WO2016176863A1 - 三维超声成像方法和装置 - Google Patents

三维超声成像方法和装置 Download PDF

Info

Publication number
WO2016176863A1
WO2016176863A1 PCT/CN2015/078494 CN2015078494W WO2016176863A1 WO 2016176863 A1 WO2016176863 A1 WO 2016176863A1 CN 2015078494 W CN2015078494 W CN 2015078494W WO 2016176863 A1 WO2016176863 A1 WO 2016176863A1
Authority
WO
WIPO (PCT)
Prior art keywords
median sagittal
sagittal section
region
orientation
fetal head
Prior art date
Application number
PCT/CN2015/078494
Other languages
English (en)
French (fr)
Inventor
邹耀贤
林穆清
陈志杰
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
Priority to PCT/CN2015/078494 priority Critical patent/WO2016176863A1/zh
Priority to CN201580071177.3A priority patent/CN107106143B/zh
Priority to CN202010990346.6A priority patent/CN112120736B/zh
Publication of WO2016176863A1 publication Critical patent/WO2016176863A1/zh
Priority to US15/800,387 priority patent/US10702240B2/en
Priority to US16/903,221 priority patent/US11534134B2/en

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/08—Detecting organic movements or changes, e.g. tumours, cysts, swellings
    • A61B8/0866—Detecting organic movements or changes, e.g. tumours, cysts, swellings involving foetal diagnosis; pre-natal or peri-natal diagnosis of the baby
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/08—Detecting organic movements or changes, e.g. tumours, cysts, swellings
    • A61B8/0808—Detecting organic movements or changes, e.g. tumours, cysts, swellings for diagnosis of the brain
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/08—Detecting organic movements or changes, e.g. tumours, cysts, swellings
    • A61B8/0808—Detecting organic movements or changes, e.g. tumours, cysts, swellings for diagnosis of the brain
    • A61B8/0816—Detecting organic movements or changes, e.g. tumours, cysts, swellings for diagnosis of the brain using echo-encephalography
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/13—Tomography
    • A61B8/14—Echo-tomography
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/46—Ultrasonic, sonic or infrasonic diagnostic devices with special arrangements for interfacing with the operator or the patient
    • A61B8/461—Displaying means of special interest
    • A61B8/466—Displaying means of special interest adapted to display 3D data
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/48—Diagnostic techniques
    • A61B8/483—Diagnostic techniques involving the acquisition of a 3D volume of data
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/5215—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data
    • A61B8/5223—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data for extracting a diagnostic or physiological parameter from medical diagnostic data
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/5215—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data
    • A61B8/523—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data for generating planar views from image data in a user selectable plane not corresponding to the acquisition plane
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/44—Constructional features of the ultrasonic, sonic or infrasonic diagnostic device
    • A61B8/4405—Device being mounted on a trolley
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/44—Constructional features of the ultrasonic, sonic or infrasonic diagnostic device
    • A61B8/4427—Device being portable or laptop-like
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/44—Constructional features of the ultrasonic, sonic or infrasonic diagnostic device
    • A61B8/4444—Constructional features of the ultrasonic, sonic or infrasonic diagnostic device related to the probe
    • A61B8/4472—Wireless probes

Definitions

  • the present invention relates to the field of medical ultrasound imaging technology, and in particular, to a three-dimensional ultrasound imaging method and apparatus.
  • Ultrasonic instruments are generally used by doctors to observe the internal tissue structure of the human body.
  • the doctor places the ultrasound probe on the surface of the skin corresponding to the human body part, and an ultrasound image of the part can be obtained.
  • Ultrasound has become one of the main aids for doctors' diagnosis because of its safety, convenience, losslessness and low cost.
  • obstetrics is one of the most widely used fields of ultrasound diagnosis. In this field, ultrasound avoids the influence of X-rays on the mother and fetus, and its application value is obviously superior to other imaging examination equipment.
  • Ultrasound can not only observe and measure fetal morphology, but also obtain a variety of physiological and pathological information such as fetal respiration and urology to evaluate the health and development of the fetus.
  • the corpus callosum and cerebellar vermis are two important examination items.
  • the corpus callosum is the largest commissural fiber in the cerebral hemisphere. It is responsible for communication between the two hemispheres of the brain. Loss or dysplasia will lead to epilepsy and intelligence. A series of complications such as low fever and motor dysfunction. Deletion or dysplasia of the cerebellum is a manifestation of the Dandy-walker syndrome. 50% of Dandy-walker patients have mental retardation and mental retardation, often accompanied by chromosomal abnormalities and other malformations, with poor prognosis and high mortality. It can be seen that the abnormalities of the corpus callosum and the cerebellum are all manifestations of major diseases.
  • the hospital as the main body of examination may also produce Medical Dispute.
  • the corpus callosum and cerebellar vermis are the most easily misdiagnosed and missed items.
  • the reason is that the median sagittal section of the fetus is the best view of the corpus callosum and cerebellar vermis, but due to the fetus.
  • the position, amniocentesis, nasal bone obstruction, doctor's technical level and other factors it is difficult to obtain the median sagittal section of the fetus under conventional two-dimensional ultrasound. Even if it can be obtained, it takes a long time to check. Many doctors can only pass Other aspects (such as cerebellar cuts, thalamic cuts, etc.) for non-intuitive diagnosis, prone to misdiagnosis and missed diagnosis.
  • One of the objects of the present invention is to provide a three-dimensional ultrasound imaging method and apparatus capable of three-dimensional imaging of a fetal head, automatically detecting a median sagittal section of the fetal head, and being able to determine the orientation of the fetal head. According to the judgment result, the median sagittal section is displayed as an image conforming to the observation habit of the person.
  • a three-dimensional ultrasound imaging method comprising: transmitting an ultrasonic wave to a fetal head; receiving an ultrasonic echo to obtain an ultrasonic echo signal; obtaining three-dimensional volume data of the fetal head according to the ultrasonic echo signal; a feature of the section, detecting a median sagittal section from the three-dimensional volume data; detecting a specific tissue in the fetal head in the section of the median sagittal section and/or parallel or intersecting the median sagittal section An image region of the region; determining an orientation orientation of the fetal head in the median sagittal section according to the image region; rotating the median sagittal section according to an orientation orientation of the fetal head, such that the median vector after rotation
  • the fetal head in the section is at a predetermined orientation, or the orientation of the fetal head is identified on the median sagittal section.
  • a three-dimensional ultrasonic imaging apparatus comprising: a probe for transmitting ultrasonic waves to a fetal head and receiving an ultrasonic echo to obtain an ultrasonic echo signal; and a three-dimensional imaging module for The ultrasonic echo signal obtains three-dimensional volume data of the fetal head, and detects a median sagittal section from the three-dimensional volume data according to characteristics of the median sagittal section of the fetal head, and determines a fetal head in the median sagittal section Orienting toward the orientation, and rotating the median sagittal section according to the orientation orientation of the fetal head such that the fetal head in the median sagittal section after rotation is in a predetermined orientation, or is identified on the median sagittal section The orientation of the fetal head; a display for displaying the median sagittal section.
  • the invention has the beneficial effects that: the fetus can be ultrasonically scanned to obtain three-dimensional body data of the fetal head, and the median sagittal section of the fetal brain is automatically detected according to the obtained three-dimensional volume data, and then the median sagittal section is automatically determined.
  • the orientation of the head of the fetus, such as whether the head is inverted or not, and the orientation of the face, according to the judgment result, the image of the median sagittal section is displayed as an image conforming to human observation habits, making it easier for the doctor to distinguish and observe the median sagittal section of the fetal brain. Case.
  • FIG. 1 is a block diagram showing the structure of a three-dimensional ultrasonic imaging apparatus according to Embodiment 1 of the present invention
  • FIG. 2 is a schematic flow chart of a three-dimensional ultrasonic imaging method according to Embodiment 1 of the present invention
  • Embodiment 3 is a schematic diagram of three-dimensional volume data in Embodiment 1 of the present invention.
  • Figure 4 is a schematic view showing the position of a median sagittal section of the fetal head
  • Figure 5 is a schematic illustration of a median sagittal section of the fetal head
  • Figure 6 is a schematic view of the L1 cut surface of Figure 5;
  • Figure 7 is a schematic view of the L2 cut surface of Figure 5;
  • Figure 8 is a schematic view of the fetal head inverted and erect in the median sagittal section
  • FIG. 9 is a flow chart showing the determination of the up and down orientation of the fetal head in the median sagittal section according to an embodiment of the present invention.
  • Figure 10 is a schematic view showing the orientation of the skull
  • FIG. 11 is a flow chart showing the determination of the up and down orientation of a fetal head in a median sagittal section in accordance with another embodiment of the present invention.
  • FIG. 12 is a schematic flow chart of a three-dimensional ultrasonic imaging method according to Embodiment 3 of the present invention.
  • FIG. 13 is a schematic flowchart of a method for determining a face orientation according to an embodiment of the present invention.
  • FIG. 14 is a schematic flow chart of a three-dimensional ultrasonic imaging method according to Embodiment 4 of the present invention.
  • 15 is a flow chart showing the steps of detecting a median sagittal plane according to an embodiment of the present invention.
  • Figure 16 is a schematic diagram of a plane in a three-dimensional space and its plane parameters
  • 17 is a schematic diagram of a three-dimensional Hough matrix in an embodiment of the present invention.
  • FIG. 18 is a schematic diagram of a process of weighted Hough transform in an embodiment of the present invention.
  • FIG. 19 is a schematic diagram of a process of random Hough transform in an embodiment of the present invention.
  • 20 is a schematic diagram of a process of detecting a plane determined by a selected feature point according to an embodiment of the present invention
  • 21 is a schematic flow chart showing the steps of detecting a median sagittal plane according to another embodiment of the present invention.
  • FIG. 22 is a schematic flow chart of extracting a midline of a brain according to an embodiment of the present invention.
  • FIG. 23 is a schematic flow chart of a step of detecting a median sagittal plane according to still another embodiment of the present invention.
  • 24 is a flow chart showing the steps of detecting a median sagittal plane according to still another embodiment of the present invention.
  • 25 and 26 are schematic illustrations of icons for identifying the orientation of a fetal head in accordance with one embodiment of the present invention.
  • the three-dimensional ultrasound imaging apparatus includes a probe 102, a transmit/receive selection switch 103, a transmitting circuit 104, a receiving circuit 105, a beam combining module 106, a signal processing module 107, a three-dimensional imaging module 108, and a display 109.
  • the transmitting circuit 104 sends a set of delayed focus pulses to the probe 102.
  • the probe 102 transmits ultrasonic waves to the body tissue (not shown), and receives the tissue reflected from the body tissue after a certain delay. The ultrasonic echo of the information is reconverted into an electrical signal.
  • the receiving circuit 105 receives the ultrasonic echo signals that have been converted into electrical signals and sends the ultrasonic echo signals to the beam combining module 106.
  • the ultrasonic echo signals are subjected to focus delay, weighting, and channel summation at beam combining module 106 for signal processing by signal processing module 107.
  • the signal processed by the signal processing module 107 is sent to the three-dimensional imaging module 108, processed by the three-dimensional imaging module 108, to obtain visual information such as a three-dimensional image, and then sent to the display 109 for display.
  • the signal processed by the signal processing module 107 forms a volume of three-dimensional volume data in the polar coordinate in the three-dimensional imaging module 108.
  • the three-dimensional volume data in the polar coordinate is reconstructed and the polar coordinates are obtained.
  • the volume data is converted into Cartesian coordinate data to obtain a volume of three-dimensional volume data in Cartesian coordinates.
  • the three-dimensional imaging module 108 then calculates the three-dimensional volume data in the Cartesian coordinate system to obtain visual information and display it on the display device 109.
  • the three-dimensional imaging module 108 further includes a sub-module for automatically detecting and processing the median sagittal section of the fetus, the sub-module being capable of automatically detecting the three-dimensional volume data of the fetal head obtained therefrom.
  • the median sagittal section of the fetus is treated to detect the orientation of the fetal head in the median sagittal section (eg, the face of the fetus and/or the head of the fetus is facing up, to the left, to the upper right, toward Left, right, down, down, down, or toward other directions, etc.), then rotate the median sagittal section such that the fetal head in the median sagittal section is in a predetermined orientation after rotation (eg, overhead Upward or towards Down or toward any other desired orientation, face up or down or toward any other desired orientation, etc., for example to facilitate the doctor's observation or to conform to the doctor's habits, etc.) and/or in the median sagittal
  • the orientation of the detected fetal head is identified in the section and the processed median sagittal section (described in detail below) is displayed.
  • the present embodiment provides a method for three-dimensional ultrasound imaging, the flow of which is shown in FIG. 2.
  • step 21 the three-dimensional ultrasound imaging device is first used to perform three-dimensional scanning on the fetal head, ultrasonic waves are transmitted to the fetal head and ultrasonic echoes are received, ultrasonic echo signals are obtained, and the ultrasonic echo signals are processed as described above, thereby Three-dimensional volume data of the fetal head (hereinafter referred to simply as "three-dimensional volume data") is obtained.
  • the specific steps of three-dimensional scanning of the scanning target and processing of the ultrasonic echo signals to obtain three-dimensional volume data may be the same as or similar to those of the three-dimensional scanning and imaging methods commonly used in the art, and will not be described in detail herein.
  • at least one volume of three-dimensional volume data of the fetal head can be obtained.
  • a volume of three-dimensional volume data can be as shown in FIG.
  • the volume data may be composed of image frames of F frame size W ⁇ H, where W is the width of the image frame and H is the height of the image frame.
  • W is the width of the image frame
  • H is the height of the image frame.
  • the width direction of the image frame is defined as the X direction
  • the height direction of the image frame is defined as the Y direction
  • the direction in which the multi-frame image frames are arranged is defined as the Z direction.
  • the X, Y, and Z directions can also be defined in different ways.
  • step 21 After obtaining the three-dimensional volume data in step 21, in the method of the embodiment of the present invention, it is desirable to be able to automatically detect the median sagittal section of the fetal head from the three-dimensional volume data.
  • FIG. 4 The position of the median sagittal section of the fetal head is shown in Figure 4, and line D in Figure 4 represents the position of the median sagittal section of the fetal brain.
  • a schematic representation of the median sagittal section of the fetal head is shown in FIG. It can be seen that in this median sagittal section, important information about the corpus callosum, cerebellar vermis, and transparent compartment of the fetus is included. In addition, the cisterna magna of the fetus can be observed from the median sagittal section of the fetal head. Thalamic adhesion, fourth ventricle and other structures.
  • the automatic detection of the median sagittal section of the fetal brain can provide a large amount of important information for the physician, greatly facilitating the physician's observation of the fetal condition.
  • 6 and 7 schematically illustrate cut planes L1 and L2 of the fetal head perpendicular to the median sagittal section, respectively.
  • the median sagittal section has some special features in the three-dimensional image of the fetal head. For example, in all the sections in the three-dimensional image of the fetal head, the median sagittal section has a larger overall area than the surrounding area. A gray value with a larger gray value, that is, in a three-dimensional image of the fetal head, the median sagittal section exhibits a gray value that is significantly larger than the vicinity thereof.
  • the section of the gray value, or the median sagittal section appears in the three-dimensional image of the fetal head as a brighter section than the surrounding area; or, the structure on both sides of the median sagittal section in the fetal head is approximate Symmetrical, so in the three-dimensional image of the fetal head, the image data on either side of the median sagittal section will exhibit approximate symmetry; or, in the fetal head, the median sagittal section is in the middle of the head, and
  • the other intersecting surface intersecting the median sagittal section includes information at the intersection of the section and the median sagittal section, and in the images of the other sections, the section and the median sagittal section The intersection line appears as a brighter line, the midline of the brain, and the collection of these midbrain lines constitutes the median sagittal section; In some embodiments of the invention, these features of the median sagittal section of the fetal head are utilized to detect
  • step 23 of the present embodiment detects from the three-dimensional volume data obtained in step 21 based on the features of the median sagittal section of the fetal head (for example, features found in the foregoing studies, such as grayscale features). A median sagittal section in the three-dimensional volume data.
  • the foregoing “detecting the median sagittal section in the three-dimensional volume data from the three-dimensional volume data obtained in step 21” may be detected in the three-dimensional volume data of all fetal heads, or may be Detection in a portion of the three-dimensional volume data of the fetal head, for example, may be detected in the region where the median sagittal section is most likely to be present, while removing the region in which the median sagittal section is clearly unlikely to be present.
  • the median sagittal section of the fetal head is a longitudinal section located in the middle of the fetal head (ie, in the direction from the head portion to the neck portion, it is clearly impossible in some areas at the edge of the head) There is a median sagittal section, and such an area can be excluded from the detection range.
  • This embodiment can detect the median sagittal section in the three-dimensional volume data using a variety of methods. For example, as described above, in the three-dimensional volume data, the median sagittal section exhibits a feature that the gray value in the median sagittal section is larger than the gray value of the surrounding area. Therefore, a specific implementation of the embodiment may be By using this feature of the median sagittal section, a median sagittal section is detected from the three-dimensional volume data using, for example, an image segmentation algorithm in a digital image processing method.
  • the automatic detection result of the sagittal section is essentially marking the position of the sagittal section in the three-dimensional volume data coordinate system, but the expression can be various, such as a plane equation and a sagittal section relative to Coordinate system origin translation (translation in X, Y, Z directions) and rotation (rotation around X, Y, Z axes), transformation matrix of sagittal section relative to the original coordinate system (usually a 4 ⁇ 4 matrix) It can represent the transformation relationship of two coordinate systems), even the coordinates of three points in space (three points determine a plane) and so on.
  • the median sagittal section image may be inverted (ie, the fetal head is facing the image below, as shown in the left diagram of Figure 8) or other directions that are not convenient for the doctor to observe; and it is difficult to obtain under two-dimensional ultrasound.
  • the orientation of the fetal head in the median sagittal section is also (for example, the fetal head and/or the fetal face are facing up, Judging to the left, to the right, to the left, to the right, to the lower, to the lower left, to the lower right, or to the other direction, etc.) to make it easier to detect the orientation of the fetal head on the median sagittal section
  • the median sagittal section is rotated such that the fetal head in the median sagittal section is in a predetermined or desired orientation after rotation (eg, the head is up or down or toward any other desired orientation, or the face is facing up) Or pointing downwards or towards any other desired orientation, for example to facilitate the doctor's observation or to conform to the doctor's habits, etc.) and/or to identify the orientation orientation of the detected fetal head in the median sagittal section, on
  • the orientation of the fetal head in the median sagittal section may be represented in the fetal head by a median sagittal section and/or a section of the three-dimensional volume data that is parallel or intersects the median sagittal section The image area of a particular tissue area is obtained.
  • the specific tissue regions referred to herein may also be at least two specific tissue regions (eg, eyes and nose, eyes and mouth, mouth, and nose) having a particular mutual positional relationship in the fetal head. Department, or eye, nose and / or mouth and transparent compartment, or eye, Nasal and / or mouth and other parts of the fetal head, etc.).
  • specific tissue regions eg, eyes and nose, eyes and mouth, mouth, and nose
  • At least two image regions representing at least two of the tissue regions may be extracted or detected from a median sagittal section and/or a section parallel or intersecting the median sagittal section, and
  • the mutual positional relationship between these image regions determines the orientation of the fetal head (eg, in the fetal head, the top of the head is always in the direction from the mouth to the eye, from the mouth to the nose, or from the nose to the eye) On, etc.).
  • the mutual positional relationship between these image regions is determined, and the orientation orientation of the fetal head can be determined based on these mutual positional relationships.
  • the particular tissue region referred to herein may be a tissue region having directional characteristics in the fetal head.
  • a tissue region having directional characteristics refers to a tissue region that itself or its location contains information indicative of the orientation of the fetal head, such as a skull or cranium (the direction of which indicates the orientation of the fetal head). Azimuth); a transparent compartment (the orientation and position of which can indicate the orientation of the fetal head); the mouth, eyes and nose (which is always on the face side of the fetal head, so its position can indicate the fetal head Orientation); and so on.
  • image regions representing one or more of the tissue regions may be extracted or detected from the median sagittal section and/or the plane parallel or intersecting the median sagittal section, according to the image regions
  • the position and/or shape feature determines the orientation of the fetal head (eg, the position or side of the eye and mouth is always the front or anterior side of the fetal head, and the head can be determined by the direction of curvature of the skull Orientation, etc.).
  • the specific tissue region is described as an example of a skull and a transparent compartment, respectively.
  • the orientation of the fetal head in the median sagittal section can be detected or identified using a structure having directional characteristics in the triad data.
  • the inventors have found through research that in the three-dimensional image of the fetal head, the skull exhibits a distinct highlight echo. Therefore, the orientation of the fetal head can be judged based on the orientation of the skull. Therefore, in the present embodiment, the orientation of the fetal head is determined by detecting the orientation of the mid-sagittal section image or the skull in the image parallel to the sagittal section, and the detection flow is as shown in FIG.
  • Step 252 Extracting a skull feature from the selected slice to obtain a candidate region for characterizing the skull. Since the skull is highlighted in the ultrasound and the echoes on both sides of the skull are slowly weakened, a variety of methods for extracting the skull features can be designed according to this feature. For example, in the present embodiment, a method for extracting a skull feature is to select a region in which a grayscale is greater than a preset grayscale threshold as a candidate region of the skull according to the characteristics of the highlighted echo of the skull, wherein the selected region is selected.
  • the cut surface refers to a median sagittal section and/or at least one section parallel to the median sagittal section, the preset
  • the grayscale threshold may be determined according to actual needs, that is, may be an empirical threshold, or may be determined according to statistical characteristics of the image, for example, the threshold may be set to an average grayscale* empirical coefficient.
  • the extraction of the skull feature may be performed by utilizing the nature of the dark side of the skull, and the operator is designed based on the differential method, and the image corresponding to the selected slice is convolved and the volume is retained. A portion whose integrated value is greater than a preset grayscale threshold is used as the feature image.
  • the operator designed based on the differential method may be one or more of the feature extraction operators including, for example, the following (1) to (5).
  • the rule may be defined as: selecting one or more feature regions having the largest sum of feature values as candidate regions.
  • the feature value may be a feature used in the foregoing extraction of the skull feature, such as a grayscale feature.
  • the feature value may also be a feature involved in feature extraction commonly used in digital image processing, such as a texture feature.
  • the rule may be defined as selecting one or more feature regions with the largest average feature value as the candidate region.
  • the rule may be defined as: adopting a machine learning method, that is, extracting features from the feature region, inputting the extracted features into a pre-trained classifier for classification, and classifying the result as a candidate region;
  • the feature extracted here may be a feature
  • the pre-trained classifier may be to extract the above features through a certain number of samples in advance and adopt PCA ( Principal Component Analysis, Principal Component Analysis, KPCA (Kernel Principal Component Analysis), ICA (Independent Component Analysis), LDA (Linear Discriminant Analysis), SVM (Support Vector Machine) Any of the classifiers, such as support vector machines, can be trained.
  • PCA Principal Component Analysis, Principal Component Analysis, KPCA (Kernel Principal Component Analysis), ICA (Independent Component Analysis), LDA (Linear Discriminant Analysis), SVM
  • the specific implementation can refer to the prior art related to image processing and pattern recognition, and will not be described in detail herein. It can be understood that, for the case where the number of selected cut planes is greater than 1 (that is, multiple cut planes are selected), the definition of the rules is similar, for example, in all selected cut planes, the sum of the feature values or the feature average is the largest.
  • One or more feature regions serve as candidate regions for the skull, or the candidate regions are identified in a machine learning manner.
  • the orientation of the skull can be determined by the direction of curvature of the skull.
  • the quadratic curve is used to fit the connected region, and the orientation of the skull is determined according to the coefficient of the quadratic curve quadratic term. For example, if the fitted quadratic coefficient is greater than 0, then the The skull is facing down, and vice versa, the skull is facing up.
  • the method of machine learning such as PCA, KPCA, LDA, SVM, etc., is obtained through learning.
  • the specific implementation may refer to the prior art related to image processing and pattern recognition, which will not be described in detail herein.
  • the orientation of each of the communication areas can be separately judged and then voted to determine the orientation of the final skull.
  • the image of the median sagittal section is rotated by 180° or the image of the median sagittal section is turned upside down and then displayed, so that the displayed image is more consistent.
  • Human observation habits help doctors observe fetal conditions.
  • a plane equation representation method is uniformly employed.
  • the present invention is not limited to the representation of the plane equation, but also includes The foregoing or other representations in the art. Any expression of the sagittal section test result is merely a difference in expression form, and does not affect the essence of the present invention, and is within the scope of protection of the present invention.
  • the three-dimensional ultrasound imaging system implementing the foregoing method is not limited to a general ultrasound imaging system (for example, a trolley-type ultrasound imaging system or a portable ultrasound imaging system) integrated as a whole device, or may be a distributed system.
  • a data communication device wireless or wirelessly
  • the steps or functions of the foregoing methods may be implemented on other devices connected to a conventional trolley-type ultrasound imaging system or portable ultrasound imaging system by a data communication device (wired or wirelessly), which may be, for example, Data processing workstations, personal computers, various smart portable devices, other ultrasound imaging devices, various network servers, and the like, such that at least some of the steps or functions are integrally formed with the trolley-type ultrasound imaging system or the portable ultrasound imaging system in this embodiment.
  • a three-dimensional ultrasound imaging system may be, for example, Data processing workstations, personal computers, various smart portable devices, other ultrasound imaging devices, various network servers, and the like, such that at least some of the steps or functions are integrally formed with the trolley-type ultrasound imaging system or the portable ultrasound imaging system in
  • the fetal can be ultrasonically scanned to obtain three-dimensional volume data of the fetal head, and according to the obtained three-dimensional volume data, the median sagittal section of the fetal brain is automatically detected and the orientation of the fetal head is determined.
  • the orientation (for example, judging whether the fetus is upside down) is displayed after rotating the inverted body to the erect position, so that the display result conforms to the observation habit of the person, and solves the problem that the doctor manually cannot accurately locate the median sagittal section, so that the doctor can Convenient observation of the median sagittal section of the fetal brain can provide physicians with a wealth of important and critical information.
  • the three-dimensional ultrasound imaging method/system provided in this embodiment is similar to Embodiment 1, and the same portions are not described herein again, except that in step 25 of Embodiment 1, the fetal position in the median sagittal section is determined by the orientation of the skull.
  • the head is oriented; and step 25 of the present embodiment determines the head orientation of the fetus in the median sagittal section based on the orientation of the transparent compartment.
  • the shape of the transparent compartment is a crescent-shaped crescent shape.
  • the transparent compartment exhibits an upward convex shape, and vice versa.
  • the transparent compartment appears as a downward convex crescent shape. Therefore, the head orientation of the fetus in the median sagittal section can be judged based on the orientation of the transparent compartment.
  • the process of determining the orientation of the fetal head in the median sagittal section by the orientation of the transparent compartment includes steps 252' and 254', as shown in FIG.
  • Step 252' detecting the communication region corresponding to the transparent compartment from the median sagittal section according to the characteristics of the transparent compartment in the ultrasound image.
  • step 252' since the transparent compartment appears as a black hypoechoic region in the ultrasound image, and the surrounding tissue is generally brighter, a plurality of methods can be used to segment the hypoechoic region according to this feature.
  • an image segmentation algorithm may be used for segmentation, such as threshold segmentation, Snake algorithm, level-set algorithm, graph-cut segmentation, etc., to segment regions that may be considered as transparent cells.
  • the number of regions obtained by segmentation may be multiple. Therefore, it is possible to select a region that is most like a transparent compartment by setting certain criteria. For example, it may be judged according to the shape, grayscale, variance, or the like of the region, or a combination of these features. Thereby, a communication area corresponding to the transparent compartment is obtained.
  • Step 254' determining the orientation of the transparent compartment based on the communication area and determining the head orientation of the fetus in the median sagittal section according to the judgment result.
  • step 254' for the communication area corresponding to the transparent compartment, the method of determining the orientation of the skull similar to that of Embodiment 1 can be used to determine the orientation of the transparent compartment.
  • the method similar to steps 254a1 to 254a3 is used to judge the orientation of the transparent compartment, that is, first, the joint region corresponding to the transparent compartment is subjected to mathematical morphology processing to extract the skeleton, obtain a skeleton image, and then search for one of the skeleton images.
  • a long and continuous curve which is a representative curve selected, and then judged according to the coordinates of at least one point in the middle position on the curve and the coordinates of at least one point located at both ends to determine the transparent partition
  • the orientation of the cavity is used to judge the orientation of the transparent compartment.
  • the method of steps 254b1 to 254b2 is used to determine the orientation of the transparent compartment, that is, first calculate the center line of the vertical direction of the communication section corresponding to the transparent compartment; then, the orientation of the transparent compartment is determined by the same method as step 254a3. .
  • the determined orientation of the transparent compartment it is obvious that if the transparent compartment is convex upward, the head of the fetus in the median sagittal section faces upward, and if the plane of the transparent compartment is convex downward, the sagittal shape is indicated. The head of the fetus is facing down. After determining the head orientation of the fetus, if the head is facing down, the image of the median sagittal section is rotated by 180° or the image of the median sagittal section is turned upside down and then displayed, so that the displayed image is more consistent. Human observation habits help doctors observe the condition of the fetus.
  • the three-dimensional ultrasound imaging method provided in this embodiment is as shown in FIG. 12, and includes: obtaining a fetal head Step 21 of the three-dimensional volume data of the part, step 23 of detecting the median sagittal section, step 27 of determining the orientation of the face, and step 29 of displaying the median sagittal section. Steps 21, 23, and 29 are the same as those of Embodiment 1 or 2, and are not described herein again.
  • the present embodiment further provides a three-dimensional ultrasonic imaging apparatus for implementing the method.
  • the structure of the apparatus other than the three-dimensional ultrasound imaging part can be referred to the foregoing embodiment 1, and will not be repeated here.
  • the transparent compartment is located in front of the skull, the face and back of the fetus (ie, the face of the fetus) can be determined by the position of the transparent compartment.
  • step 27 is as shown in FIG. 13, and includes a step 271 of detecting a transparent compartment, a step 271 of locating the intracranial center position, and a step 273 of determining the position of the transparent compartment.
  • the detection method of the transparent compartment is similar to the transparent compartment detection method of the foregoing embodiment 2, that is, the image of the median sagittal section is processed by an image segmentation algorithm to obtain at least one region, and the feature of the region is combined with the region.
  • the selection of the region most similar to the transparent compartment is selected as the transparent compartment region.
  • the intracranial center can be positioned according to the fetal skull, and the fetal skull appears in the volume data as an approximate ellipsoid or elliptical shape, and appears as a highlight echo on the ultrasound image. Therefore, in the embodiment, the positioning process of the intracranial center includes the following steps 272a1 to 272a3.
  • Step 272a1 Select at least one frame A-plane image from the volume data, that is, a horizontal cross-sectional view obtained by A-A of FIG.
  • Step 272a2 Design an operator based on a differential method, such as one or more of the aforementioned operators (1) to (5), and perform feature extraction on the A-plane image using the designed operator.
  • Step 272a3 Perform ellipsoid or ellipse detection on the extracted features, and the detection algorithm may adopt an ellipse detection algorithm involved in any known related art, such as a least squares estimation algorithm, a Hough transform algorithm, a random Hough transform algorithm, and the like. For example, it can be detected using the description of the Hough transform algorithm referred to hereinafter.
  • the center of the ellipse or ellipsoid can be determined from the relevant geometric knowledge.
  • step 273 the coordinate information of the transparent compartment area is compared with the coordinate information of the center, and the face orientation of the fetal head in the median sagittal section is obtained according to the comparison result, specifically, a certain point on the transparent compartment area is selected.
  • the position (the value of x on the X coordinate) is the position of the transparent compartment, such as the center position or other position of the transparent compartment area detected in step 271, assuming that the center of the ellipse or ellipsoid obtained in step 272 is The value on the X coordinate is x center , comparing x and x center .
  • x ⁇ x center it means that in the median sagittal section, the transparent compartment is located to the left of the center of the brain, that is, the left side of the median sagittal section is the fetus.
  • the front part (face), the right side is the back of the fetus (back), otherwise, the left side of the median sagittal section is the back of the fetus (back), and the right side is the front of the fetus (face).
  • the anterior and/or posterior portion of the fetus may be marked on the image of the median sagittal section, As shown in the right figure of Figure 8.
  • the orientation of the front/back or the top of the head or the orientation of the face of the fetus may be represented by the icon of a fetal avatar as shown in FIGS. 25 and 26.
  • the orientation of the face in the icon may be current and current.
  • the faces in the data are oriented in the same direction.
  • the median sagittal section labeled with the orientation of the fetal head (eg, the anterior and/or posterior portion of the fetus, etc.) can be displayed on the display for display to facilitate physicians' observation of the fetal condition. .
  • FIG. 14 is a schematic flow chart of the three-dimensional ultrasonic imaging method of the present embodiment, wherein each step is the same as the corresponding steps in the foregoing embodiments, and will not be repeated herein. It should be understood that the order of steps 25 and 27 in this embodiment may also be reversed, that is, the orientation of the face may be determined first, and then the orientation of the head may be determined. Obviously, the present embodiment provides a judgment of the head orientation of the fetus and the orientation of the face, so that the physician can more easily distinguish the tissue structure on the sagittal section.
  • the process of detecting the median sagittal section with respect to step 23 can be as follows.
  • the flow of detecting the median sagittal plane according to the three-dimensional volume data is as shown in FIG. 15, and includes: a step 80 of feature extraction, a step 81 of selecting a feature point, and a plane for detecting the selected feature point. Step 82.
  • the median sagittal section exhibits a feature that the gray value (e.g., gray value, etc.) in the median sagittal section is larger than the gray value of the surrounding area. Based on this, this embodiment utilizes this feature of the median sagittal section to detect the median sagittal plane from the three-dimensional volume data. Therefore, at step 80, a sagittal feature region representing a plane satisfying a condition that the gray value in the plane is larger than the gray value on both sides of the plane may be first extracted in the three-dimensional volume data.
  • the gray value e.g., gray value, etc.
  • some feature regions are extracted in the three-dimensional volume data, and in the method of the embodiment, the required feature regions represent that the grayscale values in the plane satisfying the plane in the three-dimensional volume data are larger than the sides of the plane.
  • the area of the plane of the condition of the gray value, the extracted feature area is the sagittal feature area required by the method of the present embodiment.
  • such sagittal feature regions may be extracted from the three dimensional volume data using a variety of suitable methods.
  • the feature extraction operator may be convolved with the three-dimensional volume data to obtain a convolved image, and the convolved image includes the extracted sagittal feature region.
  • each image frame constituting the three-dimensional volume data may be separately convoluted by the two-dimensional feature extraction operator, and then convolved.
  • the image frames are combined into three-dimensional volume data after convolution; or, the three-dimensional feature extraction operator can be directly designed, and the three-dimensional feature extraction operator is directly used for convolution with the three-dimensional volume data.
  • the specific steps of the convolution operation are well known in the art and will not be described in detail herein.
  • the feature extraction operator can be designed according to the image features that need to be extracted. For example, in the embodiment described above, it is necessary to extract a sagittal feature region in which the grayscale value in the region is larger than the grayscale values on both sides of the region. At this time, one or more of the feature extraction operators (1) to (5) of the foregoing first embodiment can be used.
  • the feature extraction operators obtained by the above-described feature extraction operators (1) to (5) after transposition (matrix transposition), rotation, or the like may be used, or may be used.
  • Other suitable feature extraction operators such as the Roberts operator, the Laplacian Gaussian operator and its variants, and so on.
  • the three-dimensional feature extraction operator can also be directly designed, which will not be described in detail herein.
  • the size of the feature extraction operator (two-dimensional or three-dimensional) can be set as needed.
  • step 81 feature points whose values satisfy a specific condition may be selected from the extracted sagittal feature regions, and generally, at least three feature points are selected.
  • the feature point parameters of the selected feature points are recorded, and the feature point parameters of these feature points will be used in the subsequent steps.
  • the feature point parameter may include coordinates and/or feature points of the feature point.
  • Value for example, gray value or result value after convolution, etc.
  • the aforementioned specific conditions may be determined based on the nature of the feature extraction operator employed. For example, if the foregoing feature extraction operators (1) to (5) are used, the foregoing specific condition may be set to a point where the value of the convolution result is greater than a certain threshold, and the threshold may be an empirical parameter, which may be determined according to actual needs. .
  • the head in order to reduce the pressure of the subsequent plane detecting step (described in detail below) and minimize the influence of noise, it is possible to remove some points which are obviously impossible in the head according to a certain prior knowledge.
  • the head is generally located in the middle of the three-dimensional volume data. Therefore, it is possible to select only the point in the sphere or the ellipsoid with the center of the three-dimensional volume data as the center of the sphere and the radius of a certain threshold as the feature point.
  • the threshold here can also be determined based on experience or actual conditions.
  • the selected feature points can generally determine a plane.
  • the plane is considered to be the plane where the median sagittal section is located, and the section of the three-dimensional volume data that coincides with the plane is The median sagittal section of the fetal brain.
  • the planes determined by the selected feature points are detected, and the plane in which the median sagittal section of the fetal brain is located is determined.
  • Determining a plane from a plurality of feature points can be implemented using a variety of methods, such as a weighted Hough transform method, a random Hough transform method, a least squares estimation method, a Radon transform method, and the like.
  • the method of weighted Hough transform can be used to detect the planes determined by these selected feature points, which are described in detail below.
  • ⁇ , ⁇ is a plane parameter, and its meaning can be as shown in Figure 16, a set of ⁇ , The ⁇ parameter determines a plane.
  • the plane parameter ⁇ in equation (6), ⁇ has its own range of values, and their range of values is related to how the three-dimensional Cartesian coordinate system is set. For example, for three-dimensional volume data, the origin position of the three-dimensional Cartesian coordinate system is different, and the range of the corresponding plane parameters is also different.
  • the value range of the parameter ⁇ can be expressed as follows:
  • W, H, and F are the dimensions of the three-dimensional volume data
  • F is the number of image frames in the three-dimensional volume data
  • W is the width of the image frame
  • H is the height of the image frame.
  • Hough space In the three-dimensional space corresponding to the three-dimensional volume data, there are an infinite number of planes passing through one point, that is, corresponding to an infinite number of ⁇ , ⁇ , this can construct a new parameter space, here called Space, that is, Hough space, the idea of Hough transform is to project each point in the original three-dimensional space corresponding to the three-dimensional volume data into Hough space. By detecting the peak of Hough space, the peak point corresponds to the original three-dimensional corresponding to the three-dimensional volume data. The plane in space.
  • the step of weighting the Hough transform may include S11 to S14 as shown in FIG.
  • S11 Calculate the parameter value range and sampling step size.
  • the value of the parameter ⁇ can be as shown in equation (7), ⁇ ,
  • the maximum value range can be determined with reference to FIG. 16, for example, 0° ⁇ 360°,
  • the value range may also be narrowed according to some prior knowledge.
  • S12 Generate a Hough matrix and initialize it.
  • the Hough matrix is generated and initialized to 0.
  • the size of a three-dimensional Hough matrix can be:
  • three 1-dimensional Hough matrices may also be used herein, and the sizes may be respectively
  • V i is the value of the i-th feature point P i (eg, a gray value or a convolved result value, etc.).
  • the weighted Hough transform considers that the contribution value of each feature point P i to the plane detection is different in the selected feature points, and the corresponding value V i is larger, and the corresponding contribution on the Hough matrix is larger. Big.
  • the V i value of each feature point in the foregoing method may be set to 1 regardless of the difference in contribution of each feature point. At this point, one plane determined by these feature points can still be detected.
  • the aforementioned weighted Hough transform method degenerates into a conventional Hough transform algorithm.
  • planar detection method can also be used in embodiments of the invention.
  • a plane determined by the selected feature points may be detected using a random Hough transform method.
  • the specific steps of the random Hough transform method may include S21 to S27 as shown in FIG.
  • step S22 Generate a Hough matrix and initialize it to 0.
  • a 3-dimensional Hough matrix is generated and initialized to 0. This step may be the same as or similar to step S12 in the foregoing method.
  • Step 23 to step 255 are repeated N times.
  • N is a preset parameter, which can be set according to actual needs.
  • N can take 50,000.
  • N can also take other values here.
  • another method of detecting a plane of the selected feature point determination may include S31 to S37 as shown in FIG.
  • Steps 32 to 35 are repeated N times, where N is the number of iterations, and can be set as needed.
  • the plane equation representation method of the equation (6) is adopted, and the plane detection is the coefficient ⁇ of the equation. ⁇ .
  • the representation of the equation does not affect the execution of the algorithm of the present invention.
  • the above method still applies, just do a simple modification.
  • the fetal head orientation and/or the face orientation process can be determined in combination with the foregoing embodiment, and if it is determined that the head is inverted, the adjustment is positive.
  • the face orientation is identified, and/or identified therein, and then the adjusted and/or labeled median sagittal section is displayed on the display to facilitate visualization by the physician in the fetal brain.
  • the median sagittal section is a longitudinal section located in the middle of the fetal head, and the section is included in other sections intersecting the median sagittal section
  • the information at the intersection with the median sagittal section that is, the information at the intersection.
  • the intersection of the section and the median sagittal section appears as a brighter line (because, as described above, in the three-dimensional image or three-dimensional volume data of the fetal brain, the median sagittal section appears as a ratio
  • the flow of detecting the median sagittal plane from the three-dimensional volume data is as shown in FIG. 21: a step 110 of extracting at least two slices in the three-dimensional volume data, a step 111 of extracting the midline of the brain in the slice, and The plane 112 defined by the midline of the brain is detected.
  • At step 110 at least two slices are extracted in the three-dimensional volume data.
  • the extraction of the cut surface may have different extraction modes, for example, a plane parallel to the cut surface L2 in FIG. 5 and/or parallel to the cut surface L1 in FIG. 5 may be extracted; or any other cut surface may be extracted, for example with L2 and/or L1 A cut angle at an angle.
  • step 111 the midline of the brain is extracted in each of the extracted cut surfaces, thereby obtaining a plurality of straight lines representing the midline of the brain.
  • the midline of the brain appears as a straight line on the cut surface, and its gray value is higher than the gray value on both sides. Therefore, the extraction of the midline of the brain can be achieved using this feature.
  • extracting the midbrain line therein may include steps S40 to S41 as shown in FIG.
  • a brain midline feature region that conforms to the aforementioned brain midline feature may be first extracted in the slice, that is, a line representing a gray value corresponding to the line is greater than a gray value on both sides of the line is extracted in the slice.
  • Brain midline feature area The method of extracting the feature line of the midline of the brain can be The sagittal feature region extraction method described above is similar.
  • the feature extraction operator can be used to convolve the slice, and the convolved slice includes the extracted brain midline feature region.
  • line and brain midline should not be ideally interpreted as a theoretical “line”, but rather an area that actually has a certain width and/or thickness.
  • the feature extraction operator can be designed according to the characteristics of the midline of the brain that needs to be extracted.
  • the features of the midline of the brain are similar to those of the median sagittal plane described above, and therefore, an operator similar to the feature extraction operator in the foregoing can be used, for example, with the above equations (1) to (5). Any one of the similar operators.
  • the feature point parameters of the feature points may include the coordinates of the feature points and/or the values of the feature points (eg, grayscale values or convolved values, etc.) or other suitable parameters.
  • the specific conditions mentioned herein can be determined based on the nature of the feature extraction operator employed. For example, if an operator similar to the feature extraction operators (1) to (5) described above is used, the aforementioned specific condition may be set as a point in the convolution result whose value is greater than a certain threshold, which may be an empirical parameter. Can be determined according to actual needs.
  • These selected feature points usually define a straight line.
  • the straight line determined by the selected feature points can be detected, and the straight line is considered to be a straight line of the midbrain in the cut surface.
  • the weighted Hough transform method, the random Hough transform method, the stochastic optimal energy method, etc. mentioned in the method for detecting the selected feature point determined plane in the three-dimensional space described above can be used for the line detection in this step. Just do a simple modification on the details.
  • the Hough matrix is a two-dimensional ⁇ - ⁇ matrix.
  • each iteration only wants to select from the selected feature points. You can calculate a straight line by randomly selecting two points.
  • the rest of the algorithm is basically the same as the three-dimensional plane detection method, and will not be described in detail here.
  • other methods may also be used to detect the determined line of the feature point, for example, including but not limited to the randon transform method, the phase encoding method, and the minimum two. Multiply estimates and so on.
  • these extracted midline lines will define a plane, and the plane they define is the plane in which the median sagittal section is located.
  • step 112 detecting the plane defined by the line of the midline of the brain, the plane in which the median sagittal plane is located, that is, the fetal brain is obtained.
  • a variety of methods can be used to detect the plane defined by these brain midline lines. For example, in one embodiment, three non-collinear points in the detected midline line can be substituted into the plane equation to calculate the parameters of the plane equation; the method can also be performed several times, and finally The test results are averaged as the final test result.
  • Another method may be to take N points in the detected midline line of the brain, and then fit the parameters of the plane equation by least squares estimation; or take the extracted N points as input, and adopt the three-dimensional plane detection
  • the proposed Hough transform method, random Hough transform method, stochastic optimal energy method and other methods detect the plane equation.
  • the inverted image is adjusted to be erect and/or according to the judgment result. Or identify the face orientation and then send it to the display for display.
  • this embodiment can also utilize this feature of the median sagittal section of the fetal brain to detect the median sagittal plane in the three-dimensional volume data. For example, some alternative cuts can be selected in the three-dimensional volume data, and then the symmetry of the regions on either side of the alternative cuts can be calculated, and the best alternative cuts on the sides are considered to be the desired median sagittal cuts.
  • the flow of detecting the median sagittal plane from the three-dimensional volume data is as shown in FIG. 23, including: a step 120 of selecting an alternative slice, a step 121 of calculating a symmetry index of the candidate slice, and determining Step 122 of the section of the symmetry index that satisfies the condition.
  • a set of alternative slices can be selected in the three-dimensional volume data.
  • the choice of alternative cuts can be determined as needed. For example, all of the sections of the three-dimensional volume data that are separated by a certain interval (or step size) in one or more specific directions within a certain range may be selected.
  • the "certain range” may be relative to one or more of the three-dimensional volume data.
  • the angular extent of the line and/or face may also be a range of distances relative to one or more points, lines and or faces in the three-dimensional volume data; said "in one or more directions” means the facet
  • the "interval" or "step” may be a distance interval or a step size, or an angular interval or a step size.
  • all the sections of the interval or the step size in one or more directions in the whole range of the three-dimensional volume data may be selected; or, in an embodiment, according to some prior knowledge
  • select alternative sections remove the alternative sections that are likely to be included in the median sagittal plane.
  • the median sagittal plane of the fetal head is a longitudinal section located at a mid-position of the fetal head (ie, in a direction from the fetal head portion to the fetal neck portion from the three-dimensional volume data), based on the three-dimensional volume data In the general direction of the fetal image, a longitudinal section generally at the middle of the head is selected as an alternative section.
  • the section of the section to the direction of the neck portion of the fetus, or the normal to which the normal is perpendicular to the direction from the fetal head portion of the three-dimensional volume data to the fetal neck portion is referred to as the "longitudinal section" of the three-dimensional volume data.
  • a set of longitudinal cuts in the three-dimensional volume data may be selected as the aforementioned set of alternative cuts, for example, selecting a set of longitudinal cuts substantially at the middle of the head (eg, in the middle of the head) All longitudinal sections of a particular step or spacing within a particular area of the location are included as an alternative set of cuts.
  • the user's input may also be received, the user input indicating the possible range in which the median sagittal section is located, and then the section within the range indicated by the user is selected as an alternative section.
  • all the cut surfaces in a certain step of the three-dimensional volume data may be selected, that is, all the cut surfaces in the entire range of the search for the three-dimensional volume data are traversed in a certain step size.
  • the plane parameter ⁇ when the equation (6) is used to represent the tangent equation, the plane parameter ⁇ , The range of values of ⁇ , and the selected step size ⁇ step , The value of ⁇ step achieves the choice of the alternative slice.
  • the value of the parameter ⁇ ranges from the equation (7), ⁇
  • the maximum value range can be, for example, 0° ⁇ ⁇ ⁇ 360°, (Refer to Figure 16). It is easy to understand that when the coordinate system is set differently, the range of values of the parameters will also change.
  • step 121 After the candidate slice is selected, in step 121, all the alternative slice equations ⁇ in the range of values of the plane parameters can be traversed by the aforementioned step size. ⁇ , calculate the symmetry index of each candidate slice.
  • the symmetry index is mainly used to measure the similarity of data on both sides of the alternative slice.
  • At least a pair of first and second regions may be selected on both sides of the alternate slice in the three-dimensional volume data, and the first region and the second region The region is symmetric about the candidate slice and then uses the data in the first region and the data in the second region to calculate a symmetry index for the candidate slice.
  • the "data in the first region” refers to the value of the data point in the three-dimensional volume data falling in the first region
  • the “data in the second region” refers to the three-dimensional in the second region. The value of the data point in the volume data.
  • a plurality of pairs of the first region and the second region may also be selected, and a symmetry index is calculated for each pair of the first region and the second region, and then according to the plurality of pairs of the first region
  • a plurality of symmetry indices corresponding to the second region obtain a final symmetry index of the candidate slice. For example, averaging the plurality of symmetry indices as a symmetry index of the corresponding candidate slice; or using the weighted average of the plurality of symmetry indices as a symmetry index of the corresponding candidate slice, wherein the weighting coefficient may Determined according to the location or other properties of the selected first and second pair of regions; and so on.
  • the final symmetry index of the candidate slice may be a function of a symmetry index calculated from a plurality of pairs of the first region and the second region, respectively.
  • the symmetry index can be calculated in a variety of ways.
  • the sum of the absolute values of the differences of the gray values of the corresponding points in the first region and the second region may be used as the symmetry index, namely:
  • E is the symmetry index
  • ⁇ is the symmetrical first and second regions selected on both sides of the plane
  • I L is the data value of the point in the first region
  • I R is related to the point in the first region The data value of the point in the second region where the slice is symmetrically symmetrical.
  • the "corresponding point in the first region and the second region” means a point in the first region and the second region that is symmetrical with respect to the alternative slice plane.
  • the symmetry index of the alternative slice may also be the correlation coefficient of the aforementioned first region and the second region, namely:
  • E is the symmetry index
  • ⁇ is the symmetrical first and second regions selected on both sides of the plane
  • I L is the data value of the point in the first region
  • I R is related to the point in the first region The data value of the point in the second region where the slice is symmetrically symmetrical.
  • the definition of the symmetry index includes but is not limited to the above two methods, and other similar definitions may be used, such as the Euclidean Distance of the first region and the second region, and the cosine similarity of the first region and the second region. (Cosine Similarity) and so on.
  • the symmetry index is calculated and a set of symmetry indices is obtained. Then, a feature symmetry index satisfying the feature condition is selected from the set of symmetry indexes.
  • the candidate slice corresponding to the feature symmetry index is considered to be a mid-sagittal shape of the desired fetal head. section.
  • the "characteristic condition” may be a condition indicating that the symmetry of the alternative section is optimal. This characteristic condition may vary depending on the calculation method of the symmetry index. For example, for the symmetry index calculated according to the above formula (12), it can be seen that the smaller the E value (ie, the symmetry index), the more similar the image pixels on both sides of the alternative slice, that is, the better the symmetry, therefore, At this time, the feature condition may be "the smallest symmetry index".
  • the E value ie, the similarity index
  • the characteristic condition may be "the symmetry index is closest to 1" or the "symmetry index is the largest".
  • the feature conditions can also be similarly defined.
  • the feature condition when the symmetry index is the Euclidean distance between the first region and the second region, the feature condition may be “the smallest symmetry index”, that is, the smaller the symmetry index (ie, the smaller the Euclidean distance), the first region. The better the symmetry with the second region; when the symmetry index is the first region and the second region.
  • the characteristic condition when the cosine similarity of the domain, the characteristic condition may be "the maximum symmetry index", that is, the greater the symmetry index (ie, the greater the cosine similarity), the better the symmetry of the first region and the second region; .
  • the inverted image is adjusted to be erect and/or according to the judgment result. Or identify the face orientation and then send it to the display for display.
  • the median sagittal section of the fetal head will exhibit some specific structure, that is, the image of the median sagittal section of the fetal head will have some unique structural features.
  • embodiments of the present invention may also utilize this feature of the median sagittal section of the fetal head to generate a template of the median sagittal plane of the fetal head through images of the official sagittal section of other fetal heads that have been previously obtained.
  • the section with the highest similarity to the template image is the median sagittal section of the fetal head.
  • the flow of detecting the median sagittal plane according to the three-dimensional volume data is as shown in FIG. 24, including: a step 130 of acquiring a template image, a step 131 of selecting an alternative slice, and calculating an alternative slice and a template image.
  • a template image of the median sagittal plane of the fetal head can be acquired.
  • the template image may be generated in advance according to an image of a median sagittal section of other fetal heads that has been obtained previously and stored in a storage device, in the three-dimensional imaging process of the embodiment of the present invention, directly from The reading is obtained in the storage device; or it may be generated in the three-dimensional imaging process of the embodiment of the present invention.
  • the template image may be one or plural.
  • the plurality of templates herein may be used for matching the cut surfaces in the three-dimensional volume data of different sizes.
  • each candidate slice can be matched with each template image in the subsequent matching process.
  • a set of alternative slices can be selected in the three-dimensional volume data.
  • the choice of alternative cuts can be determined as needed. For example, all of the sections of the three-dimensional volume data that are separated by a certain interval (or step size) in one or more specific directions within a certain range may be selected.
  • the "certain range” can be relative to the three-dimensional body
  • the angular extent of one or more lines and/or faces in the data may also be a range of distances relative to one or more points, lines and or faces in the three-dimensional volume data; said “in one or more "In the direction” means that the normal of the facet is in the one or more directions;
  • the "interval” or “step” may be a distance interval or a step size, or may be an angular interval or a step size.
  • all the aspects of the interval or the step size in one or more directions in the whole range of the three-dimensional volume data may be selected; or, in this embodiment, the prior knowledge may be selected according to some prior knowledge.
  • the median sagittal plane of the fetal head is a longitudinal section located at a mid-position of the fetal head (ie, a section in the direction from the head portion to the neck portion), based on the general orientation of the fetal image in the three-dimensional volume data, A set of longitudinal cuts in the three-dimensional volume data may be selected as the aforementioned set of alternative cuts, for example, selecting a set of longitudinal cuts generally at the middle of the head (eg, a particular step or spacing within a particular area of the middle of the head) All longitudinal sections of the section are used as alternative cuts for this group.
  • the user's input may also be received, the user input indicating the possible range in which the median sagittal section is located, and then the section within the range indicated by the user is selected as an alternative section.
  • all the cut surfaces of a certain step size in the three-dimensional volume data may also be selected, that is, all the cut surfaces and the template images in the entire range of the matched three-dimensional volume data are traversed in a certain step size.
  • the plane parameter ⁇ when the equation (6) is used to represent the tangent equation, the plane parameter ⁇ , The range of values of ⁇ , and the selected step size ⁇ step , The value of ⁇ step achieves the choice of the alternative slice.
  • the range of the parameter ⁇ is as shown in the formula (7), ⁇ ,
  • the maximum value range can be, for example, 0° ⁇ ⁇ ⁇ 360°, (Refer to Figure 16). It is easy to understand that when the coordinate system is set differently, the range of values of the parameters will also change.
  • the template image may also be one, in which case the template image is generated at a specific size.
  • the step of aligning the three-dimensional volume data with the template image is performed, the step of aligning the three-dimensional volume data and the template image into the same scale space, that is, making the three-dimensional volume data
  • the size of each structure in the neutralized template image is approximately the same.
  • the three-dimensional volume data has approximately the same size as each structure in the template image, so that the subsequent matching process is easier to implement, the matching effect is better, and the calculation amount of the matching process is reduced.
  • the method for aligning the three-dimensional volume data with the template image may be detecting a slice image in the three-dimensional volume data (for example, the image of the middlemost frame, that is, the image of the F/2 frame, or the adjacent frame or other frame may be taken.
  • a slice image in the three-dimensional volume data for example, the image of the middlemost frame, that is, the image of the F/2 frame, or the adjacent frame or other frame may be taken.
  • Specific structural features such as skull aura, etc.
  • the size of the image is the same size level.
  • transforming the three-dimensional volume data to the same size level as that of the template image means that the same or corresponding structural features in the three-dimensional map data and in the template image have the same size by transformation.
  • the three-dimensional volume data can also be aligned to the template image to the same scale space using any other suitable method.
  • step 132 After the candidate slice is selected, in step 132, all the alternative slice equations ⁇ in the range of values of the plane parameters can be traversed by the aforementioned step size. ⁇ , matching each of the alternative slices and the aforementioned template image. For example, a similarity index for each candidate slice and template image can be calculated.
  • the similarity index is used to measure the similarity between the alternative slice and the template image.
  • the similarity index can be calculated in a variety of ways.
  • the similarity index may be the sum of the absolute values of the differences between the alternate slice and the gray value of the corresponding point in the template image, ie:
  • E is the similarity index
  • ⁇ is the image space of the alternative slice
  • I L is the data value of the point of the alternative slice
  • I R is the data value of the point in the template image corresponding to the point in the candidate slice.
  • the corresponding slice and the corresponding point in the template image means a point having the same position in the candidate slice and in the template image.
  • the similarity index may also be a correlation coefficient of the candidate slice and the template image, namely:
  • E is the similarity index
  • ⁇ is the image space of the alternative slice
  • I L is the data value of the point of the alternative slice
  • I R is the data value of the point in the template image corresponding to the point in the candidate slice.
  • the definition of the similarity index includes but is not limited to the above two methods, and other similar definitions may be used.
  • the similarity index is calculated to obtain a set of similarity indices. Then, a feature similarity index satisfying the feature condition is selected from the set of similarity indexes.
  • the candidate slice corresponding to the feature similarity index is considered to be a mid-sagittal shape of the desired fetal head. section.
  • the "feature condition” may be a condition indicating that the similarity between the candidate slice and the template image is optimal. This characteristic condition may be different depending on the calculation method of the similarity index.
  • the feature condition may be "the smallest similarity index”.
  • the similarity index calculated according to the above formula (15) the larger the E value (ie, the similarity index) (for the equation (15), that is, the closer the E value is to 1), the image pixels of the candidate slice and the template image are illustrated.
  • the characteristic condition may be "the greatest similarity index” or "the similarity index is closest to 1".
  • the feature conditions can also be similarly defined.
  • the feature condition may be “the smallest similarity index”, that is, the smaller the similarity index (ie, the smaller the Euclidean distance), The more similar the candidate slice and the template image are;
  • the similarity index is the cosine similarity between the candidate slice and the template image, the feature condition can be “the largest symmetry index”, that is, the greater the similarity index (ie, the cosine similarity) The greater the degree, the more similar the alternative slice and template image; and so on.
  • the image is determined according to the judgment result. Adjust to the desired orientation and/or identify the orientation of the fetal head and then send it to the display for display.
  • the method of detecting the median sagittal section of the fetal brain is not limited to the methods of the foregoing embodiments, and any other suitable method of detecting the median sagittal section of the fetal brain may be used.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Pathology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Physics & Mathematics (AREA)
  • Radiology & Medical Imaging (AREA)
  • Veterinary Medicine (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Neurology (AREA)
  • Gynecology & Obstetrics (AREA)
  • Pregnancy & Childbirth (AREA)
  • Physiology (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Graphics (AREA)
  • Primary Health Care (AREA)
  • Epidemiology (AREA)
  • Databases & Information Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Ultra Sonic Daignosis Equipment (AREA)

Abstract

一种三维超声成像方法及三维超声成像装置，三维超声成像方法包括以下步骤：向胎儿头部发射超声波；接收超声回波，获得超声回波信号；根据超声回波信号获得胎儿头部的三维体数据；根据胎儿头部正中矢状切面的特征，从三维体数据中检测正中矢状切面；确定正中矢状切面中胎儿头部的朝向方位；根据胎儿头部的朝向方位，将正中矢状切面显示为适于观察的图像。将正中矢状切面的图像处理为符合人观察习惯的图像予以显示，使得医生更加容易地辨别和观察胎儿脑部的正中矢状切面的情况。

Description

三维超声成像方法和装置 技术领域
本发明涉及医用超声成像技术领域，尤其涉及一种三维超声成像方法及装置。
背景技术
超声仪器一般用于医生观察人体的内部组织结构，医生将超声探头放在人体部位对应的皮肤表面，可以得到该部位的超声图像。超声由于其安全、方便、无损、廉价等特点，已经成为医生诊断的主要辅助手段之一。其中，产科是超声诊断应用最广泛的领域之一，在该领域，超声避免了X射线等对母体及胎儿的影响，其应用价值明显优于其它影像学检查设备。超声不仅能进行胎儿形态学的观察和测量，还能获得胎儿呼吸、泌尿等生理、病理方面的多种信息，以评价胎儿的健康及发育状况。
在胎儿神经系统检查中，胼胝体和小脑蚓部是两个很重要的检查项目，其中胼胝体是大脑半球中最大的连合纤维，负责大脑两半球间的通讯，缺失或发育不良将导致癫痫、智力低下、运动功能障碍等一系列并发症。小脑蚓部的缺失或发育不良为Dandy-walker综合征的表现，50％的Dandy-walker患者精神运动发育迟滞和智力低下，且常伴有染色体异常和其它畸形，预后差，死亡率高。可见，胼胝体和小脑蚓部的异常均是重大疾病的表现，如未能在产检中发现，将给患者家庭和社会带来巨大精神和经济负担，同时，作为检查主体的医院，也可能因此产生医疗纠纷。然而，在胎儿神经系统的检查中，胼胝体和小脑蚓部是最容易误诊和漏诊的项目，究其原因，在于胎儿的正中矢状切面是观察胼胝体和小脑蚓部的最佳切面，但由于胎儿体位、羊水、鼻骨遮挡、医生技术水平等因素的影响，在常规的二维超声下很难获得胎儿的正中矢状切面，即便能获得，也需要花费很长的检查时间，很多医生只能通过其它切面(如小脑切面、丘脑切面等)进行非直观的诊断，容易出现误诊和漏诊。
近年来，随着三维超声在临床上的广泛应用，部分医生通过以双顶径切面作为起始平面对胎儿进行三维扫查，然后通过手动旋转、平移等几何变换，在3D超声中的第三平面调出正中矢状切面来，在该切面下检查胼胝体和小脑蚓部。从各种公开文献中可以看出，采用该方法得到的矢状切面的图像质量虽然比二维图像略差，但胼胝体和小脑蚓部的显示率却非常高，通过该方法可以快速、准确地判断胼胝体和小脑蚓部是否异常。然而，医生需要对三维空间有非常深刻的理解，才能够在3D 下通过手动旋转、平移几何操作调节出正中矢状切面，但大部分超声医生都是非理工科背景，对三维空间缺乏理解，很难从一个体数据中通过手动的方法将正中矢状切面调节出来。因此，尽管经过了多年的发展，也只有一小部分大医院中的小部分医生才掌握了该诊断技术。
发明内容
本发明的目的之一是，提供一种三维超声成像方法及装置，其能够对胎儿头部进行三维成像，自动检测胎儿头部的正中矢状切面，并能够判断出胎儿头部的朝向方位，根据判断结果将正中矢状切面显示为符合人的观察习惯的图像。
本发明实施例公开的技术方案包括：
提供一种三维超声成像方法，包括：向胎儿头部发射超声波；接收超声回波，获得超声回波信号；根据所述超声回波信号获得胎儿头部的三维体数据；根据胎儿头部正中矢状切面的特征，从所述三维体数据中检测正中矢状切面；检测所述正中矢状切面中和/或与所述正中矢状切面平行或者相交的切面中表示胎儿头部中的特定组织区域的图像区域；根据所述图像区域确定所述正中矢状切面中胎儿头部的朝向方位；根据所述胎儿头部的朝向方位，旋转所述正中矢状切面，使得旋转后所述正中矢状切面中的胎儿头部在预定方位，或者在所述正中矢状切面上标识胎儿头部的朝向方位。
本发明的实施例中，还提供了一种三维超声成像装置，包括：探头，用于向胎儿头部发射超声波并接收超声回波，获得超声回波信号；三维成像模块，用于根据所述超声回波信号获得胎儿头部的三维体数据，根据胎儿头部正中矢状切面的特征，从所述三维体数据中检测正中矢状切面，并确定所述正中矢状切面中胎儿头部的朝向方位，并根据所述胎儿头部的朝向方位，旋转所述正中矢状切面，使得旋转后所述正中矢状切面中的胎儿头部在预定方位，或者在所述正中矢状切面上标识胎儿头部的朝向方位；显示器，用于显示所述正中矢状切面。
本发明的有益效果是：可以对胎儿进行超声扫描获得胎儿头部的三维体数据，并根据获得的三维体数据，自动检测胎儿脑部的正中矢状切面，然后自动判断该正中矢状切面中的胎儿头部的朝向方位如头部倒立与否以及脸朝向，根据判断结果将正中矢状切面图像显示为符合人观察习惯的图像，使得医生更加容易地辨别和观察胎儿脑部正中矢状切面的情况。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，其中相同参考标号表示相同部分。
图1为本发明实施例1的三维超声成像装置的结构框图示意图；
图2为本发明实施例1的三维超声成像方法的流程示意图；
图3为本发明实施例1中三维体数据的示意图；
图4为胎儿头部的正中矢状切面的位置示意图；
图5为胎儿头部的正中矢状切面的示意图；
图6为图5中的L1切面的示意图；
图7为图5中的L2切面的示意图；
图8为正中矢状切面中胎儿头部倒立和正立的示意图；
图9为本发明一种实施例中对正中矢状切面中胎儿头部的上下朝向进行判断的流程示意图；
图10为确定颅骨朝向示意图；
图11为本发明另一种实施例中对正中矢状切面中胎儿头部的上下朝向进行判断的流程示意图；
图12为本发明实施例3的三维超声成像方法的流程示意图；
图13为本发明一种实施例中判断脸部朝向的方法流程示意图；
图14为本发明实施例4的三维超声成像方法的流程示意图；
图15为本发明一个实施例的检测正中矢状面的步骤的流程示意图；
图16为三维空间中的平面及其平面参数的示意图；
图17为本发明一个实施例中的三维Hough矩阵的示意图；
图18为本发明一个实施例中加权Hough变换的过程示意图；
图19为本发明一个实施例中随机Hough变换的过程示意图；
图20为本发明一个实施例中检测选择出的特征点确定的平面的方法的过程示意图；
图21为本发明另一个实施例的检测正中矢状面的步骤的流程示意图；
图22为本发明一种实施例中提取脑中线的流程示意图；
图23为本发明再一个实施例的检测正中矢状面的步骤的流程示意图；
图24为本发明又一个实施例的检测正中矢状面的步骤的流程示意图；
图25和26为本发明一个实施例中的用于标识胎儿头部的朝向方位的图标的示意图。
具体实施方式
下面通过具体实施例并结合附图对本发明作进一步详细说明。
[实施例1]
本实施例提供了一种三维超声成像装置，其结构框图如图1所示。三维超声成像装置包括探头102、发射/接收选择开关103、发射电路104、接收电路105、波束合成模块106、信号处理模块107、三维成像模块108和显示器109。发射电路104将一组经过延迟聚焦的脉冲发送到探头102，探头102向受测机体组织(图中未示出)发射超声波，经一定延时后接收从受测机体组织反射回来的带有组织信息的超声回波，并将此超声回波重新转换为电信号。接收电路105接收这些已转换为电信号的超声回波信号，并将这些超声回波信号送入波束合成模块106。超声回波信号在波束合成模块106完成聚焦延时、加权和通道求和，在经过信号处理模块107进行信号处理。经过信号处理模块107处理的信号送入三维成像模块108，经过三维成像模块108处理，得到三维图像等可视信息，然后送入显示器109进行显示。
当探头102扫描一个扫描周期后，经过信号处理模块107处理后的信号在三维成像模块108中形成一卷极坐标下的三维体数据，该极坐标下的三维体数据经过重建处理，将极坐标体数据转换成直角坐标体数据，从而获得一卷直角坐标下的三维体数据。然后，三维成像模块108对该直角坐标系下的三维体数据进行计算，从而获得可视信息，并在显示设备109上进行显示。
本实施例的三维超声成像装置中，三维成像模块108还包括用于自动检测和处理胎儿正中矢状切面的子模块，该子模块能够根据获得的胎儿头部的三维体数据，从中自动检测出胎儿的正中矢状切面，对其进行处理以检测正中矢状切面中胎儿头部的朝向方位(例如，胎儿头部和/或胎儿头部中的脸部朝上、朝左上、朝右上、朝左、朝右、朝下、朝左下、朝右下或者朝向其他方向，等等)，然后旋转该正中矢状切面使得旋转后该正中矢状切面中胎儿头部在预定的方位(例如，头顶朝上或者朝 下或者朝向任何其他期望的方位，脸部朝上或者朝下或者朝向任何其他期望的方位，等等，例如以便于医生观察或者以便于符合医生的习惯等等)和/或在该正中矢状切面中标识检测出的胎儿头部的朝向方位，并显示处理后的正中矢状切面(下文中详述)。
基于上述三维超声成像装置，本实施例提供了一种三维超声成像的方法，其流程如图2所示。
在步骤21中，首先使用三维超声成像装置对胎儿头部进行三维扫描，向胎儿头部发射超声波并且接收超声回波，获得超声回波信号，超声回波信号经过如前文所述的处理，从而获得胎儿头部的三维体数据(下文中简称为“三维体数据”)。对扫描目标进行三维扫描并且处理超声回波信号获得三维体数据的具体步骤可以与本领域内常用的三维扫描和成像的方法相同或相似，在此不作详述。经过步骤21，可以获得胎儿头部的至少一卷三维体数据。例如，一卷三维体数据可以如图3所示。从图3可见，该卷体数据可以是由F帧大小为W×H的图像帧构成，其中W为图像帧的宽度，H为图像帧的高度。此外，由图3可以看出，图3中将图像帧的宽度方向定义为X方向，将图像帧的高度方向定义为Y方向，多帧图像帧排列的方向定义为Z方向。可以理解，其中X、Y、Z方向也可以以不同的方式定义。
步骤21中获得三维体数据后，本发明实施例的方法中，期望能够自动从三维体数据中检测出胎儿头部的正中矢状切面。
胎儿头部的正中矢状切面的位置如图4所示，图4中的线D即代表胎儿脑部的正中矢状切面的位置。胎儿头部的正中矢状切面的一个示意图显示在图5中。可见，在这个正中矢状切面上，包含了关于胎儿的胼胝体、小脑蚓部、透明隔腔的重要信息，此外，从胎儿头部的正中矢状切面上，也能够观察胎儿的小脑延髓池、丘脑黏合、第四脑室等等结构。因此，自动检测出胎儿脑部的正中矢状切面，可以为医师提供大量重要的关键信息，极大地方便医师对胎儿状况的观察。图6和图7分别示意性地图示了胎儿头部的与正中矢状切面垂直的切面L1和L2。
发明人经研究发现，在胎儿头部的三维图像中，正中矢状切面具有一些特别的特征，例如，在胎儿头部的三维图像中的所有切面中，正中矢状切面整体具有比周围区域的灰度值更大的灰度值，也就是说，在胎儿头部的三维图像中，正中矢状切面表现为灰度值明显大于其附近区域 的灰度值的切面，或者说，正中矢状切面在胎儿头部的三维图像中表现为一个比周围的区域更亮的切面；或者，在胎儿头部中正中矢状切面两侧的结构是近似对称的，因此在胎儿头部的三维图像中，在正中矢状切面两侧的图像数据将表现出近似的对称性；或者，在胎儿头部中，正中矢状切面位于头部中间位置，而在胎儿头部的三维图像中，与正中矢状切面相交的其它切面中都会包含该切面与该正中矢状切面的相交位置处的信息，在其它切面的图像中，该切面与正中矢状切面的交线表现为比较亮的线，即脑中线，这些脑中线的集合即构成了正中矢状切面；等等。本发明的一些实施例中，即是利用胎儿头部的正中矢状切面的这些特征来检测或识别胎儿头部的三维体数据中的正中矢状切面。
基于上述研究发现，本实施例的步骤23根据胎儿头部的正中矢状切面的特征(例如，如前述研究发现的特征，比如灰度特征)，从步骤21中获得的三维体数据中检测出该三维体数据中的正中矢状切面。
本实施例中，前述的“从步骤21中获得的三维体数据中检测出该三维体数据中的正中矢状切面”，可以是在全部胎儿头部的三维体数据中检测，也可以是在胎儿头部的三维体数据中的一部分中检测，例如，可以是在正中矢状切面最可能存在于其中的区域内检测，而去除正中矢状切面明显不可能存在于其中的区域。例如，由于胎儿头部的正中矢状切面是位于胎儿头部中间位置的纵切面(即在从头顶部分到颈部部分的方向上的切面，因此位于头部边缘处的一些区域中明显不可能存在正中矢状切面，这样的区域可以剔除在检测范围之外。
本实施例可以使用多种方法根据该三维体数据检测其中的正中矢状切面。例如，如前文所述，在三维体数据中，正中矢状切面表现出该正中矢状切面内的灰度值大于周围区域的灰度值的特征，因此，本实施例的一种具体实现可以是利用正中矢状切面的这个特征，采用数字图像处理方法中的例如图像分割算法，从三维体数据中检测正中矢状切面。
本发明的实施例的方法中，矢状切面自动检测结果本质上为标记出矢状切面在三维体数据坐标系中的位置，但表现形式可以有多种，如平面方程、矢状切面相对于坐标系原点平移(X、Y、Z方向的平移量)及旋转量(绕X、Y、Z轴的旋转量)、矢状切面相对于原始坐标系的变换矩阵(通常1个4×4矩阵即可表示两个坐标系的变换关系)、甚至空间上三个点的坐标(三个点即确定一个平面)等等。这些表示方式本质均 为在三维体数据坐标系中标记出平面的位置，各种表示方式可以相互转换。本发明的各个实施例(包括本实施例)中，为了表述的方便，统一采用了平面方程表示方法。但是本发明不限制在平面方程表示方法上，而是也包含前述的或者本领域内其它的表示方法。任何矢状切面检测结果表达方式仅仅是表现形式上的差异，不影响本发明的实质，均属于本发明的保护范围。
在检测出了胎儿头部的正中矢状切面之后，虽然可在显示器上显示该正中矢状切面，以便于医生观察胎儿头部的情况，然而，考虑到由于胎儿的体位和探头的方向，检测出的矢状切面图像有可能是倒立的(即胎儿头部朝向图像下方，如图8的左图所示)或者是其他不便于医生观察的方向；而由于此前在二维超声下很难获得胎儿正中矢状切面，大部分医生对矢状切面结构并不是很熟悉，且通常矢状切面的图像分辨率也不是很好，从而导致医生对矢状切面结构的辨认存在一定的困难，如果检测出的矢状切面图像是倒立的或者其他不便于医生观察的方向，将不利于医师对矢状切面结构的辨别，因为这种图像和人的观察习惯不同，不方便医生观察其中的结果。因此，本实施例在检测出胎儿头部的正中矢状切面后，在步骤25中，还对正中矢状切面中胎儿头部的朝向方位(例如，胎儿头顶和/或胎儿脸部朝上、朝左上、朝右上、朝左、朝右、朝下、朝左下、朝右下或者朝向其他方向，等等))进行判断，以便于当检测出正中矢状切面上胎儿头部的朝向不便于医生观察时，旋转该正中矢状切面使得旋转后该正中矢状切面中胎儿头部在预定或者期望的方位(例如，头顶朝上或者朝下或者朝向任何其他期望的方位，或者脸部朝上或者朝下或者朝向任何其他期望的方位，，例如以便于医生观察或者以便于符合医生的习惯等等)和/或在该正中矢状切面中标识检测出的胎儿头部的朝向方位，在显示时(即步骤29)显示旋转或者标识了的正中矢状切面，以便于医生观察，如图8所示。
本发明的一些实施例中，正中矢状切面中胎儿头部的朝向方位可以通过正中矢状切面中和/或三维体数据中与该正中矢状切面平行或者相交的切面中表示胎儿头部中的特定组织区域的图像区域获得。
一些实施例中，这里所说的特定组织区域也可以是胎儿头部中的具有特定相互位置关系的至少两个特定组织区域(比如眼部和鼻部、眼部和嘴部、嘴部和鼻部、或者眼部、鼻部和/或嘴部与透明隔腔、或者眼部、 鼻部和/或嘴部与胎儿头部的其他部位，等等)。在这些实施例中，可以从正中矢状切面中和/或与该正中矢状切面平行或者相交的切面中提取或者检测出表示这些组织区域中的至少两个的至少两个图像区域，并根据这些图像区域之间的相互位置关系确定胎儿头部的朝向方位(例如，胎儿头部中，头顶总是在从嘴部到眼部、从嘴部到鼻部或者从鼻部到眼部的方向上，等等)。确定了这些图像区域之间的相互位置关系，即可根据这些相互位置关系确定胎儿头部的朝向方位。
另一些实施例中，这里所说的特定组织区域可以是胎儿头部中的具有方向特性的组织区域。本文中，这里所说的具有方向特性的组织区域是指其本身或者其位置含有能够指示胎儿头部的朝向方位的信息的组织区域，例如颅骨或者头盖骨(其弯曲方向指示了胎儿头部的朝向方位)；透明隔腔(其朝向和位置可以指示胎儿头部的朝向方向)；嘴部、眼部和鼻部(其总是位于胎儿头部的脸部侧，因此其位置可以指示胎儿头部的朝向方向)；等等。
在这些实施例中，可以从正中矢状切面中和/或与该正中矢状切面平行或者相交的切面中提取或者检测出表示这些组织区域中的一个或者多个的图像区域，根据这些图像区域的位置和/或形状特征确定胎儿头部的朝向方向(例如，眼部和嘴部所在的位置或者侧总是为胎儿头部的前部或者前侧，通过颅骨的弯曲方向可以确定头部的朝向方位，等等)。下文中，分别以该特定组织区域为颅骨和透明隔腔为例进行了说明。
例如，一些实施例中，正中矢状切面中胎儿头部的朝向方位可以利用三位体数据中具有方向特性的结构进行检测或识别。发明人经研究发现，在胎儿头部的三维图像中，颅骨表现为明显的高亮回声。因此可以根据颅骨的朝向来判断胎儿头部的朝向。因此，本实施例中，通过检测正中矢状切面图像或与该矢状切面平行的图像中颅骨的朝向来判断胎儿头部的朝向，检测流程如图9所示。
步骤252：从选取的切面中提取颅骨特征以得到表征颅骨的候选区域。由于在超声中颅骨表现为高亮回声而颅骨两侧回声慢慢减弱，因此可根据这一特征设计多种提取颅骨特征的方法。例如，在本实施例中，一种提取颅骨特征的方法为根据颅骨高亮回声的特点，在已选切面中选取灰度大于预设灰度阈值的区域作为颅骨的候选区域，其中，已选切面是指正中矢状切面和/或至少一个与该正中矢状切面平行的切面，该预设 灰度阈值可根据实际需要确定，即可以是经验阈值，也可以是根据图像的统计学特征确定，比如阈值可以设置为平均灰度*经验系数。又例如，另一种实施例中，颅骨特征的提取可以是利用颅骨中间亮两侧暗的性质，基于微分法设计算子，将算子和已选切面对应的图像做卷积，并保留卷积值大于预设灰度阈值的部分作为特征图像。基于微分法设计的算子可以是包括例如下面(1)～(5)特征提取算子中的一个或者多个。
Figure PCTCN2015078494-appb-000001
Figure PCTCN2015078494-appb-000002
Figure PCTCN2015078494-appb-000003
Figure PCTCN2015078494-appb-000004
Figure PCTCN2015078494-appb-000005
通常经过特征提取，在特征图像上将保留有多个联通的特征区域，因此可以通过设置一些规则来保留一块或多块联通区域作为颅骨的候选 区域。例如，规则可以定义为：选择特征值之和最大的一个或多个特征区域作为候选区域。特征值可以是前述提取颅骨特征时所采用的特征，比如灰度特征，当然，特征值还可以是数字图像处理时常用的特征提取所涉及的特征，比如纹理特征。又例如，规则可定义为：选择平均特征值最大的一个或多个特征区域作为候选区域。再例如，规则可以定义为：采用机器学习的方法，即从特征区域中提取特征，将提取出的特征输入预先训练好的分类器进行分类，分类结果作为候选区域；这里提取的特征可以是特征区域的灰度平均、特征值平均、特征区域的曲率、特征区域熵、一阶矩、二阶矩等等，预先训练好的分类器可以是事先通过一定数量的样本提取上述特征并采用PCA(Principal Component Analysis，主成分分析)、KPCA(Kernel Principal Component Analysis，核主成分分析)、ICA(Independent Component Analysis，独立成分分析)、LDA(Linear Discriminant Analysis，线性判别式分析)、SVM(Support Vector Machine，支持向量机)等任意的分类器进行训练而得到，其具体实现可以参考图像处理与模式识别相关的已有技术，在此不作详述。可以理解，对于前述的已选切面的数量大于1(即选取了多个切面)的情况，规则的定义也是类似的，例如在所有已选切面中，采用特征值之和或特征平均值最大的一个或多个特征区域作为颅骨的候选区域，或者是采用机器学习的方式识别出候选区域。
在步骤254中，对于候选区域，可以通过颅骨的弯曲的方向来确定颅骨朝向。判断颅骨弯曲方向的方法有很多，例如采用二次曲线拟合连通区域，根据二次曲线二次项的系数来判断颅骨的朝向，例如，拟合出的二次项系数大于0，则说明该颅骨朝下，反之，则说明该颅骨朝上。又例如采用机器学习的方法，如PCA，KPCA，LDA，SVM等方法通过学习得到，其具体实现可以参考图像处理与模式识别相关的已有技术，在此不作详述。对于多个候选联通区域的情况，可以分别判断每个联通区域的朝向然后投票判断出最终颅骨的朝向。
在判断出胎儿的头部朝向后，如果头部朝下，则将正中矢状切面的图像旋转180°或将正中矢状切面的图像上下翻转后再进行显示，使得显示出来的图像更为符合人的观察习惯，帮助医师观察胎儿状况。
本发明的前述的各个实施例中，为了表述的方便，统一采用了平面方程表示方法。但是本发明不限制在平面方程表示方法上，而是也包含 前述的或者本领域内其它的表示方法。任何矢状切面检测结果表达方式仅仅是表现形式上的差异，不影响本发明的实质，均属于本发明的保护范围。
本实施例中，实现前述方法的三维超声成像系统不限于通常的集成为一个整体装置的超声成像系统(例如台车式超声成像系统或者便携式超声成像系统)，也可以是分布式的系统。例如，前述方法中的至少一部分步骤或者功能可以在通过数据通信装置(有线或无线地)连接到通常的台车式超声成像系统或者便携式超声成像系统的其它设备上实现，该其它设备可以是例如数据处理工作站、个人电脑、各种智能便携设备、其它超声成像设备、各种网络服务器等等，从而这些至少一部分步骤或者功能与台车式超声成像系统或者便携式超声成像系统整体形成本实施例中的三维超声成像系统。
本实施例中的超声成像方法中，可以对胎儿进行超声扫描获得胎儿头部的三维体数据，并根据获得的三维体数据，自动检测胎儿脑部的正中矢状切面并判断胎儿头部的朝向方位(例如，判断胎儿是否倒立)，对于倒立的将其旋转为正立后予以显示，使得显示结果符合人的观察习惯，并解决了医生手动难以准确定位正中矢状切面的问题，使得医生可以方便地观察胎儿脑部正中矢状切面的情况，可以为医师提供大量重要的关键信息。
[实施例2]
本实施例提供的三维超声成像方法/系统类似实施例1，二者相同部分在此不作重述，不同之处在于：实施例1的步骤25中是通过颅骨朝向来确定正中矢状切面中胎儿的头部朝向；而本实施例的步骤25是根据透明隔腔的朝向来确定正中矢状切面中胎儿的头部朝向。
如图5和图8所示，在超声图像中，透明隔腔的形态表现为弯弯的月牙形，当超声图像中胎儿为正立时，透明隔腔表现为向上凸的月牙形，反之，当超声图像中胎儿为倒立时，透明隔腔表现为向下凸的月牙形。因此，可以根据透明隔腔的朝向来判断正中矢状切面上胎儿的头部朝向。
本实施例中，通过透明隔腔的朝向来确定正中矢状切面中胎儿头部朝向的流程包括步骤252′和步骤254′，如图11所示。
步骤252′：根据透明隔腔在超声图像中的特点从正中矢状切面中检测出透明隔腔对应的联通区域。
在步骤252′中，由于透明隔腔在超声图像中表现为黑色的低回声区域，而其周围的组织通常亮度较高，因此，可根据这一特征设计多种方法分割该低回声区域，由此检测出透明隔腔。例如可以采用图像分割算法进行分割，比如阈值分割、Snake算法、level-set算法、graph-cut分割等，分割出可能会被认为是透明隔腔的区域。通常分割得到的区域可能是多个，因此可以通过设置一定的准则来选取出最像透明隔腔的区域，例如，可以根据区域的形状、灰度、方差等特征或这些特征的组合进行判断，由此得到透明隔腔对应的联通区域。
步骤254′：基于联通区域判断透明隔腔的朝向并根据判断结果确定正中矢状切面中胎儿的头部朝向。
在步骤254′中，对于透明隔腔对应的联通区域，可以采用类似实施例1的判断颅骨朝向的方法来判断透明隔腔的朝向。
例如，采用类似步骤254a1～254a3的方法来判断透明隔腔的朝向，即首先对透明隔腔对应的联通区域进行数学形态学处理以提取出骨架，得到骨架图像，接着在骨架图像上搜索一条最长的且连续的曲线，该曲线为选取出的具有代表性的曲线，然后根据曲线上中间位置的至少一个点的坐标以及分别位于两端位置的至少一个点的坐标进行判断，确定出透明隔腔的朝向。
又例如，采用类似步骤254b1～254b2的方法来判断透明隔腔的朝向，即首先计算透明隔腔对应的联通区域竖直方向的中线；然后用同步骤254a3相同的方法确定出透明隔腔的朝向。
根据确定出的透明隔腔的朝向，显然，如果透明隔腔朝向为向上凸，说明正中矢状切面上胎儿的头部朝上，反之如果透明隔腔朝向为向下凸，则说明正中矢状切面上胎儿的头部朝下。在判断出胎儿的头部朝向后，如果头部朝下，则将正中矢状切面的图像旋转180°或将正中矢状切面的图像上下翻转后再进行显示，使得显示出来的图像更为符合人的观察习惯，帮助医师对胎儿状况进行观察。
[实施例3]
本实施例提供的三维超声成像方法如图12所示，包括：获得胎儿头 部的三维体数据的步骤21、检测正中矢状切面的步骤23、判断脸部朝向的步骤27、以及显示正中矢状切面的步骤29。其中，步骤21、23和29与实施例1或2相同，在此不作重述。基于该方法本实施例还提供了一种实现该方法的三维超声成像装置，该装置的除了三维超声成像部分的结构可参考前述实施例1，此处不再重述。
因为透明隔腔位于颅内的前面，因此，可以通过透明隔腔的位置来确定出胎儿的脸部和背部(即胎儿的脸部朝向)。
在本实施例中，步骤27的过程如图13所示，包括：检测透明隔腔的步骤271、定位颅内中心位置的步骤271、以及判断透明隔腔的位置的步骤273。
在步骤271中，透明隔腔的检测方法类似前述实施例2的透明隔腔检测方法，即对正中矢状切面的图像采用图像分割算法进行处理，得到至少一个区域，结合区域的特征对区域进行与透明隔腔最相似的区域的选择，选择结果为透明隔腔区域，具体过程参考前述步骤252′，在此不作详述。
在步骤272中，考虑到在体数据中，颅内中心可根据胎儿颅骨进行定位，而胎儿颅骨在体数据中表现为一个近似椭球体或椭圆的形状，在超声图像上表现为高亮回声。因此，实施例中，颅内中心的定位过程包括如下步骤272a1～272a3。
步骤272a1：从体数据中选取至少一帧A面图像，即按图4的A-A得到的水平剖面图。
步骤272a2：基于微分法设计算子，例如前述的算子(1)～(5)中的一个或多个，利用设计的算子对A面图像进行特征提取。
步骤272a3：对提取到的特征进行椭球体或椭圆检测，检测算法可以采用任意的已知相关技术中涉及的椭圆检测算法，比如最小二乘估计算法、Hough变换算法、随机Hough变换算法等等。例如可以采用下文中涉及的Hough变换算法的描述来检测。
在检测出椭圆或椭球体后，可以根据相关的几何知识确定出该椭圆或椭球体的中心。
在步骤273中，将透明隔腔区域的坐标信息与中心的坐标信息进行比较，根据比较结果得到正中矢状切面中胎儿头部的脸部朝向，具体地，选择透明隔腔区域上某个点的位置(其在X坐标上的值为x)作为透明 隔腔的位置，例如步骤271中检测到的透明隔腔区域的中心位置或其它位置，假设步骤272得到的椭圆或椭球体的中心在X坐标上的值为xcenter，比较x和xcenter，如果x<xcenter，则说明在正中矢状切面中，透明隔腔位于颅内中心的左边，即正中矢状切面的左边为胎儿的前部(脸部)，右边为胎儿的后部(背部)，反之，则即正中矢状切面的左边为胎儿的后部(背部)，右边为胎儿的前部(脸部)。
基于确定出的正中矢状切面中胎儿头部的朝向方位(例如，脸部朝向或者头顶朝向，等等)，可以在正中矢状切面的图像上标记出胎儿的前部和/或后部，如图8的右图所示。或者，一些实施例中，也可以采用如图25和26的方式，用一个胎儿头像的图标来表示胎儿的前部/后部或者头顶或者脸部的朝向标识，图标中脸的朝向可以和当前数据中脸部的朝向一致。然后，可以将标记有胎儿头部的朝向方位(例如，胎儿的前部和/或后部，等等)的正中矢状切面送入显示器予以显示，为帮助医师对胎儿状况的观察提供了便利。
[实施例4]
图14所示为本实施例的三维超声成像方法的流程示意图，其中各步骤同前述各实施例中对应的步骤，在此不作重述。应该理解，本实施例中步骤25和步骤27的顺序还可以颠倒，即可以先判断脸部方位，再判断头部朝向。显然，本实施例提供了胎儿的头部朝向以及脸部方位的判断，便于医师更容易地辨别矢状切面上的组织结构。
本发明的一些实施例中，关于步骤23即检测正中矢状切面的过程可以如下文所述。
在本发明一种实施例中，根据三维体数据检测正中矢状面的流程如图15所示，包括：特征提取的步骤80、选择特征点的步骤81和检测选择的特征点确定的平面的步骤82。
如前文所述，在三维体数据中，正中矢状切面表现出该正中矢状切面内的灰度值(例如，灰度值，等等)大于周围区域的灰度值的特征。基于此，该实施例利用正中矢状切面的这个特征从三维体数据中检测正中矢状面。因此，在步骤80，可以首先在三维体数据中提取代表满足平面内的灰度值大于平面两侧的灰度值的条件的平面的矢状面特征区域。 也就是说，本实施例中，在三维体数据中提取一些特征区域，并且本实施例的方法中，这些需要的特征区域是代表着三维体数据中满足平面内的灰度值大于平面两侧的灰度值的条件的平面的区域，这种提取出来的特征区域即为本实施例的方法所需要的矢状面特征区域。这样，充分利用“正中矢状切面表现为灰度值明显大于其附近区域的灰度值的切面”的特征，能够获得良好的正中矢状切面的检测效果。
本发明的实施例中，可以使用多种适合的方法从三维体数据中提取这种矢状面特征区域。例如，可以使用特征提取算子与该三维体数据做卷积，得到卷积后的图像，该卷积后的图像中即包含了提取出的矢状面特征区域。
本发明的实施例中，用特征提取算子对三维体数据做卷积时，可以分别对组成三维体数据的每个图像帧分别用二维特征提取算子做卷积，然后将卷积后的图像帧组合成卷积后的三维体数据；或者，也可以直接设计三维特征提取算子，直接用三维特征提取算子与三维体数据做卷积。卷积运算的具体步骤是本领域内熟知的，在此不再详述。
本发明的实施例中，特征提取算子可以根据需要提取的图像特征而设计。例如，如前文所述的实施例中，需要提取区域内的灰度值大于区域两侧的灰度值的矢状面特征区域。此时，可以使用前述实施例1的特征提取算子(1)～(5)中的一个或者多个。
本发明的实施例中，也可以使用上述特征提取算子(1)～(5)经过转置(矩阵转置)、旋转等变形或者相互之间组合之后获得的特征提取算子，也可以使用其它适合的特征提取算子，比如Roberts算子、拉普拉斯高斯算子及其变形、等等。
本发明的实施例中，类似地，也可直接设计三维特征提取算子，在此不再详述。
本发明的实施例中，特征提取算子(二维的或者三维的)的大小可以根据需要设定。
在步骤80中提取了矢状面特征区域之后，在步骤81中，可以从这些提取出的矢状面特征区域中选择其值满足特定条件的特征点，通常，选择至少三个特征点。记录选择出的特征点的特征点参数，这些特征点的特征点参数将用于后续步骤。
本发明的实施例中，特征点参数可以包括特征点的坐标和/或特征点 的值(例如，灰度值或者卷积后的结果值，等等)。
本发明的实施例中，前述的特定条件可以根据所采用的特征提取算子的性质确定。例如，如果采用前述的特征提取算子(1)～(5)，可将前述的特定条件设置为卷积结果中值大于某个阈值的点，该阈值可以为经验参数，可以根据实际需要确定。
此外，本发明的实施例中，为了减少后续平面检测步骤(下文详述)的压力，尽量减少噪声的影响，可以根据一定的先验知识去除一些明显不可能是头部内的点。例如，头部一般都位于三维体数据的中间。因此，可以只选择以三维体数据的中心为球心、以某个阈值为半径的球或椭球内的点为特征点。这里的阈值也可以根据经验或者实际情况确定。
步骤81中选择了特征点之后，这些选择的特征点通常可以确定一个平面，本实施例中，认为这个平面即为正中矢状切面所在的平面，三维体数据中与该平面重合的切面即为胎儿脑部的正中矢状切面。因此，本发明的实施例中，在步骤82中，检测出这些选择的特征点确定的平面，也就确定出了胎儿脑部的正中矢状切面所在的平面。
根据多个特征点确定一个平面可以使用多种方法实现，例如加权Hough变换法、随机Hough变换法、最小二乘估计法、Radon变换法等等。
例如，可以使用加权Hough变换的方法检测这些选择的特征点确定的平面，下面进行详细描述。
在三维空间中，平面方程可以用一般表达式aX+bY+cZ+d＝0或Z＝aX+bY+c或Y＝aX+bZ+c表示，其中a、b、c和d即为确定一个平面的平面参数。
三维空间中，平面方程也可以用如下的平面标准表达式进行表达：
Figure PCTCN2015078494-appb-000006
其中，式(6)中，θ、
Figure PCTCN2015078494-appb-000007
ρ为平面参数，其意义可以如图16所示，一组θ、
Figure PCTCN2015078494-appb-000008
ρ参数即确定一个平面。
式(6)中的平面参数θ、
Figure PCTCN2015078494-appb-000009
ρ有各自的取值范围，它们的取值范围与三维直角坐标系的设置方式有关。例如，对于三维体数据，三维直角坐标系的原点位置不同，则相应的平面参数的取值范围也不同。
例如，图16所示的实施例中，参数ρ的取值范围可以如下式所示:
Figure PCTCN2015078494-appb-000010
其中W、H、F为三维体数据的尺寸，F为三维体数据中的图像帧的数量，W为图像帧的宽度，H为图像帧的高度。
容易理解，当以其它的方式设置三维直角坐标系时，平面参数θ、
Figure PCTCN2015078494-appb-000011
ρ的取值范围相应地为其它值。
在三维体数据对应的三维空间中，过一个点有无数个平面，即对应无数个θ、
Figure PCTCN2015078494-appb-000012
ρ，这样可以构造一个新的参数空间，这里称为
Figure PCTCN2015078494-appb-000013
空间，也即Hough空间，Hough变换的思想为将三维体数据对应的原三维空间中的各个点投影到Hough空间中，通过检测Hough空间的峰值，峰值点就对应了三维体数据对应的原三维空间中的平面。
本发明的一个实施例中，由于θ、
Figure PCTCN2015078494-appb-000014
ρ是连续的参数，因此可以将θ、
Figure PCTCN2015078494-appb-000015
ρ采样，可以将
Figure PCTCN2015078494-appb-000016
细分成不同的单元(如图17所示)。这样，加权Hough变换的步骤可以包括如图18所示的S11～S14。
S11：计算参数取值范围及采样步长。参数ρ的取值范围可以如式(7)所示，θ、
Figure PCTCN2015078494-appb-000017
最大的取值范围可以参考图16确定，例如，0°≤θ<360°，
Figure PCTCN2015078494-appb-000018
本发明的实施例中，也可以根据一些先验知识缩小取值范围。
设最终的取值范围为θmin≤θ≤θmax，
Figure PCTCN2015078494-appb-000019
ρmin≤ρ≤ρmax，采样步长θstep、
Figure PCTCN2015078494-appb-000020
ρstep可以根据实际需要的检测精度确定，例如，一个实施例中，可以取θstep＝1、
Figure PCTCN2015078494-appb-000021
ρstep＝2。当然，也可以取适合的其它值。
S12：生成Hough矩阵并初始化。生成Hough矩阵并初始化为0，一种三维的Hough矩阵的大小可以为：
Figure PCTCN2015078494-appb-000022
本发明的实施例中，这里也可采用3个1维的Hough矩阵，其大小可以分别为
Figure PCTCN2015078494-appb-000023
Figure PCTCN2015078494-appb-000024
Figure PCTCN2015078494-appb-000025
S13：参数投票。对每个选择的特征点，以及前述参数取值范围内的每个θj及
Figure PCTCN2015078494-appb-000026
计算对应的ρl：
Figure PCTCN2015078494-appb-000027
其中(Xi,Yi,Zi)为第i个特征点Pi的坐标。
并将Hough矩阵更新为：
Figure PCTCN2015078494-appb-000028
其中Vi为第i个特征点Pi的值(例如，灰度值或者卷积后的结果值，等等)。
S14：Hough矩阵峰值检测。计算Hough矩阵H中最大值对应的θ、
Figure PCTCN2015078494-appb-000029
ρ。设Hough矩阵H中最大值的位置为
Figure PCTCN2015078494-appb-000030
则平面检测结果为：
θ＝θjθstep+θmin
Figure PCTCN2015078494-appb-000031
ρ＝ρlρstep+ρmin
这里，对于前述的采用3个1维的Hough矩阵的实施例，则分别计算每个Hough矩阵中最大值对应的θ、
Figure PCTCN2015078494-appb-000032
ρ。
本实施例中，加权Hough变换考虑到了选择的特征点中每个特征点Pi对平面检测的贡献值是不同的，其对应的值Vi越大，其在Hough矩阵上对应的贡献也越大。
本发明的实施例中，也可以不考虑每个特征点的贡献的差异，即可将前述方法中的每个特征点的Vi值都设置为1。此时，仍然可以检测出这些特征点确定的一个平面。实际上，此时前述的带权重的Hough变换方法退化为传统的Hough变换算法。
本发明的实施例中，也可以使用其它的平面检测方法。例如，一个实施例中，可以使用随机Hough变换方法检测选择出的特征点确定的一个平面。随机Hough变换方法的具体步骤可以包括如图19所示的S21～S27。
S21：计算参数取值范围及采样步长。计算平面方程参数θ、
Figure PCTCN2015078494-appb-000033
ρ取值范围及采样步长，该步骤可以与前述方法中的S11步骤相同或者类似。
S22：生成Hough矩阵并初始化为0。生成3维Hough矩阵并初始化为0，该步骤可以与前述方法中的S12步骤相同或者类似。
S23：随机选点。从选择出的特征点中随机选择3个点。
S24：平面方程求解，计算平面参数。将3个点的坐标代入平面方程，求解平面方程参数θ、
Figure PCTCN2015078494-appb-000034
ρ，平面方程参数求解方法是本领域技术人员熟知的，在此不再详述。
S25：更新Hough矩阵。将求解出的θ、
Figure PCTCN2015078494-appb-000035
ρ在Hough矩阵对应的位置上加上1。
S26：重复N次步骤23至步骤255。这里N为预先设置的参数，可根据实际需要设置。例如，一个实施例中，N可以取50000。当然，这里N也可以取其它的值。
S27：Hough矩阵峰值检测。计算Hough矩阵中值最大的位置，其对应的θ、
Figure PCTCN2015078494-appb-000036
ρ即为平面检测结果，也就是检测出的平面。
本发明的实施例中，另一种检测选择出的特征点确定的平面的方法(本文中称为随机最优能量法)的步骤可以包括如图20所示的S31～S37。
S31：初始化最优能量E_best＝0。
S32：随机选点。从选择出的特征点中随机选择3个点。
S33：方程求解。将3个点的坐标代入平面方程，求解平面方程参数θ、
Figure PCTCN2015078494-appb-000037
ρ。
S34：当前能量E计算。计算选择出的特征点中到步骤S33中求解出的平面距离小于ε的能量E。
该步骤的具体步骤可以为对选择出的特征点中的每个特征点Pi，计算该点到步骤S33中求解出的平面(θ、
Figure PCTCN2015078494-appb-000038
ρ)的距离，如果距离小于ε，则将当前特征点对应的值Vi累加到能量E中，即E＝E+Vi。ε为一参数，可根据需要进行设置，例如一个实施例中可以设置ε＝5，这里，ε也可以设置为其它的值。
S35：能量更新。如果当前能量E>E_best，则将E_best修改为E，同时将当前平面方程参数更新为最优平面方程参数，否则转到步骤36。
S36：重复步骤32至步骤35N次，这里N为迭代次数，可根据需要设置。
S37：输出方程参数。步骤S36完成后，能量最大的一次迭代对应的平面方程参数即为检测出的平面方程参数。
这样，即检测出了选择出的特征点确定的一个平面。
本实施例中，在步骤34中，也可不累加特征点的值Vi，而直接判断如果点Pi到平面的距离小于ε，则E＝E+1，即认为选择的特征点中的每个特征点对平面检测结果的贡献是一样的。
上述实施例中，采用了式(6)的平面方程表示方法，平面检测即计算方程的系数θ、
Figure PCTCN2015078494-appb-000039
ρ。但是方程的表示形式并不影响本发明所述算法的执行，事实上，对于其它形式的方程表示方法，如aX+bY+cZ+d＝0或 Z＝aX+bY+c或Y＝aX+bZ+c，上述方法仍然适用，只需做简单的修改即可。
采用本实施例的方法检测出了胎儿脑部的正中矢状切面之后，即可结合前述实施例的判断胎儿头部朝向和/或脸部朝向过程，如果判断出头部为倒立则调整为正立，和/或在其中标识出脸部朝向，然后在显示器上显示调整后的和/或是标识后的正中矢状切面，以便于医生观察胎儿脑部中的情况。
如前文所述，并且参考图4至图7，在胎儿脑部的三维图像中，正中矢状切面是位于胎儿头部正中的纵向切面，与正中矢状切面相交的其它切面中都会包含该切面与该正中矢状切面的相交位置处的信息，也就是包含交线处的信息。在其它切面的图像中，该切面与正中矢状切面的交线表现为比较亮的线(因为如前文所述，在胎儿脑部的三维图像或者三维体数据中，正中矢状切面表现为比周围区域更亮的平面)，即脑中线，这些脑中线的集合即构成了正中矢状切面。因此，本实施例可以利用这个特征从三维体数据中检测正中矢状切面。
例如，本实施例中，根据三维体数据检测正中矢状面的流程如图21所示的：在三维体数据中提取至少两个切面的步骤110、在切面中提取脑中线的步骤111、以及检测脑中线确定的平面112。
在本实施例中，在步骤110，在三维体数据中提取至少两个切面。切面的提取可以有不同的提取方式，例如，可以提取平行于图5中的切面L2和/或平行于图5中的切面L1的平面；或者提取任何其它的切面，例如与L2和/或L1成一定角度的切面。提取的切面的数量也没有限制，至少两个切面即可。
提取了切面之后，在步骤111中，在提取出的每个切面中提取脑中线，从而获得多条代表脑中线的直线。
脑中线在切面上表现为直线，并且其灰度值比两侧的灰度值高。因此，脑中线的提取可以利用这个特征实现。
在本实施例中，对于每个提取出的切面，在其中提取脑中线可以包括如图22所示的步骤S40～S41。
S40：提取脑中线特征区域。
在实施例中，可以首先在该切面中提取出符合前述的脑中线特征的脑中线特征区域，也就是在切面中提取代表满足线上的灰度值大于线两侧的灰度值的线的脑中线特征区域。脑中线特征区域提取的方法可以与 前文所述的和矢状面特征区域提取方法类似。例如，可以使用特征提取算子对切面进行卷积，卷积后的切面中即包含了提取出的脑中线特征区域。
应该理解，这里所说的“线”和“脑中线”不应该理想化地解释为理论上的“线”，而是实际上有一定的宽度和/或厚度的区域。
这里，特征提取算子可以根据需要提取的脑中线的特征设计。该实施例中，脑中线的特征与前文所述的正中矢状面特征类似，因此，这里可以使用与前文中的特征提取算子类似的算子，例如与前文中式(1)至式(5)中任何一个类似的算子。
提取了脑中线特征区域中之后，在脑中线特征区域中选择满足特定条件的至少两个特征点，并记录该至少两个特征点的特征点参数。这里，特征点的特征点参数可以包括特征点的坐标和/或特征点的值(例如，灰度值或者卷积后的值，等等)或者其它适合的参数。
这里所说的特定条件可以根据所采用的特征提取算子的性质确定。例如，如果采用与前述的特征提取算子(1)～(5)类似的算子，可将前述的特定条件设置为卷积结果中值大于某个阈值的点，该阈值可以为经验参数，可以根据实际需要确定。
S41：直线检测。
这些选择出的特征点通常确定了直线。本发明的实施例中，可以检测这些选择出的特征点确定的直线，认为该直线即为该切面内的脑中线直线。
前文所述的在三维空间中检测选择出的特征点确定的平面的方法中所提到的加权Hough变换方法、随机Hough变换方法、随机最优能量法等方法均可用于本步骤中的直线检测，只需要在细节上做简单修改即可。
例如，直线的标准方程为ρ＝cosθX+sinθY，共有两个参数θ、ρ，相比于平面方程，少一个参数
Figure PCTCN2015078494-appb-000040
例如，在采用加权Hough变换和随机Hough变换方法时，Hough矩阵为二维的ρ-θ矩阵，在随机Hough变换及随机能量最优法中，每次迭代只想要从选择出的特征点中随机选取两个点，即可计算一条直线。算法的其余部分和三维平面检测方法基本一致，在此不再详述。
此外，本实施例中，也可以使用其它的方法来检测选择出的特征点确定的直线，例如，包括但不限于randon变换法、相位编码法、最小二 乘估计等等。
基于胎儿脑部的三维图像中正中矢状切面的特征，这些提取出的脑中线直线将确定一个平面，它们确定的平面即为正中矢状切面所在的平面。
因此，在步骤111中获得了提取出的各个切面中的脑中线直线之后，在步骤112中，检测这些脑中线直线确定的平面，即可获得正中矢状面所在的平面，也就是胎儿脑部的正中矢状切面所在的平面。
可以使用多种方法检测这些脑中线直线确定的平面。例如，一个实施例中，可以在所检测到的脑中线直线中取3个不共线的点，代入平面方程中，即可计算出平面方程的参数；也可执行该方法若干次，最后对检测结果做平均作为最终的检测结果。
另一种方法可以是在所检测到的脑中线直线中取N个点，然后通过最小二乘估计拟合出平面方程的参数；也可以将提取的N个点作为输入，采用三维平面检测所提到的Hough变换方法、随机Hough变换方法、随机最优能量法等方法检测出平面方程。
采用本实施例的方法检测出了胎儿脑部的正中矢状切面之后，类似前述实施例在判断胎儿头部朝向和/或脸部朝向后，根据判断结果将倒立的图像调整为正立和/或标识出脸部朝向，然后送入显示器上显示。
如前文所述，并且参考图4，可见在胎儿脑部中，正中矢状面两侧的结构是近似对称的，因此在胎儿脑部的三维图像中，在正中矢状切面两侧的图像数据将表现出近似的对称性。因此，本实施例还可以利用胎儿脑部正中矢状切面的这个特征来检测三维体数据中的正中矢状面。例如，可以在三维体数据中选择一些备选切面，然后计算这些备选切面两侧的区域的对称性，认为两侧的对称性最好的备选切面即为所需要的正中矢状切面。
例如，本发明的实施例中，根据三维体数据检测正中矢状面的流程如图23所示，包括：选择备选切面的步骤120、计算备选切面的对称性指数的步骤121、以及确定满足条件的对称性指数的切面的步骤122。
在步骤120中，可以在三维体数据中选择一组备选切面。备选切面的选择方式可以根据需要确定。例如，可以选择三维体数据中一定范围内在一个或者多个特定方向上相距一定的间隔(或者步长)的所有切面。这里，所说的“一定范围”可以是相对于三维体数据中的一个或者多个 线和/或面的角度范围，也可以是相对于三维体数据中的一个或者多个点、线和或面的距离的范围；所说的“在一个或者多个方向上”是指切面的法线在该一个或者多个方向上；所说的“间隔”或者“步长”可以是距离间隔或者步长，也可以是角度间隔或者步长。
在本发明的实施例中，可以是选择三维体数据的全部范围内在一个或者多个方向上相距一定的间隔或者步长的全部切面；或者，一种实施例中，也可以根据一些先验知识来选择备选切面，去除正中矢状面明显不可能包含于其中的备选切面。例如，由于胎儿头部的正中矢状面是位于胎儿头部中间位置的纵切面(即在从从三维体数据中胎儿头顶部分到胎儿颈部部分的方向上的切面)，因此根据三维体数据中胎儿图像的大致方向，选择大体上在头部中间位置的纵切面作为备选切面。本文中，将三维体数据中或者三维体数据的至少一部分中在从从三维体数据中胎儿头顶部分到胎儿颈部部分的方向上的切面(也就是大体上平行于从三维体数据中胎儿头顶部分到胎儿颈部部分的方向的切面，或者其法线大体上垂直于从从三维体数据中胎儿头顶部分到胎儿颈部部分的方向的切面)称为该三维体数据的“纵切面”。
因此，本发明的实施例中，可以选择三维体数据中的一组纵切面作为前述的一组备选切面，例如，选择大体上在头部中间位置的一组纵切面(例如，头部中间位置特定区域内以特定步长或者间距的所有纵切面)作为该一组备选切面。
或者，本发明的实施例中，也可以接收用户的输入，该用户输入指示正中矢状切面所处的可能的范围，然后选择这个用户指出的范围内的切面作为备选切面。
本实施例中，也可以选择三维体数据中相距一定步长的所有切面，即以一定的步长遍历搜索三维体数据的全部范围内的所有切面。
例如，一个实施例中，当采用式(6)表示切面方程时，选定其中的平面参数θ、
Figure PCTCN2015078494-appb-000041
ρ的取值范围，并且选定步长θstep、
Figure PCTCN2015078494-appb-000042
ρstep的值，即实现了对备选切面的选择。
类似地，当采用一般表达式aX+bY+cZ+d＝0或Z＝aX+bY+c或Y＝aX+bZ+c表示切面方程时，选定其中的平面参数a、b、c和d的范围和各自的步长，即实现了对备选切面的选择。
例如，当选择的三维体数据中的以一定步长的所有切面为备选切面 时，参数ρ的取值范围如式(7)所示，θ、
Figure PCTCN2015078494-appb-000043
最大的取值范围例如可以为0°≤θ<360°，
Figure PCTCN2015078494-appb-000044
(参考图16)。容易理解，当坐标系设置方式不同时，参数的取值范围也会变化。
步长θstep、
Figure PCTCN2015078494-appb-000045
ρstep可以根据实际需要的检测精度来定，本发明对此没有限制。例如，一个实施例中，可以是θstep＝1、
Figure PCTCN2015078494-appb-000046
ρstep＝2。容易理解，根据需要的检测精度，步长也可以设置为任意其它的值。
选定了备选切面之后，在步骤121中，即可以前述的步长遍历平面参数取值范围内的所有的备选切面方程θ、
Figure PCTCN2015078494-appb-000047
ρ，计算每个备选切面的对称性指数。
对称性指数主要用于衡量备选切面两侧的数据的相似性。
因此，例如，一个实施例中，对于每个备选切面，可以在三维体数据中该备选切面的两侧选择至少一对第一区域和第二区域，并且该第一区域和该第二区域关于该备选切面对称，然后用第一区域中的数据和第二区域中的数据来计算该备选切面的对称性指数。
这里，“第一区域中的数据”是指落入第一区域中的三维体数据中的数据点的值，类似地，“第二区域中的数据”是指落入第二区域中的三维体数据中的数据点的值。
本发明的实施例中，对于每个备选切面，也可以选择多对第一区域和第二区域，对于每对第一区域和第二区域分别计算对称性指数，然后根据多对第一区域和第二区域对应的多个对称性指数，获得最终的该备选切面的对称性指数。例如，将该多个对称性指数取平均值作为对应的备选切面的对称性指数；或者将该多个对称性指数的加权平均值作为对应的备选切面的对称性指数，其中加权系数可以根据选择的第一区域和第二区域对的位置或者其它性质确定；等等。本发明的实施例中，最终的该备选切面的对称性指数可以是分别根据多对第一区域和第二区域计算获得的对称性指数的函数。
对称性指数可以用多种方法计算。
例如，一个实施例中，可以用前述的第一区域和第二区域中对应的点的灰度值的差的绝对值之和作为对称性指数，即：
Figure PCTCN2015078494-appb-000048
其中，E为对称性指数，Ω为平面两侧选定的对称的第一区域和第 二区域，IL为第一区域中的点的数据值，IR为与第一区域中的点关于备选切面对称的第二区域中的点的数据值。这里，“第一区域和第二区域中对应的点”是指第一区域和第二区域中关于备选切面对称的点。
本发明的实施例中，备选切面的对称性指数也可以是前述的第一区域和第二区域的相关系数，即：
Figure PCTCN2015078494-appb-000049
其中，E为对称性指数，Ω为平面两侧选定的对称的第一区域和第二区域，IL为第一区域中的点的数据值，IR为与第一区域中的点关于备选切面对称的第二区域中的点的数据值。
对称性指数的定义包括但不局限于上述两种方法，也可以使用其它类似的定义，如第一区域和第二区域的欧式距离(Euclidean Distance)，第一区域和第二区域的余弦相似度(Cosine Similarity)等等。
对于所有备选切面，均计算出其对称性指数，即可获得一组对称性指数。然后，从该组对称性指数中选择满足特征条件的特征对称性指数，本发明的实施例中，认为该特征对称性指数所对应的备选切面即为所需要的胎儿头部的正中矢状切面。
这里，所说的“特征条件”可以是表明备选切面的对称性最优的条件。该特征条件可以根据对称性指数的计算方法的不同而不同。例如，对于前述按照式(12)计算的对称性指数，可以看出，E值(即对称性指数)越小，说明备选切面两侧的图像像素越相似，即对称性越好，因此，此时，特征条件可以是“对称性指数最小”。而对于前述按照式(13)计算的对称性指数，E值(即相似性指数)越大(对于式(13)，也就是E值越接近1)，说明备选切面两侧的图像像素越相似，即对称性越好，因此，此时，特征条件可以是“对称性指数最接近1”或者“对称性指数最大”。
当按照其它方式计算对称性指数时，特征条件也可以类似地定义。例如，当对称性指数为第一区域和第二区域的欧式距离时，特征条件可以为“对称性指数最小”，即此时对称性指数越小(即欧式距离越小)，则第一区域和第二区域对称性越好；当对称性指数为第一区域和第二区 域的余弦相似度时，特征条件可以为“对称性指数最大”，即此时对称性指数越大(即余弦相似度越大)，则第一区域和第二区域对称性越好；等等。
采用本实施例的方法检测出了胎儿脑部的正中矢状切面之后，类似前述实施例在判断胎儿头部朝向和/或脸部朝向后，根据判断结果将倒立的图像调整为正立和/或标识出脸部朝向，然后送入显示器上显示。
如前文所述，胎儿头部的正中矢状切面中将表现一些特定的结构，也就是说，胎儿头部的正中矢状切面的图像将具有一些特有的结构特征。因此，本发明的实施例也可以利用胎儿头部的正中矢状切面的这个特征，通过先前已经获得的其它的胎儿头部的正式矢状切面的图像，生成胎儿头部正中矢状面的模板图像(或者标准参考图像)，然后在三维成像过程中，将获得的三维体数据中的切面与该模板图像进行匹配，计算三维体数据中的切面与目标图像的相似度，认为三维体数据中与模板图像相似度最高的切面即为胎儿头部的正中矢状切面。
例如，本发明的实施例中，根据三维体数据检测正中矢状面的流程如图24所示，包括：获取模板图像的步骤130、选择备选切面的步骤131、计算备选切面与模板图像的相似度的步骤132、以及确定满足条件的相似度的切面的步骤133。
在步骤130中，可以获取胎儿头部的正中矢状面的模板图像。本发明的实施例中，这个模板图像可以是预先根据先前已经获得的其它的胎儿头部的正中矢状切面的图像生成并且存储在存储设备中，本发明实施例的三维成像过程中，直接从该存储设备中读取获得；也可以是在本发明实施例的三维成像过程中生成。
本发明的实施例中，模板图像可以是一个，也可以是多个，例如，这里的多个模板可以是每个用于与不同尺寸的三维体数据中的切面匹配。
当是多个模板图像时，后续的匹配过程中可以用每个备选切面与每个模板图像匹配。
获得了模板图像后，在步骤131中，即可在三维体数据中选择一组备选切面。备选切面的选择方式可以根据需要确定。例如，可以选择三维体数据中一定范围内在一个或者多个特定方向上相距一定的间隔(或者步长)的所有切面。这里，所说的“一定范围”可以是相对于三维体 数据中的一个或者多个线和/或面的角度范围，也可以是相对于三维体数据中的一个或者多个点、线和或面的距离的范围；所说的“在一个或者多个方向上”是指切面的法线在该一个或者多个方向上；所说的“间隔”或者“步长”可以是距离间隔或者步长，也可以是角度间隔或者步长。
本发明的实施例中，可以是选择三维体数据的全部范围内在一个或者多个方向上相距一定的间隔或者步长的全部切面；或者，本实施例中，也可以根据一些先验知识来选择备选切面，去除正中矢状面明显不可能包含于其中的备选切面。例如，由于胎儿头部的正中矢状面是位于胎儿头部中间位置的纵切面(即在从头顶部分到颈部部分的方向上的切面)，因此根据三维体数据中胎儿图像的大致方向，可以选择三维体数据中的一组纵切面作为前述的一组备选切面，例如，选择大体上在头部中间位置的一组纵切面(例如头部中间位置特定区域内以特定步长或者间距的所有纵切面)作为该组备选切面。
或者，本发明的实施例中，也可以接收用户的输入，该用户输入指示正中矢状切面所处的可能的范围，然后选择这个用户指出的范围内的切面作为备选切面。
本发明的实施例中，也可以选择三维体数据中相距一定步长的所有切面，即以一定的步长遍历匹配三维体数据的全部范围内的所有切面与模板图像。
例如，一个实施例中，当采用式(6)表示切面方程时，选定其中的平面参数θ、
Figure PCTCN2015078494-appb-000050
ρ的取值范围，并且选定步长θstep、
Figure PCTCN2015078494-appb-000051
ρstep的值，即实现了对备选切面的选择。
类似地，当采用一般表达式aX+bY+cZ+d＝0或Z＝aX+bY+c或Y＝aX+bZ+c表示切面方程时，选定其中的平面参数a、b、c和d的范围和各自的步长，即实现了对备选切面的选择。
例如，当选择的三维体数据中的以一定步长的所有切面为备选切面时，参数ρ的取值范围如式(7)所示，θ、
Figure PCTCN2015078494-appb-000052
最大的取值范围例如可以为0°≤θ<360°，
Figure PCTCN2015078494-appb-000053
(参考图16)。容易理解，当坐标系设置方式不同时，参数的取值范围也会变化。
步长θstep、
Figure PCTCN2015078494-appb-000054
ρstep可以根据实际需要的检测精度来定，本发明对此没有限制。例如，一个实施例中，可以是θstep＝1、
Figure PCTCN2015078494-appb-000055
ρstep＝2。容易理解，根据需要的检测精度，步长也可以设置为任意其它的值。
如前文所述，本发明的实施例中，模板图像也可以是一个，在这种情况下，模板图像在一个特定的尺寸下生成。此时，在三维体数据中选择备选切面之前，还包括将三维体数据与模板图像对齐的步骤，该步骤是将三维体数据和模板图像对齐到同一尺度空间，也就是说使得三维体数据中和模板图像中的各个结构的大小是近似一致的。经过这种对齐，三维体数据与模板图像中的各个结构具有近似一致的尺寸，从而后续匹配过程更容易实现，匹配效果更好，并且减小了匹配过程的计算量。
使三维体数据与模板图像对齐的方法可以是检测三维体数据中的切面图像(例如，可以取最中间一帧图像，即第F/2帧的图像，也可取其临近的帧或者其它的帧的图像或者其它的切面图像)中的特定结构特征(例如颅骨光环，等的)，然后根据检测到的特定结构特征的大小将三维体数据通过旋转、平移和/或缩放等方式变换到与模板图像的尺寸相同的尺寸水平。
这里，将三维体数据变换到与模板图像的尺寸相同的尺寸水平是指通过变换使三维图数据中和模板图像中相同或者相应的结构特征具有相同的尺寸。
这里，“相同”是指大体上或者基本上相同或者相近，不是严格限制是绝对地相同，而是可以允许有一定的差别。也就是说，这里的“相同”不应该严格地理想化地解释。
本发明的实施例中，也可以使用任何其它适合的方法将三维体数据与模板图像对齐到同一尺度空间。
这里，“同一”是指大体上或者基本上相同或者相近，不是严格限制是绝对地相同，而是可以允许有一定的差别。也就是说，这里的“同一”不应该严格地理想化地解释。
选定了备选切面之后，在步骤132中，即可以前述的步长遍历平面参数取值范围内的所有的备选切面方程θ、
Figure PCTCN2015078494-appb-000056
ρ，匹配每个备选切面和前述的模板图像。例如，可以计算每个备选切面和模板图像的相似性指数。
相似性指数用于衡量备选切面与模板图像的相似性。本发明的实施例中，相似性指数可以用多种方式计算。
例如，一个实施例中，相似性指数可以是备选切面和模板图像中对应的点的灰度值的差的绝对值之和，即：
Figure PCTCN2015078494-appb-000057
其中，E为相似性指数，Ω为备选切面的图像空间，IL为备选切面的点的数据值，IR为与备选切面中的点对应的模板图像中的点的数据值。这里，“备选切面和模板图像中对应的点”是指在备选切面中和在模板图像中具有相同位置的点。
或者，本发明另外的实施例中，相似性指数也可以是备选切面和模板图像的相关系数，即：
Figure PCTCN2015078494-appb-000058
其中，E为相似性指数，Ω为备选切面的图像空间，IL为备选切面的点的数据值，IR为与备选切面中的点对应的模板图像中的点的数据值。
相似性指数的定义包括但不局限于上述两种方法，也可以使用其它类似的定义。
对于所有备选切面，均计算出其相似性指数，即可获得一组相似性指数。然后，从该组相似性指数中选择满足特征条件的特征相似性指数，本发明的实施例中，认为该特征相似性指数所对应的备选切面即为所需要的胎儿头部的正中矢状切面。
这里，所说的“特征条件”可以是表明备选切面与模板图像的相似性最优的条件。该特征条件可以根据相似性指数的计算方法的不同而不同。
例如，对于前述按照式(14)计算的相似性指数，可以看出，E值(即相似性指数)越小，说明备选切面与模板图像的图像像素越相似，即相似性越好，因此，此时，特征条件可以是“相似性指数最小”。
而对于前述按照式(15)计算的相似性指数，E值(即相似性指数)越大(对于式(15)，也就是E值越接近1)，说明备选切面与模板图像的图像像素越相似，即相似性越好，因此，此时，特征条件可以是“相似性指数最大”或者“相似性指数最接近1”。
当按照其它方式计算相似性指数时，特征条件也可以类似地定义。例如，当相似性指数为备选切面和模板图像的欧式距离时，特征条件可以为“相似性指数最小”，即此时相似性指数越小(即欧式距离越小)， 则备选切面和模板图像越相似；当相似性指数为备选切面和模板图像的余弦相似度时，特征条件可以为“对称性指数最大”，即此时相似性指数越大(即余弦相似度越大)，则备选切面和模板图像越相似；等等。
采用本实施例的方法检测出了胎儿脑部的正中矢状切面之后，类似前述实施例在判断胎儿头部的朝向方位(例如，头顶朝向和/或脸部朝向)后，根据判断结果将图像调整为期望的方位和/或标识出胎儿头部的朝向，然后送入显示器上显示。
本发明的一些实施例中，检测胎儿脑部的正中矢状切面的方法不限于前述实施例中的方法，也可以使用任何其他适合的检测胎儿脑部的正中矢状切面的方法。
应该理解，对于本发明各个实施例中的“朝上”、“朝下”、“前”、“后”、“左”、“右”等位置关系术语，其中“朝上”和“朝下”是按照人的观察习惯而言分别对应头部在图像中为正立和倒立，“前”和“后”是按照人的观察习惯而言分别对应胎儿的脸部和背部，“左”和“右”是具有相对性的，可以将“左”和“前”对应，也可以将“左”和“后”对应，不能认为其是对本发明的限制。
本领域技术人员可以理解，上述实施方式中各种方法的全部或部分步骤可以通过程序来指令相关硬件完成，该程序可以存储于一计算机可读存储介质中，存储介质可以包括：只读存储器、随机存储器、磁盘或光盘等。
以上内容是结合具体的实施方式对本发明所作的进一步详细说明，不能认定本发明的具体实施只局限于这些说明。对于本发明所属技术领域的普通技术人员来说，在不脱离本发明构思的前提下，还可以做出若干简单推演或替换。

Claims (10)

  1. 一种三维超声成像方法，其特征在于，包括：
    向胎儿头部发射超声波；
    接收超声回波，获得超声回波信号；
    根据所述超声回波信号获得胎儿头部的三维体数据；
    根据胎儿头部正中矢状切面的特征，从所述三维体数据中检测正中矢状切面；
    检测所述正中矢状切面中和/或与所述正中矢状切面平行或者相交的切面中表示胎儿头部中的特定组织区域的图像区域；
    根据所述图像区域确定所述正中矢状切面中胎儿头部的朝向方位；
    根据所述胎儿头部的朝向方位，旋转所述正中矢状切面，使得旋转后所述正中矢状切面中的胎儿头部在预定方位，或者在所述正中矢状切面上标识胎儿头部的朝向方位。
  2. 如权利要求1所述的方法，其特征在于，检测所述正中矢状切面中和/或与所述正中矢状切面平行或者相交的切面中表示胎儿头部中的特定组织区域的图像区域并根据所述图像区域确定正中矢状切面中胎儿头部的朝向方位的步骤包括：
    检测所述正中矢状切面中和/或与所述正中矢状切面平行或者相交的切面中表示胎儿头部中具有特定相互位置关系的至少两个特定组织区域的至少两个图像区域，并根据所述至少两个图像区域之间的相互位置关系确定胎儿头部的朝向方位。
  3. 如权利要求1所述的方法，其特征在于，检测所述正中矢状切面中和/或与所述正中矢状切面平行或者相交的切面中表示胎儿头部中的特定组织区域的图像区域并根据所述图像区域确定正中矢状切面中胎儿头部的朝向方位的步骤包括：
    检测所述正中矢状切面中和/或与所述正中矢状切面平行或者相交的切面中表示胎儿头部中具有方向特性的特定组织区域的图像区域，并根据所述图像区域的位置和/或形状特征确定胎儿头部的朝向方位。
  4. 如权利要求3所述的方法，其特征在于，所述图像区域为表示胎儿头部中的颅骨的区域，其中检测所述正中矢状切面中和/或与所述正中矢状切面平行或者相交的切面中表示胎儿头部中具有方向特性的特定组织区域的图像区域并根据所述图像区域确定正中矢状切面中胎 儿头部的朝向方位的步骤包括：
    选取正中矢状切面和/或至少一个与所述正中矢状切面平行的切面，根据颅骨在超声图像中的特点，从选取出的切面中提取颅骨对应的特征区域，获得表征颅骨的候选区域；
    基于所述候选区域确定颅骨的朝向；
    根据颅骨的朝向确定在所述正中矢状切面中胎儿头部的朝向方位。
  5. 如权利要求4所述的方法，其特征在于，
    所述提取颅骨对应的特征区域的步骤包括：在所述选取出的切面中提取满足灰度值大于预设灰度阈值的区域作为特征区域；或者基于微分法设计算子，将设计的算子和所述选取出的切面对应的图像作卷积，选取卷积值大于预设灰度阈值的区域作为特征区域；
    所述获得表征颅骨的候选区域的步骤包括：选择特征值之和最大的至少一个特征区域作为候选区域；或者，选择平均特征值最大的至少一个特征区域作为候选区域；或者，从所述特征区域中提取特征，将提取出的特征输入预先训练好的分类器进行分类，分类结果作为候选区域；
    所述判断颅骨朝向的步骤包括：根据颅骨的弯曲方向确定颅骨朝向。
  6. 如权利要求3所述的方法，其特征在于，所述图像区域为表示胎儿头部中的透明隔腔的区域，其中检测所述正中矢状切面中和/或与所述正中矢状切面平行或者相交的切面中表示胎儿头部中具有方向特性的特定组织区域的图像区域并根据所述图像区域确定正中矢状切面中胎儿头部的朝向方位的步骤包括：
    根据透明隔腔在超声图像中的特点，从所述正中矢状切面中检测出透明隔腔对应的联通区域；
    基于所述联通区域确定透明隔腔的朝向；
    根据透明隔腔的朝向确定在所述正中矢状切面中胎儿头部的朝向方位。
  7. 如权利要求6所述的方法，其特征在于，
    所述从所述正中矢状切面中检测出透明隔腔对应的联通区域的步骤包括：对所述正中矢状切面的图像采用图像分割算法进行处理，得到至少一个区域，结合所述区域的特征从所述区域中选择与透明隔腔最相似的区域，选择结果为透明隔腔对应的联通区域。
  8. 如权利要求3所述的方法，其特征在于，所述图像区域为表示 胎儿头部中的透明隔腔的区域，其中检测所述正中矢状切面中和/或与所述正中矢状切面平行或者相交的切面中表示胎儿头部中具有方向特性的特定组织区域的图像区域并根据所述图像区域确定正中矢状切面中胎儿头部的朝向方位的步骤包括：
    根据透明隔腔在超声图像中的特点，从所述正中矢状切面中检测出透明隔腔区域；
    从所述三维体数据中选取至少一帧水平剖面图，对所述水平剖面图进行椭圆或椭球体检测，并获得所述椭圆或椭球体的中心；
    将所述透明隔腔区域的坐标信息与所述中心的坐标信息进行比较，根据比较结果得到所述正中矢状切面中胎儿头部的朝向方位。
  9. 如权利要求8所述的方法，其特征在于，
    所述从所述正中矢状切面中检测出透明隔腔区域的步骤包括：对所述正中矢状切面的图像采用图像分割算法进行处理，得到至少一个区域，结合所述区域的特征从所述区域中选择与透明隔腔最相似的区域，选择结果为透明隔腔区域；
    所述对所述水平剖面图进行椭圆或者椭球体检测的步骤包括：基于微分法设计算子，利用所述算子对所述水平剖面图进行特征提取，对提取到的特征进行椭圆或椭球体检测。
  10. 一种三维超声成像装置，其特征在于，包括：
    探头，用于向胎儿头部发射超声波并接收超声回波，获得超声回波信号；
    三维成像模块，用于根据所述超声回波信号获得胎儿头部的三维体数据，根据胎儿头部正中矢状切面的特征，从所述三维体数据中检测正中矢状切面，并确定所述正中矢状切面中胎儿头部的朝向方位，并根据所述胎儿头部的朝向方位，旋转所述正中矢状切面，使得旋转后所述正中矢状切面中的胎儿头部在预定方位，或者在所述正中矢状切面上标识胎儿头部的朝向方位；
    显示器，用于显示所述正中矢状切面。
PCT/CN2015/078494 2015-05-07 2015-05-07 三维超声成像方法和装置 WO2016176863A1 (zh)

Priority Applications (5)

Application Number Priority Date Filing Date Title
PCT/CN2015/078494 WO2016176863A1 (zh) 2015-05-07 2015-05-07 三维超声成像方法和装置
CN201580071177.3A CN107106143B (zh) 2015-05-07 2015-05-07 三维超声成像方法和装置
CN202010990346.6A CN112120736B (zh) 2015-05-07 2015-05-07 三维超声成像方法和装置
US15/800,387 US10702240B2 (en) 2015-05-07 2017-11-01 Three-dimensional ultrasound imaging method and device
US16/903,221 US11534134B2 (en) 2015-05-07 2020-06-16 Three-dimensional ultrasound imaging method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2015/078494 WO2016176863A1 (zh) 2015-05-07 2015-05-07 三维超声成像方法和装置

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US15/800,387 Continuation US10702240B2 (en) 2015-05-07 2017-11-01 Three-dimensional ultrasound imaging method and device

Publications (1)

Publication Number Publication Date
WO2016176863A1 true WO2016176863A1 (zh) 2016-11-10

Family

ID=57217341

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/078494 WO2016176863A1 (zh) 2015-05-07 2015-05-07 三维超声成像方法和装置

Country Status (3)

Country Link
US (2) US10702240B2 (zh)
CN (2) CN107106143B (zh)
WO (1) WO2016176863A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110176066A (zh) * 2019-05-28 2019-08-27 中山大学附属第三医院 颅骨缺损结构的重建方法、装置及电子设备
CN110432929A (zh) * 2019-07-11 2019-11-12 暨南大学 基于超声图像的产时头盆关系自动测量方法和装置
CN112017189A (zh) * 2020-10-26 2020-12-01 腾讯科技（深圳）有限公司 图像分割方法、装置、计算机设备和存储介质
CN112862944A (zh) * 2019-11-09 2021-05-28 无锡祥生医疗科技股份有限公司 人体组织超声建模方法、超声设备及存储介质
US11521363B2 (en) * 2017-05-12 2022-12-06 Shenzhen Mindray Bio-Medical Electronics Co., Ltd. Ultrasonic device, and method and system for transforming display of three-dimensional ultrasonic image thereof
WO2023133935A1 (zh) * 2022-01-14 2023-07-20 汕头市超声仪器研究所股份有限公司 超声颅脑异常区域自动检测及显示方法

Families Citing this family (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107766874B (zh) * 2017-09-07 2021-06-04 深圳度影医疗科技有限公司 一种超声容积生物学参数的测量方法及测量系统
CN114305499A (zh) * 2018-10-19 2022-04-12 深圳迈瑞生物医疗电子股份有限公司 超声成像方法、设备和存储介质
WO2020087532A1 (zh) * 2018-11-02 2020-05-07 深圳迈瑞生物医疗电子股份有限公司 超声成像方法及系统、存储介质、处理器和计算机设备
US10672510B1 (en) * 2018-11-13 2020-06-02 Biosense Webster (Israel) Ltd. Medical user interface
CN111368586B (zh) * 2018-12-25 2021-04-20 深圳迈瑞生物医疗电子股份有限公司 超声成像方法及系统
EP3730060A1 (en) * 2019-04-24 2020-10-28 Koninklijke Philips N.V. Fetal ultrasound processing unit for separating heart rate signals
CN111839515B (zh) * 2019-04-26 2024-03-08 西门子医疗系统有限公司 利用磁共振成像的胎儿胼胝体体积的测量方法及磁共振成像装置
US11830192B2 (en) 2019-07-31 2023-11-28 The Joan and Irwin Jacobs Technion-Cornell Institute System and method for region detection in tissue sections using image registration
CN110464380B (zh) * 2019-09-12 2021-10-29 李肯立 一种对中晚孕期胎儿的超声切面图像进行质量控制的方法
CN111462055B (zh) * 2020-03-19 2024-03-08 东软医疗系统股份有限公司 颅骨检测方法及装置
CN112258533B (zh) * 2020-10-26 2024-02-02 大连理工大学 一种超声图像中小脑蚓部的分割方法
CN112370078B (zh) * 2020-11-10 2024-01-26 安徽理工大学 一种基于超声成像和贝叶斯优化的图像检测方法
WO2022099705A1 (zh) * 2020-11-16 2022-05-19 深圳迈瑞生物医疗电子股份有限公司 早孕期胎儿的超声成像方法和超声成像系统
CN116568223A (zh) * 2020-12-25 2023-08-08 深圳迈瑞生物医疗电子股份有限公司 胎儿颅骨的超声成像方法和超声成像系统
KR20240032563A (ko) * 2022-09-02 2024-03-12 삼성메디슨 주식회사 초음파 진단 장치 및 진단 방법

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH04295346A (ja) * 1991-03-22 1992-10-20 Hitachi Medical Corp 超音波断層装置
CN101069647A (zh) * 2006-05-09 2007-11-14 株式会社东芝 超声波图像取得装置及超声波图像显示方法
US20110125016A1 (en) * 2009-11-25 2011-05-26 Siemens Medical Solutions Usa, Inc. Fetal rendering in medical diagnostic ultrasound
US20110224546A1 (en) * 2010-03-10 2011-09-15 Medison Co., Ltd. Three-dimensional (3d) ultrasound system for scanning object inside human body and method for operating 3d ultrasound system
CN104114096A (zh) * 2012-11-27 2014-10-22 株式会社东芝 医用图像诊断装置、医用图像处理装置以及医用图像处理方法
CN104414680A (zh) * 2013-08-21 2015-03-18 深圳迈瑞生物医疗电子股份有限公司 一种三维超声成像方法及系统

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
MXPA05011120A (es) * 2003-04-16 2005-12-15 Eastern Viriginai Medical Scho Sistema y metodo para generar imagenes de ultrasonido independientes del operador.
KR101229490B1 (ko) * 2010-05-31 2013-02-04 삼성메디슨 주식회사 ３차원 초음파 검사기 및 ３차원 초음파 검사기의 동작 방법
KR101194290B1 (ko) * 2010-09-24 2012-10-29 삼성메디슨 주식회사 이미지 필터링을 이용한 3차원 초음파 검사기 및 3차원 초음파 검사기의 동작 방법
CN102949206B (zh) * 2011-08-26 2015-12-02 深圳迈瑞生物医疗电子股份有限公司 一种三维超声成像的方法及装置
KR20130072810A (ko) * 2011-12-22 2013-07-02 삼성전자주식회사 초음파 영상을 이용하여 정중 시상면을 자동으로 검출하는 방법 및 그 장치
CN104156967B (zh) * 2014-08-18 2017-09-08 深圳开立生物医疗科技股份有限公司 一种胎儿颈部透明层图像分割方法、装置及超声成像系统

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH04295346A (ja) * 1991-03-22 1992-10-20 Hitachi Medical Corp 超音波断層装置
CN101069647A (zh) * 2006-05-09 2007-11-14 株式会社东芝 超声波图像取得装置及超声波图像显示方法
US20110125016A1 (en) * 2009-11-25 2011-05-26 Siemens Medical Solutions Usa, Inc. Fetal rendering in medical diagnostic ultrasound
US20110224546A1 (en) * 2010-03-10 2011-09-15 Medison Co., Ltd. Three-dimensional (3d) ultrasound system for scanning object inside human body and method for operating 3d ultrasound system
CN104114096A (zh) * 2012-11-27 2014-10-22 株式会社东芝 医用图像诊断装置、医用图像处理装置以及医用图像处理方法
CN104414680A (zh) * 2013-08-21 2015-03-18 深圳迈瑞生物医疗电子股份有限公司 一种三维超声成像方法及系统

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11521363B2 (en) * 2017-05-12 2022-12-06 Shenzhen Mindray Bio-Medical Electronics Co., Ltd. Ultrasonic device, and method and system for transforming display of three-dimensional ultrasonic image thereof
CN110176066A (zh) * 2019-05-28 2019-08-27 中山大学附属第三医院 颅骨缺损结构的重建方法、装置及电子设备
CN110432929A (zh) * 2019-07-11 2019-11-12 暨南大学 基于超声图像的产时头盆关系自动测量方法和装置
CN112862944A (zh) * 2019-11-09 2021-05-28 无锡祥生医疗科技股份有限公司 人体组织超声建模方法、超声设备及存储介质
CN112862944B (zh) * 2019-11-09 2024-04-12 无锡祥生医疗科技股份有限公司 人体组织超声建模方法、超声设备及存储介质
CN112017189A (zh) * 2020-10-26 2020-12-01 腾讯科技（深圳）有限公司 图像分割方法、装置、计算机设备和存储介质
WO2023133935A1 (zh) * 2022-01-14 2023-07-20 汕头市超声仪器研究所股份有限公司 超声颅脑异常区域自动检测及显示方法

Also Published As

Publication number Publication date
CN112120736B (zh) 2023-04-18
CN107106143B (zh) 2020-10-20
US20180185003A1 (en) 2018-07-05
CN112120736A (zh) 2020-12-25
US20200383659A1 (en) 2020-12-10
US11534134B2 (en) 2022-12-27
CN107106143A (zh) 2017-08-29
US10702240B2 (en) 2020-07-07

Similar Documents

Publication Publication Date Title
WO2016176863A1 (zh) 三维超声成像方法和装置
US11969288B2 (en) Three-dimensional ultrasonic imaging method and system
CN106102585B (zh) 三维成像数据的显示处理方法和三维超声成像方法及系统
Rueda et al. Evaluation and comparison of current fetal ultrasound image segmentation methods for biometric measurements: a grand challenge
CN111368586B (zh) 超声成像方法及系统
US20060235301A1 (en) 3D ultrasound-based instrument for non-invasive measurement of amniotic fluid volume
EP3432803A1 (en) Ultrasound system and method for detecting lung sliding
CN111374712B (zh) 一种超声成像方法及超声成像设备
KR20210014267A (ko) 초음파 이미지상의 특징점들을 이용한 초음파 지방간 자동 진단 장치 및 이를 이용한 원격 의료 진단 방법
Cuingnet et al. Where is my baby? A fast fetal head auto-alignment in 3D-ultrasound
US20220249060A1 (en) Method for processing 3d image data and 3d ultrasonic imaging method and system
WO2020133236A1 (zh) 一种脊柱的成像方法以及超声成像系统
Challenge Evaluation and comparison of current fetal ultrasound image segmentation methods for biometric measurements: a grand challenge
US20230301621A1 (en) Systems and methods for detection of traumatic brain injury using combined 3d computational modeling and elastography
Khazendar Computer-aided diagnosis of gynaecological abnormality using B-mode ultrasound images
JP6761365B2 (ja) 超音波診断装置及びプログラム
CN115886876A (zh) 胎儿姿态的评估方法、超声成像方法及超声成像系统
Rahmatullah Assessment of obstetric ultrasound images using machine learning

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: 15891115

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 15891115

Country of ref document: EP

Kind code of ref document: A1