡的器官。因此，对外侧的器官进行透明渲染呈像可以更加清晰地表现出被遮挡器官的形状结构，以及两个器官的位置关系等，从而更好的协助医生进行临床诊断。例如，卵巢和卵泡三维混合渲染可以协助医生进行卵巢容受性分析，宫颈和宫体三维混合渲染可以使医生根据渲染出的宫颈和宫体的位置关系判断子宫的倾曲程度等。
在目前的混合渲染方式中，对需要透明显示的器官设置固定的透明系数，这使得器官渲染的立体感不够，无法体现器官在不同深度或角度的变换，视觉体验感较差，所呈现的器官之间的位置关系也不够清晰，可能会影响医生临床诊断的结果。
发明内容
在发明内容部分中引入了一系列简化形式的概念，这将在具体实施方式部分中进一步详细说明。本发明的发明内容部分并不意味着要试图限定出所要求保护的技术方案的关键特征和必要技术特征，更不意味着试图确定所要求保护的技术方案的保护范围。
鉴于上述技术问题，本发明的第一个方面提出了一种超声成像方法，包括：
获取被测对象的卵巢和所述卵巢包裹的多个卵泡的待渲染的三维数据；
在所述三维数据中对所述卵巢对应的卵巢区域和所述多个卵泡对应的 多个卵泡区域进行定位检测，以得到定位检测结果；
基于所述定位检测结果，确定所述卵巢区域的不同部分对应的透明系数；以及
根据所确定的透明系数对所述三维数据进行渲染，以得到所述卵巢区域和所述多个卵泡区域对应的渲染图像，其中，在所述渲染图像中，所述卵巢区域的不同部分呈现不同的透明度。
本发明的第二个方面提出了一种超声成像方法，包括：
获取被测对象的第一器官和第二器官的待渲染的三维数据，其中所述第一器官至少部分地包裹所述第二器官；
在所述三维数据中对所述第一器官对应的第一器官区域和所述第二器官对应的第二器官区域进行定位检测，以得到定位检测结果；
基于所述定位检测结果，确定所述第一器官区域的不同部分对应的透明系数；以及
根据所确定的透明系数对所述三维数据进行渲染，以得到所述第一器官区域和所述第二器官区域的渲染图像，其中，在所述渲染图像中，所述第一器官区域的不同部分呈现不同的透明度。
本发明的第三个方面提出了一种超声设备，包括：
超声探头；
发射和接收电路，其被配置为控制所述超声探头向被测对象的目标组织发射超声波，并控制所述超声探头接收所述超声波的回波信号；
存储器，其用于存储计算机可执行指令；
处理器，其被配置为在执行所述计算机可执行指令时，根据所述回波信号获得三维数据，并执行上述实施例中任一项所述的超声成像方法以生成渲染图像；以及
显示器，其被配置为显示所述渲染图像。
根据本发明，当需要渲染存在包裹或遮挡关系的器官时，可以确定外侧器官不同部分对应的透明系数，从而使得外侧器官的不同部分呈现不同的透明度，增加了外侧器官的立体感，因此能够呈现给用户位置关系清晰的渲染效果，使得用户能够清楚地看到器官之间的位置、大小和包含关系等，提升了视觉体验感，并有助于提高工作效率和准确性。
附图说明
为了更清楚地说明本发明实施例中的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
在附图中：
图1示出了根据本发明实施例的超声设备的示意性框图；
图2示出了根据本发明实施例的超声成像方法的一个示意性流程图；
图3a-3c示出了根据本发明实施例的渲染图像的示意图；
图4示出了根据本发明实施例的超声成像方法的另一个示意性流程图；
图5示出了根据本发明实施例的计算设备的示意图。
具体实施方式
为了使得本发明的目的、技术方案和优点更为明显，下面将参照附图详细描述根据本发明的示例实施例。显然，所描述的实施例仅仅是本发明的一部分实施例，而不是本发明的全部实施例，应理解，本发明不受这里描述的示例实施例的限制。基于本发明中描述的本发明实施例，本领域技术人员在没有付出创造性劳动的情况下所得到的所有其它实施例都应落入本发明的保护范围之内。
在下文的描述中，给出了大量具体的细节以便提供对本发明更为彻底的理解。然而，对于本领域技术人员而言显而易见的是，本发明可以无需一个或多个这些细节而得以实施。在其他的例子中，为了避免与本发明发生混淆，对于本领域公知的一些技术特征未进行描述。
应当理解的是，本发明能够以不同形式实施，而不应当解释为局限于这里提出的实施例。相反地，提供这些实施例将使公开彻底和完全，并且将本发明的范围完全地传递给本领域技术人员。
在此使用的术语的目的仅在于描述具体实施例并且不作为本发明的限制。在此使用时，单数形式的“一”、“一个”和“所述/该”也意图包括复数形式，除非上下文清楚指出另外的方式。还应明白术语“组成”和/或“包 括”，当在该说明书中使用时，确定所述特征、整数、步骤、操作、元件和/或部件的存在，但不排除一个或更多其它的特征、整数、步骤、操作、元件、部件和/或组的存在或添加。在此使用时，术语“和/或”包括相关所列项目的任何及所有组合。
为了彻底理解本发明，将在下列的描述中提出详细的结构，以便阐释本发明提出的技术方案。本发明的可选实施例详细描述如下，然而除了这些详细描述外，本发明还可以具有其他实施方式。
本发明提供了一种超声成像方法和超声设备，当需要渲染存在包裹或遮挡关系的器官时，可以确定外侧器官不同部分对应的透明系数，从而使得外侧器官的不同部分呈现不同的透明度，增加了外侧器官的立体感，因此能够呈现给用户位置关系清晰的渲染效果，使得用户能够清楚地看到器官之间的位置、大小和包含关系等，提升了视觉体验感，并有助于增加诊断分析的准确性。
下面，首先参考图1描述根据本发明一个实施例的超声设备。图1示出了根据本发明实施例的超声设备的示意性框图。
如图1所示，超声设备100包括超声探头110、发射和接收电路111、处理器112、存储器113和显示器114。进一步地，超声设备100还可以包括波束合成电路和发射/接收选择开关等。
超声探头110通常包括多个阵元的阵列。在每次发射超声波时，超声探头110的所有阵元或者部分阵元参与超声波的发射。此时，这些参与超声波发射的阵元中的每个阵元或者每部分阵元分别受到发射脉冲的激励并分别发射超声波，这些阵元分别发射的超声波在传播过程中发生叠加，形成被发射到被测对象的感兴趣区域所在区域的合成超声波束。例如，感兴趣区域可以为卵巢区域、子宫区域等。
发射和接收电路111可以通过发射和接收选择开关与超声探头110耦合。发射和接收选择开关也可以被称为发送和接收控制器，其可以包括发送控制器和接收控制器。发送控制器用于激励超声探头110经由发射电路向被测对象的感兴趣区域发射超声波；接收控制器用于通过超声探头110经由接收电路接收从被测对象的感兴趣区域返回的超声回波，从而获得超声波的回波信号。之后，发射和接收电路111将回波信号送入波束合成电 路，波束合成电路对该电信号进行聚焦延时、加权和通道求和等处理，然后将处理后的超声回波数据送入处理器112。
可选地，处理器112可以通过软件、硬件、固件或其任意组合来实现，可以使用电路、单个或多个专用集成电路(Application Specific Integrated Circuit，ASIC)、单个或多个通用集成电路、单个或多个微处理器、单个或多个可编程逻辑器件、或者前述电路或器件的任意组合、或者其他适合的电路或器件，从而使得处理器112可以执行本说明书中的各个实施例中的方法的相应步骤。并且，处理器112可以控制超声设备100中的其它组件以执行期望的功能。
处理器112对其接收到的超声波的回波信号进行处理，得到被测对象的感兴趣区域的三维超声图像。在此过程中，超声探头110在一系列扫描平面内发射或接收超声波，由处理器112根据其三维空间关系进行整合，实现被测对象的感兴趣区域在三维空间的扫描以及三维图像的重建。最后，由处理器114对其进行去噪、平滑、增强等部分或全部图像后处理步骤后，获取被测对象的感兴趣区域的三维超声图像。得到的三维超声图像可以存储在存储器113中，也可以显示在显示器114上，还可以经由有线或无线通信线路传送到其它存储设备以进行存储。
存储器113用于存储处理器执行的指令、存储接收到的超声波的回波信号、存储超声图像，等等。存储器可以为闪存卡、固态存储器、硬盘等。其可以为易失性存储器和/或非易失性存储器，为可移除存储器和/或不可移除存储器等。
显示器114与处理器112通信耦合。显示器112可以是触摸显示屏、液晶显示屏等。尽管在本实施例中，显示器112被示出为是超声设备100的一部分，但是在其它实施例中，显示器112也可以是独立于超声设备100的液晶显示器、电视机等独立的显示设备；或者，显示器112还可以是智能手机、平板电脑等电子设备的显示屏，等等。显示器112的数量可以为一个或多个。例如，显示器112可以包括主屏和触摸屏，主屏主要用于显示超声图像，触摸屏主要用于人机交互。
显示器114可以显示处理器112得到的超声图像。此外，显示器114在显示超声图像的同时还可以向用户提供用于人机交互的图形界面。在图 形界面上设置有一个或多个被控对象。用户利用人机交互装置输入操作指令来控制这些被控对象，从而执行相应的控制操作。例如，在图形界面上显示图标，利用人机交互装置可以对该图标进行操作，用来执行特定的功能。
可选地，超声设备100还可以包括显示器114之外的其他人机交互装置，其与处理器112通信耦合。例如，处理器112可以通过外部输入/输出端口与人机交互装置连接，外部输入/输出端口可以是无线通信模块，也可以是有线通信模块，或者两者的组合。外部输入/输出端口也可基于USB、如CAN等总线协议、和/或有线网络协议等来实现。
人机交互装置可以包括输入设备，用于检测用户的输入信息，该输入信息例如可以是对超声波发射/接收时序的控制指令，可以是在超声图像上绘制出点、线或框等的操作输入指令，或者还可以包括其他指令类型。输入设备可以包括键盘、鼠标、滚轮、轨迹球、移动式输入设备(比如带触摸显示屏的移动设备、手机等等)、多功能旋钮等等其中之一或者多个的结合。人机交互装置还可以包括诸如打印机之类的输出设备。
应理解，图1所示的超声设备100所包括的部件只是示意性的，其可以包括更多或更少的部件。本发明对此不限定。
下面将参考图2描述根据本发明实施例的一个超声成像方法。图2示出了根据本发明实施例的超声成像方法的一个示意性流程图。在一些实施例中，可以由图1中的超声设备100的处理器112执行图2中的超声成像方法200，得到渲染图像并经由显示器114进行显示。在另一些实施例中，可以由任何其它计算设备的处理器执行该超声成像方法100，得到渲染图像并经由与该处理器通信耦合的显示器进行显示。
在图2的方法中，确定外侧的卵巢区域的不同部分对应的透明系数，从而使外侧的卵巢区域的不同部分呈现不同的透明度，增加了卵巢区域的立体感，呈现给用户位置关系清晰的渲染效果，使用户能够清楚地看到卵巢和卵泡之间的位置、大小和包含关系等，提升了视觉体验感，并有助于提高工作效率和准确性。
参考图2，根据本发明实施例的超声成像方法200包括如下步骤：
在步骤201中，获取被测对象的卵巢和卵巢包裹的多个卵泡的待渲染 的三维数据。所获取的三维数据应当包含被测对象完整的卵巢和卵泡结构。三维数据可以是三维原始数据或者三维图像数据。
在一些实施方式中，可以由用户通过超声设备的探头对被测对象的卵巢组织进行扫查，采集三维数据或四维数据。具体地，首先，由超声设备的超声探头发射超声波至被测对象的卵巢组织。卵巢组织可以是包括卵巢和卵巢包裹的多个卵泡。发射该超声波的时候需要将该超声波发射至被测对象的内部。在一种可行的实施方式中，可以是将超声探头放置在被测对象体表对应卵巢组织的位置。在另一种可行的实施方式中，也可以是通过腔内超声扫描的方式使超声探头的扫描区域覆盖卵巢组织，然后通过超声探头发射超声波，进而实现将超声波发射值被测对象的卵巢内部。被测对象可以是人体器官或人体组织结构等包括卵巢组织的对象。接着，超声探头接收从被测对象的卵巢组织返回的超声波的回波信号，并对回波信号进行波束合成、三维重建等处理得到三维超声图像或四维超声图像。
在一些实施方式中，可以从存储器中获取预先存储的三维数据或四维数据。可以由用户通过超声设备的探头对被测对象的卵巢组织进行扫查，获得三维数据或四维数据后存储在存储器中或发送给其它计算设备进行存储。在后续需要图像渲染时，由其它计算设备从存储器中获取三维数据或四维数据。
在一些实施方式中，如果获得四维数据，需要从四维数据中选取一个时刻的三维数据作为待渲染的三维数据。可以通过手动或自动的方式选择三维数据。用户可以通过移动鼠标、轨迹球之类的输入设备对四维数据进行查看，选择包含完整的卵巢和卵泡结构的三维数据作为待渲染的三维数据。自动的三维数据选择方式可以通过应用机器学习或深度学习算法对四维数据进行自动识别，选取待渲染的三维数据。
在一些实施方式中，可以根据目标识别算法从四维数据中选取待渲染的三维数据。目标识别算法通过直接对每一时刻的三维数据提取特征，然后对数据进行评分分类实现。该算法的主要步骤为：1)构建数据库：数据库包括大量的三维数据集合及其对应的标定结果。标定结果为是否是标准的待渲染的三维数据(如是否包含了完整的器官结构)。2)待渲染的三维数据的识别、定位步骤。目标识别算法可以包括基于传统机器学习的目标 识别算法和基于深度学习的目标识别算法。
基于传统机器学习的目标识别算法通过对三维数据进行特征的提取，如局部的上下文信息、纹理信息、Harr特征等，再整合相关的特征输入与之级联的分类器，如支持向量机(SVM)、Adaboost、随机森林(Random Forest)等，通过对特征进行判别分类评分。在遍历四维数据中的所有时刻的三维数据之后，选取评分最高的三维数据作为待渲染的三维数据。
基于深度学习的目标识别算法通过神经网络架构实现，属于分类网络，其主要结构为卷积层、激活层、池化层和全连接层的堆叠，通过浅层的卷积层对三维数据进行相关特征的提取，再在全连接层对提取到的特征进行线性组合，最后输出当前图像的概率。在遍历四维数据中的所有时刻的三维数据之后，选取概率最大的三维数据作为待渲染的三维数据。常见的分类网络有3D FCN等。
在一些实施方式中，可以根据目标检测算法从四维数据中选取待渲染的三维数据。通过目标检测算法检测出每一时刻的三维数据所包含的卵巢区域和卵泡区域，根据检测到的卵巢区域和卵泡区域的数量、体积等选择待渲染的三维数据。该算法的主要步骤为：1)构建数据库：数据库包括大量的三维数据集合及其对应的标定结果。标定结果为三维数据的卵巢区域和卵泡区域的边界框(bounding box)。2)标准待渲染的三维数据的检测、定位步骤。目标检测算法可以包括基于传统机器学习的目标检测算法和基于深度学习的目标检测算法。
基于传统机器学习的目标检测算法主要分为以下步骤：1)首先通过移动不同的尺度和不同的长宽比滑动窗进行区域选择；2)基于区域内的图像块提取相关特征(如Harr特征，HOG特征等)，3)将提取的特征送入分类器(如SVM，Adaboost等)中进行分类确定卵巢区域或卵泡区域。
基于深度学习的目标检测算法分为利用候选区域和深度学习分类的目标检测方法、基于深度学习的回归方法和基于点云的目标检测方法。
利用候选区域和深度学习分类的目标检测方法通过提取候选区域，对相应区域进行深度学习为主的分类方案，如将R-CNN(Selective Search+CNN+SVM)，SPP-Net(ROI Pooling)，Fast R-CNN(Selective Search+CNN+ROI)，Faster R-CNN(RPN+CNN+ROI)，R-FCN等的卷积核改造成三 维来进行分类。
基于深度学习的回归方法的主要步骤是首先将图像切分为S×S×S个网格，然后让每个网格负责检测物体中心落在这个网格上的物体(如卵巢区域和卵泡区域)，最后网络输出这个物体的每个位置的坐标(如中心点坐标以及长宽等)和该物体所属类别的概率，如将YOLO系列(YOLO V1,YOLO V2,YOLO V3)的卷积核改造为三维、SSD、DenseBox等算法来实现。
基于点云的目标检测方法分为基于点和基于体素的两种类型的检测方式。基于点的检测方式是直接在点云数据上应用深度学习模型的方法，输入是n×3的点云张量，网络对点云数据进行特征提取后，在基于特征在规范坐标中修改获得最终得检测结果，网络例如为Point-RCNN。基于体素的方法则是将点云划分为3D体素，由三维的CNN处理，再使用多个检测头来探索位置以提升检测性能。另外，还可以将基于点和基于体素的检测方式结合起来，将三维体素卷积神经网络(CNN)和基于点网的集合抽象技术结合起来，以准确估算三维对象的位置(如PV-RCNN，point-voxel RCNN)。
通过上述目标检测算法，可以得到卵巢区域的位置、卵泡区域的位置和数量等信息，通过从四维数据中选取卵巢区域的体积(通过边界框的大小估算)最大的三维数据，也可以同时将卵泡区域的数量作为选择标准。所选取的三维数据即为待渲染的三维数据。
在一些实施方式中，可以根据目标分割算法从四维数据中选取待渲染的三维数据。通过目标分割算法分别将每一时刻的三维数据所包含的卵巢区域和卵泡区域进行分割，根据分割到的卵巢区域和卵泡区域的数量、体积等选择待渲染的三维数据。在分割时，可以直接采用目标分割算法对三维数据所包含的卵巢区域和卵泡区域进行分割，也可以将三维数据拆分成数个二维数据，例如以卵巢区域中心为轴进行放射性采用或平行采样，对二维数据分割后，再将数个分割结果融合得到最终的三维分割结果。
传统的分割算法有如基于水平集(Level Set)的分割算法，随机游走(Random Walker)，图割(Graph Cut)，Snake等。目标分割算法还可以包括基于传统机器学习的目标分割算法和基于深度学习的目标分割算法。
基于传统机器学习的目标分割算法的主要步骤为：1)构建数据库：数 据库包括大量的超声数据集合及其对应的标定结果(在对三维数据直接分割的情况下，为三维数据集合，在对三维数据包括的多个二维数据进行分割的情况下，为二维数据集合)。标定结果为超声数据的卵巢区域和卵泡区域的掩膜(mask)，即分割结果。2)分割步骤：将超声数据分为切分为多个数据块(在对三维数据直接分割的情况下，为S×S×S大小的图像块，在对三维数据包括的二维数据进行分割的情况下，为S×S大小的图像块)，然后对数据块进行特征提取，特征的提取方式包括传统的PCA、LDA、Harr特征、纹理特征等，也可以使用深度神经网络(如Overfeat网络)来进行特征的提取，然后对提取的特征使用级联的分类器，如KNN、SVM、随机森林等判别器进行分类，从而确定当前数据块是否为卵巢区域或卵泡区域，将该分类结果作为当前数据块的中心点标记结果，最后得到整个超声数据的分割结果。
基于深度学习的目标分割算法的主要步骤为：1)构建数据库：数据库包括大量的超声数据集合及其对应的标定结果(在对三维数据直接分割的情况下，为三维数据集合，在对三维数据包括的多个二维数据进行分割的情况下，为二维数据集合)。标定结果为超声数据的卵巢区域和卵泡区域的掩膜(mask)，即分割结果。2)卵巢区域和卵泡区域的分割，即端到端的语义分割算法。输入可以是图像，也可以是三维点云(在对三维数据直接分割的情况下)。在图像作为输入的情况下，通过卷积层、池化层、上采样或者反卷积层等的堆叠，得到一个和输入图像尺寸一致的输出图像，该输出图像直接分割出需要的目标器官区域，该方法是一个监督学习，常见的二维网络有FCN、U-Net、Mask R-CNN等，三维分割网络有3D U-Net、3D FCN、Medical-Net等。在点云作为输入的情况下，a.首先将数据表现为一个点云的集合，表示为n×3的张量，其中n表示点云的数量，b.将输入数据通过一个学习到的转换矩阵对齐，保证模型对特征空间转换的不变性，c.对点云数据进行特征提取，d.对特征进行对齐，e.将全局特征和之前学习到的点云局部特征进行串联，在进行上采样等，得到每个数据点的分类结果作为分割结果，如PointNet或者pointNet++。
如果将三维数据拆分成数个二维数据，对二维数据进行分割，则需要对二维分割结果进行融合得到最终的三维分割结果。融合方式可以直接融 合得到三维分割结果，例如直接加在一起，或者取像素点最大值、平均值等作为分割结果，也可以通过插值方式(适用于放射状采样方式)融合得到三维分割结果。
通过上述目标分割算法，可以得到三维数据中卵巢区域的位置、卵泡区域的位置和数量等信息，通过从四维数据中选取卵巢区域的体积(通过掩膜的区域估算)最大的三维数据，也可以同时将卵泡的数量作为选择标准。所选取的三维数据即为待渲染的三维数据。
继续参考图2，在步骤202中，在三维数据中对卵巢对应的卵巢区域和多个卵泡对应的多个卵泡区域进行定位检测，以得到定位检测结果。定位检测结果包括器官区域的类型(如卵巢和卵泡)和位置。
在一些实施例中，可以根据目标检测算法对卵巢区域和卵泡区域进行定位检测。目标检测算法通过检测出每一时刻的三维数据所包含的卵巢区域和卵泡区域，从而得到器官区域的类型和位置信息。该算法的主要步骤为：1)构建数据库：数据库包括大量的三维数据集合及其对应的标定结果。标定结果为卵巢区域和卵泡区域的边界框(bounding box)。2)卵巢区域和卵泡区域的检测、定位步骤。目标检测算法可以分为基于传统机器学习的目标检测算法、基于深度学习的目标检测算法和基于点云的目标检测算法。算法的具体步骤与步骤201中采用的目标检测算法类似，在上面进行了详细描述，在此不再赘述。
在一些实施方式中，可以根据目标分割算法对卵巢区域和卵泡区域进行定位检测。与步骤201中采用的目标分割算法类似，在分割时，可以直接采用目标分割算法对三维数据所包含的卵巢区域和卵泡区域进行分割，也可以将三维数据拆分成数个二维数据，例如以卵巢区域中心为轴进行放射性采用或平行采样，对二维切面分割后，再将数个分割结果融合得到最终的三维分割结果。传统的分割算法有如基于水平集(Level Set)的分割算法，随机游走(Random Walker)，图割(Graph Cut)，Snake等。目标分割算法还可以包括基于传统机器学习的目标分割算法和基于深度学习的目标分割算法。二维或三维目标分割算法的具体步骤与步骤201中采用的目标分割算法类似，在上面进行了详细描述，在此不再赘述。
在一些实施方式中，用户可以手动地对卵巢区域和卵泡区域进行描迹， 例如，用户通过移动鼠标、轨迹球之类的输入设备在三维超声图像中描迹出卵巢区域和卵泡区域，处理器接收到用户的描迹操作，在三维超声图像中对卵巢区域和卵泡区域进行分割，得到分割结果。
在一些实施方式中，可以根据目标检测算法，在三维数据中对卵巢区域和多个卵泡区域中的一部分区域进行定位检测，并根据目标分割算法，在三维数据中对卵巢区域和多个卵泡区域中的另一部分区域进行定位检测。也就是说，可以根据不同的定位检测算法对卵巢区域和多个卵泡区域进行定位检测。例如，根据目标检测算法在三维数据中对卵巢区域进行定位检测，并根据目标分割算法在三维数据中对多个卵泡区域进行定位检测。又例如，根据目标检测算法在三维数据中对卵巢区域和一部分卵泡区域进行定位检测，并根据目标分割算法在三维数据中对其它卵泡区域进行定位检测。再例如，根据目标分割算法在三维数据中对卵巢区域进行定位检测，并根据目标检测算法在三维数据中对多个卵泡区域进行定位检测。
继续参考图2，在步骤203中，基于定位检测结果，确定卵巢区域的不同部分对应的透明系数。
在一些实施方式中，对整个卵巢区域都进行透明渲染，且各部分具有不同的透明系数。在这些实施方式中，可以自适应地为整个卵巢区域的每个三角面片(采用面绘制的渲染方式)或每个像素(采用体绘制的渲染方式)确定对应的透明系数，并预先设定透明系数的确定规则。例如，卵巢区域表面的法线和视线越垂直，透明系数越小，越不透明。又例如，卵巢区域表面的法线与某个固定方向夹角越垂直，透明系数越小，越不透明。再例如，将卵巢区域表面的法线与不同的方向计算角度，确定透明系数。
在一些实施方式中，卵巢区域在用户视角下的前侧部分呈现多种不同的透明度，而卵巢区域在用户视角下的其它部分呈现相同的透明度或不透明。可以自动检测卵巢区域在用户视角下的前侧部分，并为卵巢区域的前侧部分所对应的每个三角面片或每个像素确定对应的透明系数。在这些实施方式中，方法200还可以包括：基于定位检测结果，确定卵巢区域在用户视角下的前侧部分，并分别为前侧部分和其它部分确定对应的透明系数。具体的，可以基于定位检测结果，获取卵巢区域的空间深度信息或者卵巢轮廓的法线方向。卵巢区域的空间深度信息或者轮廓的法线方向，例如可 以从用于构建卵巢区域的模型顶点坐标信息获得。之后，基于空间深度信息或者轮廓的法线方向，确定卵巢区域在用户视角下的前侧部分，并分别为前侧部分和其它部分确定对应的透明系数。固定的透明系数例如可以为0(即不透明)，或者可以小于前侧部分的透明系数(即透明度小于前侧部分的透明度)。固定的透明系数可以是预先设定的，也可以由用户进行输入。
在一些实施方式中，卵巢区域在用户视角下的目标部分(如用户视角下左边或右边的一半，或者一个子区域)呈现多种不同的透明度，而卵巢区域在用户视角下的其它部分呈现相同的透明度或不透明。可以先确定该目标部分，再为该目标部分所对应的每个三角面片或每个像素确定对应的透明系数。目标部分可以是自动确定的(如根据用户视角)，也可以由用户进行选择，处理器响应于用户的选择操作而进行确定。在这些实施方式中，方法200还可以包括：在卵巢区域中确定待透明渲染的目标部分，并分别为目标部分和其它部分确定对应的透明系数。固定的透明系数例如可以为0(即不透明)，或者可以小于目标部分的透明系数(即透明度小于前侧部分的透明度)。固定的透明系数可以是预先设定的，也可以由用户进行输入。
接下来描述为整个卵巢区域或其一部分自适应地确定透明系数的过程。
在一些实施方式中，采用面绘制的方式渲染图像，可以基于定位检测结果，从三维数据中提取卵巢区域的模型顶点坐标信息。模型顶点坐标信息用于形成构建卵巢区域的网格模型的多个三角面片。在提取模型顶点坐标信息之后，确定多个三角面片中的至少部分三角面片对应的透明系数。具体地，如果对整个卵巢区域都自适应地确定透明系数，则对卵巢区域的网格模型的所有三角面片都确定对应的透明系数。如果仅对卵巢区域的前侧部分或目标部分自适应地确定透明系数，则对卵巢区域的网格模型中与前侧部分或目标部分对应的那些三角面片确定对应的透明系数。
在一种可能的实施方式中，可以先根据三角面片的三个顶点的坐标信息计算三角面片的法向量，再计算法向量与预设参考方向之间的夹角，最后根据夹角的大小确定该三角面片对应的透明系数。
在一种可能的实施方式中，可以先根据三角面片的三个顶点的坐标信息计算三角面片的三个顶点与模型重心之间的方向向量，再计算方向向量与预设参考方向之间的夹角，最后根据夹角的大小确定该三角面片对应的 透明系数。
在一种可能的实施方式中，根据三角面片的三个顶点的坐标信息计算三个顶点与预设参考方向之间的夹角，再根据夹角的大小确定该三角面片对应的透明系数。
在上述实施方式中，可以使用相同的预设参考方向，例如视线方向或某个固定方向。或者，对于不同的三角面片，也可以使用不同的预设参考方向。或者，还可以将三角面片进行分组，对于每组三角面片使用一个预设参考方向。此外，可以根据预设参考方向设定透明系数的计算规则。例如，当预设参考方向为视线方向时，可以将透明系数的计算规则设定为夹角越大，透明系数越小。
在一些实施方式中，采用体绘制的方式渲染图像，可以确定用于绘制卵巢区域的多个像素中的至少部分像素对应的透明系数。具体地，如果对整个卵巢区域都自适应地确定透明系数，则对卵巢区域的所有像素都确定对应的透明系数。如果仅对卵巢区域的前侧部分或目标部分自适应地确定透明系数，则对与前侧部分或目标部分对应的那些像素确定对应的透明系数。
在一种可能的实施方式中，可以基于定位检测结果，分割出卵巢区域的边界轮廓，然后根据体绘制对应的算法确定卵巢区域的边界轮廓内多个像素中至少部分像素对应的透明系数。例如采用光线追踪算法，在体绘制的一个示例中，基于视线方向发射多根穿过三维数据的光线，每一根光线按固定步长进行递进，对光线路径上的三维数据进行采样，根据每个采样点的灰度值确定每个采样点的不透明度，再对每一根光线路径上各采样点的不透明度进行累积得到累积不透明度，最后将每一根光线路径上的累积不透明度映射为一个透明系数，再将该透明系数映射到二维图像的一个像素上，通过如此方式得到所有光线路径各自对应的像素的透明系数，即可得到不同透明度的渲染图像。
在一种可能的实施方式中，可以预先设定灰度值与透明系数之间的映射关系，从三维数据中提取卵巢区域的边界轮廓内多个像素对应的灰度值，再根据该映射关系确定各个像素对应的透明系数。
接下来，在步骤204中，根据所确定的透明系数对三维数据进行渲染， 以得到卵巢区域和多个卵泡区域对应的渲染图像。在该渲染图像中，卵巢区域的不同部分呈现不同的透明度。
在一些实施方式中，对整个卵巢区域都进行透明渲染，且各部分呈现不同的透明度。图3a示出了一个示意图。从图3a中可以看到，整个卵巢区域与视线越垂直的部分(如边缘部分)，透明度越低，即越不透明，而与用户视线夹角越小的部分(如靠近卵巢区域中间的部分)，透明度越高，即越透明。
在一些实施方式中，卵巢区域在用户视角下的前侧部分呈现不同的透明度，而卵巢区域在用户视角下的其它部分呈现相同的透明度或不透明。以这种方式，能够在不影响用户观察的情况下减小透明系数的计算量，加快渲染处理的速度。
在一些实施方式中，卵巢区域在用户视角下的目标部分呈现不同的透明度，而卵巢区域在用户视角下的其它部分呈现相同的透明度或不透明。图3b示出了一个示意图。从图3b中可以看到，卵巢区域右上侧部分呈现多种不同的透明度，其中与用户视线越垂直的部分(如边缘部分)透明度越低，即越不透明，而与用户视线夹角越小的部分(如靠近卵巢区域中间的部分)透明度越高，即越透明。卵巢区域的其它部分呈现相同的透明度且小于右上侧部分的透明度，即不如右上侧部分透明。图3c示出了另一个示意图。在图3c中可以看到，相比于图3b，卵巢区域右侧透明呈现的部分比图3b中稍大。通过自适应地为卵巢区域的目标部分确定对应的透明系数，能够在不影响用户观察的情况下减小透明系数的计算量，加快渲染处理的速度。
在一些实施方式中，可以采用任意颜色渲染卵巢区域和多个卵泡区域。卵巢区域和多个卵泡区域的渲染颜色可以不同。另外，每个卵泡区域的渲染颜色也可以不同，以对不同大小和位置的卵泡区域进行区分。在一个可能的实施方式中，方法200还可以包括：确定卵巢区域和多个卵泡区域的渲染颜色。之后，根据所确定的渲染颜色对三维数据进行渲染。以这种方式，能够更清晰地向用户呈现卵巢区域和各卵泡区域的位置和大小关系，从而进一步提高用户的视觉体验。
在一些实施方式中，除了卵巢区域，还可以对目标卵泡区域进行透明 渲染并确定对应的透明系数。目标卵泡区域可以是遮挡了其它卵泡区域的卵泡区域，也可以是根据需要选择的卵泡区域。目标卵泡区域可以是自动确定的，也可以由用户进行选择，处理器响应于用户的选择操作而进行确定。在渲染图像中，目标卵泡区域的不同部分呈现不同的透明度。在一个可能的实施方式中，方法200还可以包括：基于定位检测结果，在多个卵泡区域中确定在用户视角下遮挡了一个或多个其它卵泡区域的目标卵泡区域，再确定目标卵泡区域的不同部分对应的透明系数。具体的，可以基于定位检测结果，获取多个卵泡的空间深度信息或者轮廓的法线方向。卵泡区域的空间深度信息或者轮廓的法线方向例如可以从用于构建卵泡区域的模型顶点坐标信息获得。之后，基于空间深度信息或者轮廓的法线方向，在多个卵泡区域中确定在用户视角下遮挡了一个或多个其它卵泡区域的目标卵泡区域。在确定目标卵泡区域后，再确定目标卵泡区域的不同部分对应的透明系数。与卵巢区域类似，同样也可以对目标卵泡区域的整体或目标部分自适应地确定透明系数。例如，可以仅对目标卵泡区域与其它卵泡区域的重合部分(即遮挡部分)透明渲染，自适应地确定透明系数，而其它部分不进行透明渲染。具体地，可以为目标卵泡区域整体或目标部分所对应的每个三角面片或像素计算对应的透明系数。计算透明系数的过程与上面描述的相同，在此不再赘述。
下面将参考图4描述根据本发明实施例的另一个超声成像方法。图4示出了根据本发明实施例的超声成像方法的一个示意性流程图。在一些实施例中，可以由图1中的超声设备100的处理器112执行图4中的超声成像方法300，得到渲染图像并经由显示器114进行显示。在另一些实施例中，可以由任何其它计算设备的处理器执行该超声成像方法300，得到渲染图像并经由与该处理器通信耦合的显示器进行显示。
在图4的方法中，当需要渲染存在包裹或遮挡关系的器官时，可以确定外侧器官不同部分对应的透明系数，从而使得外侧器官的不同部分呈现不同的透明度，增加了外侧器官的立体感，因此能够呈现给用户位置关系清晰的渲染效果，使得用户能够清楚地看到器官之间的位置、大小和包含关系等，提升了视觉体验感，并有助于提高工作效率和准确性。
参考图4，根据本发明实施例的超声成像方法300包括如下步骤：
在步骤301中，获取被测对象的第一器官和第二器官的待渲染的三维数据，其中第一器官至少部分地包裹所述第二器官。第一器官和第二器官可以是具有包裹关系的任何器官或组织结构。例如，第一器官为卵巢，第二器官为卵泡；第一器官为宫体，第二器官为子宫内膜；第一器官为胎儿颅脑，第二器官为颅脑内部结构(例如小脑、透明隔腔、丘脑和/或侧脑室等)；第一器官为胎儿腹部，第二器官为腹部内部结构(例如胃泡、胎心、脊柱、肾脏和/或血管等)；第一器官为肝脏，第二器官为肝内血管；第一器官为心脏，第二器官为心脏内部结构(例如心脏内部腔室和/或血管等)。此处仅对第一器官和第二器官的各种组合做举例说明，对类似具有包裹关系的第一器官和第二器官的组合都应当属于本申请所保护的范围。所获取的三维数据应当包含被测对象的完整的第一器官和第二器官结构。三维数据可以是三维原始数据或者三维图像数据。
与上述关于图2描述的类似，可以由用户通过超声设备的探头直接采集三维数据或四维数据，也可以从存储器中获取预先存储的三维数据或四维数据。如果获得四维数据，可以通过手动或自动的方式从四维数据中选取一个时刻的三维数据。该三维数据可以是四维数据中质量最优的三维超声数据。
可以根据目标识别算法、目标检测算法或目标分割算法选取三维数据。在一些实施方式中，可以通过目标识别算法对每一时刻的三维数据提取特征，然后对数据进行评分分类，在遍历四维数据中的所有时刻的三维数据之后，选取评分最高的三维数据作为待渲染的三维数据。在一些实施方式中，可以通过目标检测算法检测出每一时刻的三维数据所包含的卵巢区域和卵泡区域，根据检测到的卵巢区域和卵泡区域的数量、体积等选择待渲染的三维数据。在一些实施方式中，通过目标分割算法分别将每一时刻的三维数据所包含的卵巢区域和卵泡区域进行分割，根据分割到的卵巢区域和卵泡区域的数量、体积等选择待渲染的三维数据。具体的选取过程和算法与关于图2描述的内容类似，在此不再赘述。
在步骤302中，在三维数据中对第一器官对应的第一器官区域和第二器官对应的第二器官区域进行定位检测，以得到定位检测结果。定位检测结果包括器官区域的类型(如第一器官和第二器官)和位置。
可以根据目标检测算法或目标分割算法对第一器官区域和第二器官区域进行定位检测。在一些实施方式中，可以通过目标检测算法检测出每一时刻的三维数据所包含的第一器官区域和第二器官区域，从而得到器官区域的类型和位置信息。在一些实施方式中，可以通过目标分割算法对三维数据所包含的第一器官区域和第二器官区域进行分割(直接分割或对二维切面进行分割后再进行融合)，从而得到器官区域的类型和位置信息。在一些实施方式中，用户可以手动地对第一器官区域和第二器官区域进行描迹，处理器接收到用户的描迹操作，在三维数据中对第一器官区域和第二器官区域进行分割，得到分割结果。在一些实施方式中，可以根据目标检测算法，在三维数据中对第一器官区域和第二器官区域中的一个区域进行定位检测，并根据目标分割算法，在三维数据中对第一器官区域和第二器官区域中的另一个区域进行定位检测。例如，根据目标检测算法在三维数据中对外侧的第一器官区域进行定位检测，并根据目标分割算法在三维数据中对内侧的第二器官区域进行定位检测。具体的定位检测过程和算法与关于图2描述的内容类似，在此不再赘述。
在步骤303中，基于定位检测结果，确定第一器官区域的不同部分对应的透明系数。可以基于定位检测结果，自动确定第一器官区域为外侧包裹的器官，即需要透明渲染的器官，并确定其不同部分对应的透明系数。也可以由用户选择第一器官区域，处理器响应于用户的选择操作，将第一器官区域作为需要透明渲染的器官，再确定其不同部分对应的透明系数。
在一些实施方式中，对整个第一器官区域都进行透明渲染，且各部分具有不同的透明系数。在这些实施方式中，可以自适应地为整个第一器官区域的每个三角面片(采用面绘制的渲染方式)或每个像素(采用体绘制的渲染方式)确定对应的透明系数，并预先设定透明系数的确定规则。例如，第一器官区域表面的法线和视线越垂直，透明系数越小，越不透明。又例如，第一器官区域表面的法线与某个固定方向夹角越垂直，透明系数越小，越不透明。再例如，将第一器官区域表面的法线与不同的方向计算角度，确定透明系数。
在一些实施方式中，第一器官区域在用户视角下的前侧部分呈现多种不同的透明度，而第一器官区域在用户视角下的其它部分呈现相同的透明 度。在这些实施方式中，可以自动检测第一器官区域在用户视角下的前侧部分，并为第一器官区域前侧部分的每个三角面片或每个像素确定对应的透明系数。在这些实施方式中，方法300还可以包括：基于定位检测结果，确定第一器官区域在用户视角下的前侧部分，并分别为前侧部分和其它部分确定对应的透明系数。具体的，可以基于定位检测结果，获取第一器官区域的空间深度信息或者轮廓的法线方向。第一器官区域的空间深度信息或者轮廓的法线方向，例如可以从用于构建第一器官区域的模型顶点坐标信息获得。之后，基于空间深度信息或者轮廓的法线方向，确定第一器官区域在用户视角下的前侧部分，并分别为前侧部分和其它部分确定对应的透明系数。固定的透明系数例如可以为0(即不透明)，或者可以小于前侧部分的透明系数(即透明度小于前侧部分的透明度)。固定的透明系数可以是预先设定的，也可以由用户进行输入。
在一些实施方式中，第一器官区域在用户视角下的目标部分(如用户视角下左边或右边的一半，或者一个子区域)呈现多种不同的透明度，而第一器官区域在用户视角下的其它部分呈现相同的透明度或不透明。可以先确定第一器官区域的待透明渲染的目标部分，再为第一器官区域的目标部分所对应的每个三角面片或每个像素确定对应的透明系数。目标部分可以是自动确定的(如用户视角下包裹第二器官的部分)，也可以由用户进行选择，处理器响应于用户的选择操作而进行确定。在这些实施方式中，方法200还可以包括：在第一器官区域中确定待透明渲染的目标部分，并分别为目标部分和其它部分确定对应的透明系数。固定的透明系数例如可以为0(即不透明)，或者可以小于目标部分的透明系数(即透明度小于前侧部分的透明度)。固定的透明系数可以是预先设定的，也可以由用户进行输入。
接下来描述为整个第一器官区域或其一部分自适应地确定透明系数的过程。
在一些实施方式中，采用面绘制的方式渲染图像，可以基于定位检测结果，从三维数据中提取第一器官区域的模型顶点坐标信息。模型顶点坐标信息用于形成构建第一器官区域的网格模型的多个三角面片。在提取模型顶点坐标信息之后，确定多个三角面片中的至少部分三角面片对应的透 明系数。具体地，如果对整个第一器官区域都自适应地确定透明系数，则对第一器官区域的网格模型的所有三角面片都确定对应的透明系数。如果仅对第一器官区域的前侧部分或目标部分自适应地确定透明系数，则对第一器官区域的网格模型中与前侧部分或目标部分对应的那些三角面片确定对应的透明系数。
在一种可能的实施方式中，可以先根据三角面片的三个顶点的坐标信息计算三角面片的法向量，再计算法向量与预设参考方向之间的夹角，最后根据夹角的大小确定该三角面片对应的透明系数。
在一种可能的实施方式中，可以先根据三角面片的三个顶点的坐标信息计算三角面片的三个顶点与模型重心之间的方向向量，再计算方向向量与预设参考方向之间的夹角，最后根据夹角的大小确定该三角面片对应的透明系数。
在一种可能的实施方式中，根据三角面片的三个顶点的坐标信息计算三个顶点与预设参考方向之间的夹角，再根据夹角的大小确定该三角面片对应的透明系数。
在上述实施方式中，可以使用相同的预设参考方向，例如视线方向或某个固定方向。或者，对于不同的三角面片，也可以使用不同的预设参考方向。或者，还可以将三角面片进行分组，对于每组三角面片使用一个预设参考方向。此外，可以根据预设参考方向设定透明系数的计算规则。例如，当预设参考方向为视线方向时，可以将透明系数的计算规则设定为夹角越大，透明系数越小。
在一些实施方式中，采用体绘制的方式渲染图像，可以确定用于绘制第一器官区域的多个像素中的至少部分像素对应的透明系数。具体地，如果对整个第一器官区域都自适应地确定透明系数，则对第一器官区域的所有像素都确定对应的透明系数。如果仅对第一器官区域的前侧部分或目标部分自适应地确定透明系数，则对与前侧部分或目标部分对应的那些像素确定对应的透明系数。
在一种可能的实施方式中，可以基于定位检测结果，分割出第一器官区域的边界轮廓，然后根据体绘制对应的算法确定第一器官区域的边界轮廓内多个像素中至少部分像素对应的透明系数。例如采用光线追踪算法， 在体绘制的一个示例中，基于视线方向发射多根穿过三维数据的光线，每一根光线按固定步长进行递进，对光线路径上的三维数据进行采样，根据每个采样点的灰度值确定每个采样点的不透明度，再对每一根光线路径上各采样点的不透明度进行累积得到累积不透明度，最后将每一根光线路径上的累积不透明度映射为一个透明系数，再将该透明系数映射到二维图像的一个像素上，通过如此方式得到所有光线路径各自对应的像素的透明系数，即可得到不同透明度的渲染图像。
在一种可能的实施方式中，可以预设灰度值与透明系数之间的映射关系，从三维数据中提取第一器官区域的边界轮廓内多个像素对应的灰度值，再根据该映射关系确定各个像素对应的透明系数。
接下来，在步骤304中，根据所确定的透明系数对三维数据进行渲染，以得到第一器官区域和第二器官区域的渲染图像。在该渲染图像中，第一器官区域的不同部分呈现不同的透明度。
如上所述，可以对整个第一器官区域都进行透明渲染，且使得各部分呈现不同的透明度，也可以使第一器官区域在用户视角下的前侧部分呈现不同的透明度，而第一器官区域在用户视角下的其它部分呈现相同的透明度或不透明，还可以使第一器官区域在用户视角下的目标部分呈现不同的透明度，而卵巢区域在用户视角下的其它部分呈现相同的透明度或不透明。
本发明还提出了一种计算设备。图5示出了根据本发明一个实施例的计算设备的示意图。从图5中可以看出，计算设备500包括处理器(例如，中央处理单元(CPU))501以及与处理器501耦合的存储器502。存储器502用于存储计算机可执行指令，当计算机可执行指令被执行时使得处理器501执行以上实施例中的方法。处理器501和存储器502通过总线彼此相连，输入/输出(I/O)接口也连接至总线。计算设备500还可以包括连接至I/O接口的多个部件(图5中未示出)，包括但不限于：输入单元，例如键盘、鼠标等；输出单元，例如各种类型的显示器、扬声器等；存储单元，例如磁盘、光盘等；以及通信单元，例如网卡、调制解调器、无线通信收发机等。通信单元允许该计算设备500通过诸如因特网的计算机网络和/或各种电信网络与其他设备交换信息/数据。
此外，替代地，上述方法能够通过计算机可读存储介质来实现。计算 机可读存储介质上载有用于执行本发明的各个实施例的计算机可读程序指令。计算机可读存储介质可以是可以保持和存储由指令执行设备使用的指令的有形设备。计算机可读存储介质例如可以是但不限于电存储设备、磁存储设备、光存储设备、电磁存储设备、半导体存储设备或者上述的任意合适的组合。计算机可读存储介质的更具体的例子(非穷举)包括：便携式计算机盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、静态随机存取存储器(SRAM)、便携式压缩盘只读存储器(CD-ROM)、数字多功能盘(DVD)、记忆棒、软盘、机械编码设备、例如其上存储有指令的打孔卡或凹槽内凸起结构、以及上述的任意合适的组合。这里所使用的计算机可读存储介质不被解释为瞬时信号本身，诸如无线电波或者其他自由传播的电磁波、通过波导或其他传输媒介传播的电磁波(例如，通过光纤电缆的光脉冲)、或者通过电线传输的电信号。
因此，在另一个实施例中，本发明提出了一种计算机可读存储介质，该计算机可读存储介质具有存储在其上的计算机可执行指令，计算机可执行指令用于执行本发明的各个实施例中的方法。
一般而言，本发明的各个示例实施例可以在硬件或专用电路、软件、固件、逻辑，或其任何组合中实施。某些方面可以在硬件中实施，而其他方面可以在可以由控制器、微处理器或其他计算设备执行的固件或软件中实施。当本发明的实施例的各方面被图示或描述为框图、流程图或使用某些其他图形表示时，将理解此处描述的方框、装置、系统、技术或方法可以作为非限制性的示例在硬件、软件、固件、专用电路或逻辑、通用硬件或控制器或其他计算设备，或其某些组合中实施。
用于执行本发明的各个实施例的计算机可读程序指令或者计算机程序产品也能够存储在云端，在需要调用时，用户能够通过移动互联网、固网或者其他网络访问存储在云端上的用于执行本发明的一个实施例的计算机可读程序指令，从而实施依据本发明的各个实施例所公开内容的技术方案。
在本发明所提供的几个实施例中，应该理解到，所揭露的设备和方法，可以通过其它的方式实现。例如，以上所描述的设备实施例仅仅是示意性的，例如，所述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以 有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个设备，或一些特征可以忽略，或不执行。
在此处所提供的说明书中，说明了大量具体细节。然而，能够理解，本发明的实施例可以在没有这些具体细节的情况下实践。在一些实例中，并未详细示出公知的方法、结构和技术，以便不模糊对本说明书的理解。
类似地，应当理解，为了精简本发明并帮助理解各个发明方面中的一个或多个，在对本发明的示例性实施例的描述中，本发明的各个特征有时被一起分组到单个实施例、图、或者对其的描述中。然而，并不应将该本发明的方法解释成反映如下意图：即所要求保护的本发明要求比在每个权利要求中所明确记载的特征更多的特征。更确切地说，如相应的权利要求书所反映的那样，其发明点在于可以用少于某个公开的单个实施例的所有特征的特征来解决相应的技术问题。因此，遵循具体实施方式的权利要求书由此明确地并入该具体实施方式，其中每个权利要求本身都作为本发明的单独实施例。
本领域的技术人员可以理解，除了特征之间相互排斥之外，可以采用任何组合对本说明书(包括伴随的权利要求、摘要和附图)中公开的所有特征以及如此公开的任何方法或者设备的所有过程或单元进行组合。除非另外明确陈述，本说明书(包括伴随的权利要求、摘要和附图)中公开的每个特征可以由提供相同、等同或相似目的替代特征来代替。
此外，本领域的技术人员能够理解，尽管在此所述的一些实施例包括其它实施例中所包括的某些特征而不是其它特征，但是不同实施例的特征的组合意味着处于本发明的范围之内并且形成不同的实施例。例如，在权利要求书中，所要求保护的实施例的任意之一都可以以任意的组合方式来使用。
还应该注意的是，上述实施例对本发明进行说明而不是对本发明进行限制，并且本领域技术人员在不脱离所附权利要求的范围的情况下可设计出替换实施例。在权利要求中，不应将位于括号之间的任何参考符号构造成对权利要求的限制。本发明可以借助于包括有若干不同元件的硬件以及借助于适当编程的计算机来实现。在列举了若干装置的单元权利要求中，这些装置中的若干个可以是通过同一个硬件项来具体体现。词语第一、第 二、以及第三等的使用不表示任何顺序。可将这些单词解释为名称。
以上所述仅为本发明的具体实施方式或对具体实施方式的说明，本发明的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本发明揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本发明的保护范围之内。本发明的保护范围应以权利要求的保护范围为准。

Claims (26)

  1. 一种超声成像方法，其特征在于，包括：
    获取被测对象的卵巢和所述卵巢包裹的多个卵泡的待渲染的三维数据；
    在所述三维数据中对所述卵巢对应的卵巢区域和所述多个卵泡对应的多个卵泡区域进行定位检测，以得到定位检测结果；
    基于所述定位检测结果，确定所述卵巢区域的不同部分对应的透明系数；以及
    根据所确定的透明系数对所述三维数据进行渲染，以得到所述卵巢区域和所述多个卵泡区域对应的渲染图像，其中，在所述渲染图像中，所述卵巢区域的不同部分呈现不同的透明度。
  2. 根据权利要求1所述的方法，其特征在于，还包括：
    基于所述定位检测结果，确定所述卵巢区域在用户视角下的前侧部分，并且其中，
    确定所述卵巢区域的不同部分对应的透明系数进一步包括：分别确定所述前侧部分和所述卵巢区域的其它部分对应的透明系数，并且，
    在所述渲染图像中，所述卵巢区域的所述前侧部分呈现多种不同的透明度，所述其它部分呈现相同的透明度或不透明。
  3. 根据权利要求1所述的方法，其特征在于，还包括：
    在所述卵巢区域中确定待透明渲染的目标部分，并且其中，
    确定所述卵巢区域的不同部分对应的透明系数进一步包括：分别确定所述目标部分和所述卵巢区域的其它部分对应的透明系数，并且，
    在所述渲染图像中，所述卵巢区域的所述目标部分呈现多种不同的透明度，所述其它部分呈现相同的透明度或不透明。
  4. 根据权利要求1所述的方法，其特征在于，还包括：
    确定所述卵巢区域和所述多个卵泡区域的渲染颜色；以及
    根据所确定的渲染颜色对所述三维数据进行渲染。
  5. 根据权利要求1所述的方法，其特征在于，还包括：
    基于所述定位检测结果，在所述多个卵泡区域中确定在用户视角下遮挡了一个或多个其它卵泡区域的目标卵泡区域；
    确定所述目标卵泡区域的不同部分对应的透明系数，并且其中，
    在所述渲染图像中，所述目标卵泡区域的不同部分呈现不同的透明度。
  6. 根据权利要求1所述的方法，其特征在于，对所述三维数据进行渲染的渲染模式为面绘制，并且，基于所述定位检测结果，确定所述卵巢区域的不同部分对应的透明系数进一步包括：
    基于所述定位检测结果，从所述三维数据中提取所述卵巢区域的模型顶点坐标信息，所述模型顶点坐标信息用于形成构建所述卵巢区域的网格模型的多个三角面片；以及
    对于所述多个三角面片中的至少部分三角面片，执行以下步骤：
    根据所述三角面片的三个顶点的坐标信息计算所述三角面片的法向量；
    计算所述三角面片的法向量与预设参考方向的夹角；以及
    根据所述夹角的大小确定所述三角面片对应的透明系数。
  7. 根据权利要求1所述的方法，其特征在于，对所述三维数据进行渲染的渲染模式为面绘制，并且，基于所述定位检测结果，确定所述卵巢区域的不同部分对应的透明系数进一步包括：
    基于所述定位检测结果，从所述三维数据中提取所述卵巢区域的模型顶点坐标信息和模型重心坐标信息，所述模型顶点坐标信息用于形成构建所述卵巢区域的网格模型的多个三角面片；以及
    对于所述多个三角面片中的至少部分三角面片，执行以下步骤：
    根据所述三角面片的三个顶点的坐标信息计算所述三角面片的三个顶点与模型重心之间的方向向量；
    计算所述方向向量与预设参考方向之间的夹角；以及
    根据所述夹角的大小确定所述三角面片对应的透明系数。
  8. 根据权利要求1所述的方法，其特征在于，对所述三维数据进行渲染的渲染模式为面绘制，并且，基于所述定位检测结果，确定所述卵巢区域的不同部分对应的透明系数进一步包括：
    基于所述定位检测结果，从所述三维数据中提取所述卵巢区域的模型顶点坐标信息，所述模型顶点坐标信息用于形成构建所述卵巢区域的网格模型的多个三角面片；以及
    对于所述多个三角面片中的至少部分三角面片，执行以下步骤：
    根据所述三角面片的三个顶点的坐标信息计算所述三角面片的所述三个顶点与预设参考方向的夹角；以及
    根据所述夹角的大小确定所述三角面片对应的透明系数。
  9. 根据权利要求1所述的方法，其特征在于，对所述三维数据进行渲染的渲染模式为体绘制，并且，基于所述定位检测结果，确定所述卵巢区域的不同部分对应的透明系数进一步包括：
    基于所述定位检测结果，确定所述卵巢区域的边界轮廓；
    根据体绘制对应的算法确定所述卵巢区域的边界轮廓内多个像素中至少部分像素对应的透明系数。
  10. 根据权利要求1所述的方法，其特征在于，对所述三维数据进行渲染的渲染模式为体绘制，并且，基于所述定位检测结果，确定所述卵巢区域的不同部分对应的透明系数进一步包括：
    基于所述定位检测结果，确定所述卵巢区域的边界轮廓；
    从所述三维数据中提取所述卵巢区域的边界轮廓内多个像素中至少部分像素对应的灰度值；以及
    根据灰度值与透明系数之间的预设映射关系，确定所述至少部分像素对应的透明系数。
  11. 根据权利要求1所述的方法，其特征在于，在所述三维数据中对所述卵巢对应的卵巢区域和所述多个卵泡对应的多个卵泡区域进行定位检 测进一步包括：
    根据目标检测算法，在所述三维数据中对所述卵巢区域和所述多个卵泡区域进行定位检测；或者
    根据目标分割算法，在所述三维数据中对所述卵巢区域和所述多个卵泡区域进行定位检测；或者
    根据所述目标检测算法，在所述三维数据中对所述卵巢区域和所述多个卵泡区域中的一部分区域进行定位检测，并根据所述目标分割算法，在所述三维数据中对所述卵巢区域和所述多个卵泡区域中的另一部分区域进行定位检测。
  12. 根据权利要求1所述的方法，其特征在于，获取被测对象的卵巢和所述卵巢包裹的多个卵泡的待渲染的三维数据进一步包括：
    获取所述被测对象的所述卵巢和所述多个卵泡的四维超声图像；以及
    根据目标识别算法、目标检测算法和目标分割算法中的一个，从所述四维超声图像中选取一个时刻的三维超声图像作为所述待渲染的三维数据。
  13. 一种超声成像方法，其特征在于，包括：
    获取被测对象的第一器官和第二器官的待渲染的三维数据，其中所述第一器官至少部分地包裹所述第二器官；
    在所述三维数据中对所述第一器官对应的第一器官区域和所述第二器官对应的第二器官区域进行定位检测，以得到定位检测结果；
    基于所述定位检测结果，确定所述第一器官区域的不同部分对应的透明系数；以及
    根据所确定的透明系数对所述三维数据进行渲染，以得到所述第一器官区域和所述第二器官区域的渲染图像，其中，在所述渲染图像中，所述第一器官区域的不同部分呈现不同的透明度。
  14. 根据权利要求13所述的超声成像方法，其特征在于，还包括：
    基于所述定位检测结果，确定所述第一器官区域在用户视角下的前侧部分，并且其中，
    确定所述第一器官区域的不同部分对应的透明系数进一步包括：分别确定所述前侧部分和所述第一器官区域的其它部分的透明系数，并且，
    在所述渲染图像中，所述第一器官区域的所述前侧部分呈现多种不同的透明度，所述其它部分呈现相同的透明度或不透明。
  15. 根据权利要求13所述的方法，其特征在于，还包括：
    在所述第一器官区域中确定待透明渲染的目标部分，并且其中，
    确定所述第一器官区域的不同部分对应的透明系数进一步包括：分别确定所述目标部分和所述第一器官区域的其它部分的透明系数，并且，
    在所述渲染图像中，所述第一器官区域的所述目标部分呈现多种不同的透明度，所述其它部分呈现相同的透明度或不透明。
  16. 根据权利要求13所述的方法，其特征在于，还包括：
    确定所述第一器官区域和所述第二器官区域的渲染颜色；以及
    根据所确定的渲染颜色对所述三维数据进行渲染。
  17. 根据权利要13所述的方法，其特征在于，对所述三维数据进行渲染的渲染模式为面绘制，并且，基于所述定位检测结果，确定所述第一器官区域的不同部分对应的透明系数进一步包括：
    基于所述定位检测结果，从所述三维数据中提取所述第一器官区域的模型顶点坐标信息，所述模型顶点坐标信息用于形成构建所述第一器官区域的网格模型的多个三角面片；以及
    对于所述多个三角面片中的至少部分三角面片，执行以下步骤：
    根据所述三角面片的三个顶点的坐标信息计算所述三角面片的法向量；
    计算所述三角面片的法向量与预设参考方向的夹角；以及
    根据所述夹角的大小确定所述三角面片对应的透明系数。
  18. 根据权利要求13所述的方法，其特征在于，对所述三维数据进行渲染的渲染模式为面绘制，并且，基于所述定位检测结果，确定所述第一 器官区域的不同部分对应的透明系数进一步包括：
    基于所述定位检测结果，从所述三维数据中提取所述第一器官区域的模型顶点坐标信息和模型重心坐标信息，所模型顶点坐标信息用于形成构建所述第一器官区域的网格模型的多个三角面片；以及
    对于所述多个三角面片中的至少部分三角面片，执行以下步骤：
    根据所述三角面片的三个顶点的坐标信息计算所述三角面片的三个顶点与模型重心之间的方向向量；
    计算所述方向向量与预设参考方向之间的夹角；以及
    根据所述夹角的大小确定所述三角面片对应的透明系数。
  19. 根据权利要求13所述的方法，其特征在于，对所述三维数据进行渲染的渲染模式为面绘制，并且，基于所述定位检测结果，确定所述第一器官区域的不同部分对应的透明系数进一步包括：
    基于所述定位检测结果，从所述三维数据中提取所述第一器官区域的模型顶点坐标信息，所述模型顶点坐标信息用于形成构建所述第一器官区域的网格模型的多个三角面片；以及
    对于所述多个三角面片中的至少部分三角面片，执行以下步骤：
    根据所述三角面片的三个顶点的坐标信息计算所述三角面片的所述三个顶点与预设参考方向的夹角；以及
    根据所述夹角的大小确定所述三角面片对应的透明系数。
  20. 根据权利要求13所述的方法，其特征在于，对所述三维数据进行渲染的渲染模式为体绘制，并且，基于所述定位检测结果，确定所述第一器官区域的不同部分对应的透明系数进一步包括：
    基于所述定位检测结果，确定所述第一器官区域的边界轮廓；
    根据体绘制对应的算法确定所述第一器官区域的边界轮廓内多个像素中至少部分像素对应的透明系数。
  21. 根据权利要求13所述的方法，其特征在于，对所述三维数据进行渲染的渲染模式为体绘制，并且，基于所述定位检测结果，确定所述第一 器官区域的不同部分对应的透明系数进一步包括：
    基于所述定位检测结果，确定所述第一器官区域的边界轮廓；
    从所述三维数据中提取所述第一器官区域的边界轮廓内多个像素中至少部分像素对应的灰度值；以及
    根据灰度值与透明系数之间的预设映射关系，确定所述至少部分像素对应的透明系数。
  22. 根据权利要求13所述的方法，其特征在于，在所述三维数据中对所述第一器官和所述第二器官进行定位检测进一步包括：
    根据目标检测算法，在所述三维数据中对所述第一器官区域和所述第二器官区域进行定位检测；或者
    根据目标分割算法，在所述三维数据中对所述第一器官区域和所述第二器官区域进行定位检测；或者
    根据所述目标检测算法，在所述三维数据中对所述第一器官区域和所述第二器官区域中的一个区域进行定位检测，并根据所述目标分割算法，在所述三维数据中对所述第一器官区域和所述第二器官区域中的另一个区域进行定位检测。
  23. 根据权利要求13所述的方法，其特征在于，获取被测对象的第一器官和第二器官的待渲染的三维数据进一步包括：
    获取所述被测对象的所述第一器官和所述第二器官的四维超声图像；以及
    根据目标识别算法、目标检测算法和目标分割算法中的一个，从所述四维超声图像中选取一个时刻的三维超声图像作为所述待渲染的三维数据。
  24. 根据权利要求13-23中任一项所述的超声成像方法，其特征在于，
    所述第一器官为卵巢，所述第二器官为卵泡；或者
    所述第一器官为宫体，所述第二器官为子宫内膜；或者
    所述第一器官为胎儿颅脑，所述第二器官为颅脑内部结构；或者
    所述第一器官为胎儿腹部，所述第二器官为腹部内部结构；或者
    所述第一器官为肝脏，所述第二器官为肝内血管；或者
    所述第一器官为心脏，所述第二器官为心脏内部结构。
  25. 一种超声设备，其特征在于，包括：
    超声探头；
    发射和接收电路，其被配置为控制所述超声探头向被测对象的卵巢组织发射超声波，并控制所述超声探头接收所述超声波的回波信号；
    存储器，其用于存储计算机可执行指令；
    处理器，其被配置为在执行所述计算机可执行指令时，根据所述回波信号获得三维数据，并执行根据权利要求1-12中任一项所述的超声成像方法以生成渲染图像；以及
    显示器，其被配置为显示所述渲染图像。
  26. 一种超声设备，其特征在于，包括：
    超声探头；
    发射和接收电路，其被配置为控制所述超声探头向被测对象的目标组织发射超声波，并控制所述超声探头接收所述超声波的回波信号；
    存储器，其用于存储计算机可执行指令；
    处理器，其被配置为在执行所述计算机可执行指令时，根据所述回波信号获得三维数据，并执行根据权利要求13-24中任一项所述的超声成像方法以生成渲染图像；以及
    显示器，其被配置为显示所述渲染图像。
PCT/CN2023/127760 2022-10-31 2023-10-30 一种超声成像方法及超声设备 WO2024093911A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211352167.5A CN115619941A (zh) 2022-10-31 2022-10-31 一种超声成像方法及超声设备
CN202211352167.5 2022-10-31

Publications (1)

Publication Number Publication Date
WO2024093911A1 true WO2024093911A1 (zh) 2024-05-10

Family

ID=84876248

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/127760 WO2024093911A1 (zh) 2022-10-31 2023-10-30 一种超声成像方法及超声设备

Country Status (2)

Country Link
CN (1) CN115619941A (zh)
WO (1) WO2024093911A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115619941A (zh) * 2022-10-31 2023-01-17 深圳迈瑞生物医疗电子股份有限公司 一种超声成像方法及超声设备

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140343404A1 (en) * 2013-03-14 2014-11-20 Inneroptic Technology, Inc. Medical device guidance
US20190307417A1 (en) * 2016-10-28 2019-10-10 Samsung Electronics Co., Ltd. Method and apparatus for follicular quantification in 3d ultrasound images
CN114617581A (zh) * 2020-12-08 2022-06-14 深圳迈瑞生物医疗电子股份有限公司 一种早孕期胎儿的超声成像方法和系统
CN114723893A (zh) * 2022-04-26 2022-07-08 广州柏视医疗科技有限公司 一种基于医疗影像的器官组织空间关系渲染方法及系统
CN115619941A (zh) * 2022-10-31 2023-01-17 深圳迈瑞生物医疗电子股份有限公司 一种超声成像方法及超声设备

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140343404A1 (en) * 2013-03-14 2014-11-20 Inneroptic Technology, Inc. Medical device guidance
US20190307417A1 (en) * 2016-10-28 2019-10-10 Samsung Electronics Co., Ltd. Method and apparatus for follicular quantification in 3d ultrasound images
CN114617581A (zh) * 2020-12-08 2022-06-14 深圳迈瑞生物医疗电子股份有限公司 一种早孕期胎儿的超声成像方法和系统
CN114723893A (zh) * 2022-04-26 2022-07-08 广州柏视医疗科技有限公司 一种基于医疗影像的器官组织空间关系渲染方法及系统
CN115619941A (zh) * 2022-10-31 2023-01-17 深圳迈瑞生物医疗电子股份有限公司 一种超声成像方法及超声设备

Also Published As

Publication number Publication date
CN115619941A (zh) 2023-01-17

Similar Documents

Publication Publication Date Title
CN110325119B (zh) 卵巢卵泡计数和大小确定
JP6629094B2 (ja) 超音波診断装置、医用画像処理装置及び医用画像処理プログラム
CN110753517A (zh) 基于概率映射的超声扫描
JP5226978B2 (ja) 超音波診断装置及び画像処理プログラム
KR102539901B1 (ko) 2차원 초음파 이미지를 음영화하는 방법 및 시스템
CN111368586B (zh) 超声成像方法及系统
WO2024093911A1 (zh) 一种超声成像方法及超声设备
JP2012155723A (ja) 三次元医療映像から最適の二次元医療映像を自動的に生成する方法及び装置
US20140031691A1 (en) Ultrasound diagnostic device
US20230210501A1 (en) Ultrasound contrast imaging method and device and storage medium
US11696740B2 (en) Method and system for defining cut lines to generate a 3D fetal representation
WO2022099705A1 (zh) 早孕期胎儿的超声成像方法和超声成像系统
EP3579757B1 (en) Ultrasound evaluation of anatomical features
WO2022134049A1 (zh) 胎儿颅骨的超声成像方法和超声成像系统
WO2020133236A1 (zh) 一种脊柱的成像方法以及超声成像系统
CN113229850A (zh) 超声盆底成像方法和超声成像系统
JP2018149055A (ja) 超音波画像処理装置
CN116327237A (zh) 超声成像系统及方法、超声图像处理系统及方法
CN111403007A (zh) 一种超声成像的优化方法、超声成像系统和计算机可读存储介质
JP2018153561A (ja) 超音波画像処理装置
CN113974688B (zh) 超声成像方法和超声成像系统
CN117982169A (zh) 一种子宫内膜厚度的确定方法及超声设备
CN116115267A (zh) 超声成像方法、超声成像系统和存储介质
CN116091516A (zh) 医学图像配准方法、医学影像系统及超声成像系统
CN115778435A (zh) 胎儿颜面部的超声成像方法和超声成像系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23884841

Country of ref document: EP

Kind code of ref document: A1