WO2018095058A1

WO2018095058A1 - Three-dimensional ultrasonic fetal face profile image processing method and system

Info

Publication number: WO2018095058A1
Application number: PCT/CN2017/093457
Authority: WO
Inventors: 黄柳倩; 孙慧; 艾金钦; 刘旭江; 喻美媛
Original assignee: 深圳开立生物医疗科技股份有限公司
Priority date: 2016-11-22
Filing date: 2017-07-19
Publication date: 2018-05-31
Also published as: CN106725593B; CN106725593A

Abstract

A three-dimensional ultrasonic fetal face profile image processing method and system. The method comprises: S101: detecting a multi-frame slice of fetal volume data in a predetermined direction to acquire a target area in each frame of the slice, the target area comprising a fetal head area; S102: screening out the slice comprising the target area and performing facial boundary detection on the screened-out slice to obtain trusted boundary points; and S103: cropping the fetal volume data according to the trusted boundary points to obtain a three-dimensional ultrasonic fetal face profile image. According to the image processing method, the volume data is cropped by using the trusted boundary points so that a shaded part of the fetal face can be automatically cut out; and thus the operation of a detector is simplified and three-dimensional ultrasonic imaging rate is improved.

Description

Ultrasonic three-dimensional fetal facial contour image processing method and system

The present application claims priority to Chinese Patent Application No. 201611055976.4, the entire disclosure of which is incorporated herein in .

Technical field

The present invention relates to the field of ultrasonic imaging technology, and in particular, to an ultrasonic three-dimensional fetal facial contour image processing method and system.

Background technique

The three-dimensional ultrasound imaging system is based on the traditional two-dimensional ultrasound, and acquires two-dimensional ultrasonic images of multiple sequential spatial sequences. According to the spatial positional relationship of the data acquisition, the volume data is reproduced by steps such as scanning conversion. The three-dimensional ultrasound imaging system provides high-dimensional spatial information that traditional two-dimensional ultrasound can not provide, making clinical diagnosis and observation more intuitive and flexible, and making the communication between the doctors and patients more smooth. Due to its informative and intuitive information, the three-dimensional ultrasound imaging system is currently mainly used for observation of fetal morphology in the field of obstetrics, especially for facial observation. However, due to the particularity of the imaging environment, when the fetal face is three-dimensionally visualized, the front of the fetal face may be blocked by the placenta, umbilical cord, arm, uterine wall, etc., so that the collected three-dimensional volume data may include placenta and amniocentesis. Substances, umbilical cords, uterine tissue, etc., cause obstruction to the imaging target, which brings difficulties to the target observation.

Although the current volumetric cutting method usually has an interactive volume cutting method for the inspector to cut the shielding portion, the operation is cumbersome and takes a long time.

Summary of the invention

The invention provides an ultrasonic three-dimensional fetal facial contour image processing method and system, which automatically cuts the occlusion portion of the fetal face by cutting the volume data by using the trusted boundary point, simplifies the operation of the detecting personnel, and improves the three-dimensional imaging rate of the ultrasound.

To achieve the above design, the present invention adopts the following technical solutions:

In one aspect, an ultrasonic three-dimensional fetal facial contour image processing method is provided, the method comprising:

Detecting a plurality of frame slices of the fetal volume data in a predetermined direction to obtain a target region of each frame slice, wherein the target region includes a fetal head region;

Screening the slice containing the target area, and performing face boundary detection on the selected slice to obtain a trusted boundary point;

The fetal volume data is cropped according to the trusted boundary point to obtain an ultrasound three-dimensional fetal facial contour image.

Wherein, after the step of filtering out the slice including the target area, the method further includes:

When the difference between the position of the current frame slice target area and the position of the adjacent slice target area exceeds a predetermined threshold, the position of the current frame slice target area is corrected.

The step of correcting the location of the current frame slice target area includes:

Traversing all of the selected slices to obtain a sequence of <frame number, target area>;

Deviating the position of the target area of each frame slice in the <frame number, target area> sequence from the position of the adjacent slice target area, respectively;

Calculating the mean of the deviations;

When the deviation of the position of the current frame slice target area from the position of the adjacent slice target area is greater than the mean value, the center point coordinate of the current frame slice target area is replaced with the center point coordinate of the target slice target area.

The step of performing facial boundary detection on the selected slice to obtain a trusted boundary point includes:

Obtaining a transition boundary from the darker region to the brighter region of the selected slice in each frame;

Determining a face region of the sliced slice of each frame according to a gray value and a contour shape of the connected region where the transition boundary is located, and using an upper surface boundary of the face region as an alternative segmentation boundary;

Obtaining a plurality of trusted boundary points according to the alternative segmentation boundary.

Wherein, before the step of acquiring a plurality of trusted boundary points according to the candidate segmentation boundary, the method further includes: a step of boundary growth.

The step of acquiring multiple trusted boundary points according to the candidate segmentation boundary includes:

Constructing a boundary moment corresponding to the selected slice of each frame according to the candidate segmentation boundary Array

Superimposing all of the boundary matrix and the accumulation matrix to obtain a voting matrix;

The maximum value of each column in the voting matrix is counted, and the point corresponding to the maximum value is determined as a trusted boundary point.

The step of cropping the fetal volume data according to the trusted boundary point to obtain an ultrasound three-dimensional fetal facial contour image includes:

Making a cropping template according to the trusted boundary point;

The fetal volume data is cropped according to the cutting module to obtain an ultrasound three-dimensional fetal facial contour image.

The step of detecting the multi-frame slice of the fetal volume data in a predetermined direction to obtain the target area of each frame slice includes:

Detecting, by using a preset classifier, the target area of each frame slice of the fetal volume data in a predetermined direction;

The target area and its corresponding slice are saved.

In another aspect, an ultrasound three-dimensional fetal facial contour image processing system is provided, the system comprising:

a target area detecting module, configured to detect a multi-frame slice of the fetal body data in a predetermined direction to obtain a target area of each frame slice, wherein the target area includes a fetal head area;

a trusted boundary point acquiring module, configured to filter out a slice including a target area, and perform face boundary detection on the selected slice to obtain a trusted boundary point;

And a cropping module, configured to crop the fetal volume data according to the trusted boundary point to obtain an ultrasound three-dimensional fetal facial contour image.

The system further includes: a correction module, configured to correct a position of the current frame slice target area when a difference between a position of the current frame slice target area and a position of the adjacent slice target area exceeds a predetermined threshold.

The correction module is further configured to: traverse all the selected slices to obtain a sequence of <frame number, target area>; respectively obtain a position of a frame target area of each frame in the <frame number, target area> sequence and a neighboring slice thereof Deviation of the position of the target area; calculating the mean of the deviation; when the current frame slice target area When the deviation of the position of the domain from the position of the adjacent slice target area is greater than the mean value, the center point coordinate of the current frame slice target area is replaced with the center point coordinate of the target slice target area.

The trusted boundary point obtaining module is further configured to: obtain a transition boundary from the darker region to the brighter region of the selected slice in each frame; according to the gray value and the contour shape of the connected region where the transition boundary is located, Determining a face region of the sliced slice of each frame, and using an upper surface boundary of the face region as an alternative segmentation boundary; and acquiring a plurality of trusted boundary points according to the candidate segmentation boundary.

Wherein the trusted boundary point acquisition module comprises a boundary growth unit, and the boundary growth unit is used for boundary growth.

The trusted boundary point obtaining module is further configured to: construct a boundary matrix corresponding to the filtered slice in each frame according to the candidate segmentation boundary; and superimpose all the boundary matrix and the accumulation matrix, A voting matrix is obtained; the maximum value of each column in the voting matrix is counted, and the point corresponding to the maximum value is determined as a trusted boundary point.

The cutting module is further configured to: create a cropping template according to the trusted boundary point; and perform cropping on the fetal body data according to the cutting module to obtain an ultrasound three-dimensional fetal facial contour image.

The target area detecting module is further configured to: detect, by using a preset classifier, the target area of each frame slice of the fetal body data in a predetermined direction; and save the target area and the corresponding slice thereof.

Compared with the prior art, the beneficial effects of the present invention are: detecting a multi-frame slice of a fetal body data in a predetermined direction to obtain a target region of each frame slice, wherein the target region includes a fetal head region; A slice including the target area is extracted, and the selected slice is subjected to face boundary detection to obtain a trusted boundary point; the fetal body data is cropped according to the trusted boundary point to obtain an ultrasound three-dimensional fetal facial contour image. The invention automatically cuts the occlusion portion of the fetal face by cutting the volume data by using the trusted boundary point, simplifies the operation of the detecting personnel, and improves the three-dimensional imaging rate of the ultrasound.

DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative work Other drawings may also be obtained in accordance with the teachings of the embodiments of the present invention and the drawings.

1 is a flow chart of a method of an embodiment of an ultrasonic three-dimensional fetal facial contour image processing method provided in an embodiment of the present invention.

2 is a schematic diagram of a multi-frame slice of fetal volume data in a predetermined direction provided in an embodiment of the present invention.

FIG. 3 is a schematic diagram of a process for correcting the position of a current frame slice target area provided in an embodiment of the present invention.

Figure 4a is a schematic illustration of a target area prior to correction in accordance with an embodiment of the present invention.

Figure 4b is a schematic illustration of the target area of Figure 4a after correction.

FIG. 5 is a schematic diagram of a process for performing facial boundary detection on a selected slice to obtain a trusted boundary point according to an embodiment of the present invention.

Figure 6a is a schematic illustration of alternative segmentation boundaries provided in an embodiment of the invention.

Figure 6b is a schematic illustration of the alternate segmentation boundary of Figure 6a after boundary growth.

FIG. 7 is a schematic diagram of a process for acquiring multiple trusted boundary points according to an alternative segmentation boundary according to an embodiment of the present invention.

FIG. 8 is a schematic diagram of a process of cropping fetal body data according to a trusted boundary point to obtain an ultrasound three-dimensional fetal facial contour image according to an embodiment of the present invention.

9 is a cut template provided in an embodiment of the present invention.

Figure 10a is a schematic illustration of a sagittal direction slice prior to trimming in accordance with an embodiment of the present invention.

Figure 10b is a schematic view of the cut of Figure 10a.

11 is a block diagram showing the structure of an embodiment of an ultrasonic three-dimensional fetal facial contour image processing system provided in an embodiment of the present invention.

detailed description

The technical solutions of the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, and it is obvious that the described embodiments are only the present invention. Some embodiments, but not all of the embodiments. Based on the embodiments of the present invention, all of those obtained by those skilled in the art without creative efforts The embodiments thereof are all within the scope of protection of the present invention.

Please refer to FIG. 1 , which is a flowchart of an ultrasonic three-dimensional fetal facial contour image processing method provided in an embodiment of the present invention. As shown in Figure 1, the method includes:

Step S101: detecting a multi-frame slice of the fetal volume data in a predetermined direction to obtain a target area of each frame slice, wherein the target area includes a fetal head area.

The target area of the multi-frame (slice) slice of the fetal volume data in a predetermined direction is detected to acquire a target area of at least one frame slice. In this embodiment, the target area includes a fetal head region, and the predetermined direction may be a facet direction of the fetal volume data, such as a planar direction parallel to the transducer array direction. Usually the 3D/4D imaging examiner uses this plane direction to obtain the sagittal plane of the fetus, and the sagittal plane can also be replaced with a characteristic section such as the coronal plane. The predetermined direction can also be used to traverse the plane direction of the three axes of the volume data. Of course, it can also be a predetermined direction determined by other algorithms, which will not be repeated here.

In this embodiment, the target region detection adopts a Histogram of Oriented Gradient (HOG) feature extraction algorithm and an adboost classifier algorithm. The classifier is trained in advance using the data of the sagittal head region of the fetus, and the classifier is set according to the training result, and the target region of the sagittal plane is automatically positioned on each frame slice of the predetermined direction of the fetal volume data by using a preset classifier. , save the target area and its corresponding slice.

Step S102: Filter out a slice including the target area, and perform face boundary detection on the selected slice to obtain a trusted boundary point.

The multi-frame slice of the fetal volume data in a predetermined direction may not include all the slices in the target area, so it is necessary to screen out the slice containing the target area.

Performing a face boundary detection on a slice including the target area, and using a boundary detection operator to find a transition boundary of each frame slice from a darker region to a brighter region, where the boundary detection operator includes but is not limited to the prewitt operator and Sobel operator.

The gray value and the contour form of the connected region where the transition boundary of each frame slice is located are analyzed, and the gray value and the contour form are compared with the pre-stored face gray value and the face contour form to find the face region of each frame slice. And using the upper surface boundary of the face area as the face boundary, voting for each face boundary point according to the face boundary of each frame slice, and determining the trusted boundary by the point where the face boundary passes the most point.

Step S103: crop the fetal volume data according to the trusted boundary point to obtain an ultrasound fetal facial contour image.

According to the trusted boundary point making module acquired in the above step S102, the fetal body data is cropped according to the template to obtain an image of the ultrasonic three-dimensional fetal facial contour. The automatic cutting of the occlusion portion of the fetal face is realized, which simplifies the operation of the detecting person and improves the three-dimensional imaging rate of the ultrasound.

The ultrasonic fetal facial contour image processing method of the above embodiment detects the multi-frame slice of the fetal volume data in a predetermined direction to acquire a target region of each frame slice; selects a slice including the target region, and performs face on the selected slice Boundary detection to obtain a trusted boundary point; the fetal body data is cropped according to the trusted boundary point to obtain an ultrasound three-dimensional fetal facial contour image. The invention automatically cuts the occlusion portion of the fetal face by cutting the volume data by using the trusted boundary point, so that the operation is simple and quick, the error caused by the manual operation is reduced, and the image quality is higher.

In one embodiment, after the step of filtering out the slice including the target area, the method further includes:

When the difference between the position of the current frame slice target area and the position of the adjacent slice target area exceeds a predetermined threshold, the position of the current target area is corrected.

After filtering out the slice including the target area, it is necessary to correct the position of the target area whose difference from the position of the target area of the adjacent slice reaches a predetermined threshold to increase the accuracy of the fetal face contour detection. The predetermined threshold is not specifically limited herein, and the accuracy requirement may be selected according to an actual application. In the present embodiment, the adjacent slice refers to a slice that is temporally adjacent to the current frame slice. For example, as shown in FIG. 2, on the slice sequence of the Z-axis, slices numbered "1", "3", "5", and "6" can detect a region of interest (ROI). Then, the neighboring ROI of the slice ROI numbered "3" is the ROI of the slice numbered "1" and "numbered 5".

Preferably, in one embodiment, as shown in FIG. 3, the step of correcting the position of the current frame slice target area includes:

Step S301: Traverse all the selected slices to obtain a <frame number, target area> sequence.

Step S302: Find the deviation between the position of each frame slice target area in the <frame number, target area> sequence and the position of the adjacent slice target area.

In this embodiment, the deviation includes a center point of each frame slice target area and a neighboring slice target area thereof. The X coordinate deviation and the Y coordinate deviation of the center point of the domain. The obtained X coordinate deviation and Y coordinate deviation are respectively stored in the <frame number, the X coordinate deviation from the center point of the adjacent slice target area> and <the frame number, and the Y coordinate deviation from the center point of the adjacent slice target area> In the sequence.

Step S303: Calculate the mean value of the deviation.

In the present embodiment, the mean value includes the mean value of the X coordinate deviation and the mean value of the Y coordinate deviation. The <frame number obtained according to the above steps, the X coordinate deviation from the center point of the target area of the adjacent slice> and the <frame number, and the Y coordinate deviation from the center point of the target area of the adjacent slice> are two sequences, respectively, are eliminated. The maximum value of the two deviation sequences is then calculated, and the mean value of the X coordinate deviation and the mean value of the Y coordinate deviation of the two sequences are calculated, that is, the mean value of the deviation between the target region of each frame slice and the target region of the adjacent slice is obtained.

Step S304: When the deviation of the position of the current frame slice target area from the position of the adjacent slice target area is greater than the mean value, the center point coordinate of the current frame slice target area is replaced with the center point coordinate of the target slice target area. The distance between the target slice and the current frame slice is the smallest, and the deviation between the position of the target slice target area and the position of the adjacent slice target area is less than the mean value.

In this embodiment, there are three cases:

(1) When the X coordinate deviation of the center point of the current frame slice target area and the adjacent slice target area is greater than the mean value of the X coordinate deviation, the X coordinate of the center point of the current frame slice target area is replaced with the center of the target slice target area. The X coordinate of the point.

(2) When the Y coordinate deviation of the center point of the current frame slice target area and the adjacent slice target area is greater than the mean value of the Y coordinate deviation, the Y coordinate of the current frame slice target area center point is replaced with the target slice target area center point. Y coordinate.

(3) When the X coordinate deviation of the target area of the current frame slice and the center point of the adjacent slice target area is greater than the mean value of the X coordinate deviation, and the Y coordinate deviation of the center point of the current frame slice target area and the adjacent slice target area is greater than the Y coordinate When the mean value of the deviation is obtained, the X coordinate and the Y coordinate of the center point of the current frame slice target area are replaced with the X coordinate and the Y coordinate of the center point of the target slice target area, respectively. As shown in FIG. 4a and FIG. 4b, the small squares in the figure represent the target area, the large squares represent the slices, and FIG. 4a and FIG. 4b are respectively before and after correction of a facial target area provided in the embodiment of the present invention. A comparison of the diagrams.

In one embodiment, as shown in FIG. 5, facial segmentation detection is performed on the selected slice to obtain The steps to take a trusted boundary point include:

Step S501: Obtain a transition boundary from a darker region to a brighter region in the slice selected by each frame.

The boundary detection operator is used to find the transition boundary of each frame slice from the darker region to the brighter region. The boundary detection operator here includes but is not limited to the following operator prewitt operator and sobel operator.

Step S502, determining a face region of each frame slice according to the gray value and the contour shape of the connected region where the transition boundary is located, and using the upper surface boundary of the face region as an alternative segmentation boundary, as shown in FIG. 6a.

In this embodiment, the gray value and the contour form of the connected region where the transition boundary of each frame slice is located are analyzed, and the gray value and the contour form are compared with the pre-stored face gray value and the face contour form to find each frame. The sliced face area and the upper surface boundary of the face area is used as an alternative segmentation boundary.

Step S503, acquiring a plurality of trusted boundary points according to the candidate segmentation boundary acquired in the above step.

Each of the face boundary points is voted according to an alternative segmentation boundary of each frame slice, and the point at which the candidate segmentation boundary passes the most is determined as a trusted boundary point. In one embodiment, before the step of acquiring a plurality of trusted boundary points according to the candidate segmentation boundary of each frame slice, the step of: boundary growth is further included.

Since the starting and ending points of the candidate segmentation boundaries of each frame slice may be inconsistent, it will affect the subsequent acquisition of the trusted boundary. Therefore, by using the boundary growth, the complete face boundary of the slice can be obtained from left to right, thereby improving the accuracy of the boundary detection. .

In the present embodiment, the boundary growth is performed based on the gradient image. Taking the left and right endpoints of the candidate segmentation boundary as the growth point in the lateral direction, searching for the boundary point on the gradient image based on the neighborhood of the current growth point, and adding the searched boundary point to the current candidate segmentation boundary to obtain each frame Slice the full face border. The contrast effects before and after the boundary growth are shown in Figures 6a and 6b.

In an embodiment, as shown in FIG. 7, the step of acquiring a plurality of trusted boundary points according to the candidate segmentation boundary includes:

Step S701: Construct a boundary matrix corresponding to the slice selected by each frame according to the candidate segmentation boundary.

The dimensions of each boundary matrix are the same, and the specific size can be determined as the case may be. In this embodiment, the background point of the boundary matrix is set to 0, and the boundary point is set to 1. Of course, it can be set to other values as long as the background and boundary points can be distinguished. The boundary points are points corresponding to alternative segmentation boundaries in each frame slice.

Step S702, superimposing all the boundary matrices with the accumulating matrix to obtain a voting matrix.

In this embodiment, the accumulation matrix may be a zero matrix of the same dimension as the boundary matrix. Superimpose all boundary matrices onto an accumulating matrix so that alternate partition boundaries that traverse the same point will vote for that point.

Step S703, the maximum value of each column in the voting matrix is counted, and the point corresponding to the maximum value is determined as a trusted boundary point.

The alternative segmentation boundary will vote for the point through the same point. The background point is 0 and the boundary point is 1. If the number of 1s accumulated on an element (point) is more, the maximum of each column element (point) is counted separately. Value, you can get the point where each column is the most divided by the candidate partition boundary, that is, the point with the highest brightness, and determine that the point is a trusted boundary point, you can obtain the trusted boundary points of each column, and the trusted boundary of each column Point to obtain a trusted boundary of the ultrasound three-dimensional fetal facial contour.

In one embodiment, as shown in FIG. 8, the step of cropping the fetal volume data according to the trusted boundary point to obtain an ultrasound three-dimensional fetal facial contour image includes:

Step S801, a cropping template is created according to the trusted boundary point.

The main operation of cropping the fetal body data is to make a cropping template, which is to fill the trusted boundary point after the voting, the lower right corner of the image, and the lower left corner of the image as a closed area, as shown in FIG. 9 , which is a specific embodiment of the present invention. A cropping template provided in .

Step S802, cutting the fetal volume data according to the cropping module to obtain an ultrasound three-dimensional fetal facial contour image.

The trimming template obtained in the above step is “ANDed” with the selected frame slices—that is, the data of each frame slice and the white area of the cropping template are retained, and the data of the black portion is deleted. 10a and 10b are schematic views of the sagittal direction slice before and after cutting, respectively, provided in the embodiment of the present invention. By cutting the fetal body data with the trimming template, an image of the contour of the fetal face can be obtained, and the occlusion portion of the fetal face is automatically cropped, which simplifies the operation and improves the drawing rate.

In one embodiment, after the step of cropping the fetal volume data according to the trusted boundary point to obtain the fetal facial contour image, the method further comprises: three-dimensional rendering of the cropped fetal volume data.

The cropped volume data is rendered, and the rendering can be performed by a three-dimensional rendering method such as the well-known "ray casting method" to obtain a more intuitive image of the fetal facial contour.

In this embodiment, face detection is performed on each frame slice that includes the target area, and an alternative segmentation boundary of each frame slice is obtained, and each of the face boundary points is performed by using an alternative segmentation boundary of each slice. Voting results in the point where the candidate segmentation boundary passes the most - the trusted boundary point. To improve the accuracy of the boundary detection, the boundary growth can be performed after the candidate segmentation boundary is obtained, and the cropping template is created according to the trusted boundary point, and the cropping is performed. The template automatically cuts the fetal body data, and automatically cuts the occlusion portion of the fetal face, which makes the operation simple and quick, reduces the error caused by the manual operation, and the image quality is higher.

The following is an embodiment of an ultrasonic three-dimensional fetal facial contour image processing system provided in an embodiment of the present invention. The embodiment of the ultrasonic three-dimensional fetal facial contour image processing system is implemented based on the above-described embodiment of the ultrasonic three-dimensional fetal facial contour image processing method. For an exhaustive description of the ultrasound three-dimensional fetal facial contour image processing system, please refer to the aforementioned embodiment of the ultrasonic three-dimensional fetal facial contour image processing method.

Please refer to FIG. 11, which is a structural block diagram of an embodiment of an ultrasonic three-dimensional fetal facial contour image processing system provided in an embodiment of the present invention. As shown, the system includes:

The target area detecting module 111 is configured to detect a multi-frame slice of the fetal volume data in a predetermined direction to obtain a target area of each frame slice, wherein the target area includes a fetal head area.

The trusted boundary point obtaining module 112 is configured to filter the slice including the target area, and perform face boundary detection on the selected slice to obtain a trusted boundary point.

The cropping module 113 is further configured to: create a cropping template according to the trusted boundary point; and perform cropping on the fetal volume data according to the cropping module to obtain an ultrasound three-dimensional fetal facial contour image.

In one embodiment, the system further includes: a correction module. The correction module is configured to correct the position of the current frame slice target area when the difference between the position of the current frame slice target area and the position of the adjacent slice target area exceeds a predetermined threshold.

The correction module is further configured to traverse all the selected slices to obtain a sequence of <frame number, target area>; respectively obtain the position of the frame target area of each frame in the <frame number, target area> sequence and the position of the adjacent slice target area Deviation of the deviation; calculating the mean value of the deviation; when the deviation of the position of the current frame slice target area from the position of the adjacent slice target area is greater than the mean value, the center point coordinate of the current frame slice target area is replaced with the center of the target slice target area Point coordinates.

In an embodiment, the trusted boundary point obtaining module is further configured to: obtain a transition boundary from the darker region to the brighter region in the filtered slice of each frame; according to the gray of the connected region where the transition boundary is located And a contour shape, determining a face region of the selected slice of each frame, and using an upper surface boundary of the face region as an alternative segmentation boundary; and acquiring a plurality of trusted boundary points according to the candidate segmentation boundary.

In one embodiment, the trusted boundary point acquisition module includes a boundary growth unit for boundary growth.

In an embodiment, the trusted boundary point obtaining module is further configured to: construct, according to the candidate partitioning boundary, a boundary matrix corresponding to the filtered slice in each frame; and The accumulation matrix is superimposed to obtain a voting matrix; the maximum value of each column in the voting matrix is counted, and the point corresponding to the maximum value is determined as a trusted boundary point.

In one embodiment, the target area detecting module 111 is further configured to: detect, by using a preset classifier, the target area of each frame slice of the fetal volume data in a predetermined direction; and save the target area and its corresponding slice.

In one embodiment, the system further includes a rendering module for three-dimensional rendering of the cropped fetal volume data.

The ultrasonic three-dimensional fetal facial contour image processing system of the present embodiment is used to implement the aforementioned ultrasonic three-dimensional fetal facial contour image processing method. Therefore, the specific embodiment in the ultrasonic three-dimensional fetal facial contour image processing system can be seen in the foregoing ultrasonic three-dimensional fetal facial contour image. An embodiment of the processing method, for example, the target area detecting module 111, the trusted boundary point acquiring module 112, and the trimming module 113 are respectively used to implement steps S101, S102, and S103 in the above-described ultrasonic three-dimensional fetal facial contour image processing method, For the specific implementation manners, reference may be made to the descriptions of the respective partial embodiments, and details are not described herein again.

The ultrasonic three-dimensional fetal facial contour image processing system provided by the embodiment performs face detection on the slice including the target region, obtains a face boundary of each frame slice, and sets a deviation from a position difference with a target region of the adjacent slice. The position of the target area of the value is corrected to improve the accuracy of the detection, and then the face boundary detection is performed on the target area of the slice, and the point at which the face boundary passes the most is obtained by voting on the face boundary point - credible At the boundary point, the fetal body data is cut by the trusted boundary point, and the occlusion portion of the fetal face is automatically cut, which simplifies the operation of the detecting person and improves the three-dimensional imaging rate of the ultrasound. The technical principles of the present invention have been described above in connection with specific embodiments. These descriptions are only for the purpose of explaining the principles of the present invention and are not to be construed as Limit the scope of protection. Based on the explanation herein, those skilled in the art can devise various other embodiments of the present invention without departing from the scope of the invention.

Claims

An ultrasonic three-dimensional fetal facial contour image processing method, comprising:

Detecting a plurality of frame slices of the fetal volume data in a predetermined direction to obtain a target region of each frame slice, wherein the target region includes a fetal head region;

Screening the slice containing the target area, and performing face boundary detection on the selected slice to obtain a trusted boundary point;

The fetal volume data is cropped according to the trusted boundary point to obtain an ultrasound three-dimensional fetal facial contour image.
The method according to claim 1, wherein after the step of filtering out the slice including the target area, the method further comprises:

When the difference between the position of the current frame slice target area and the position of the adjacent slice target area exceeds a predetermined threshold, the position of the current frame slice target area is corrected.
The method according to claim 2, wherein the step of correcting the position of the current frame slice target area comprises:

Traversing all of the selected slices to obtain a sequence of <frame number, target area>;

Deviating the position of the target area of each frame slice in the <frame number, target area> sequence from the position of the adjacent slice target area, respectively;

Calculating the mean of the deviations;

When the deviation of the position of the current frame slice target area from the position of the adjacent slice target area is greater than the mean value, the center point coordinate of the current frame slice target area is replaced with the center point coordinate of the target slice target area.
The method according to claim 1, wherein the step of performing facial boundary detection on the selected slice to obtain a trusted boundary point comprises:

Obtaining a transition boundary from the darker region to the brighter region of the selected slice in each frame;

Determining a face region of the sliced slice of each frame according to a gray value and a contour shape of the connected region where the transition boundary is located, and using an upper surface boundary of the face region as an alternative segmentation boundary;

Obtaining a plurality of trusted boundary points according to the alternative segmentation boundary.
The method of claim 4, wherein said said partitioning boundary is based on said alternative Before the step of obtaining multiple trusted boundary points, the steps include: boundary growth.
The method according to claim 4, wherein the step of acquiring a plurality of trusted boundary points according to the candidate segmentation boundary comprises:

Constructing, according to the candidate segmentation boundary, a boundary matrix corresponding to the selected slice of each frame;

Superimposing all of the boundary matrix and the accumulation matrix to obtain a voting matrix;

The maximum value of each column in the voting matrix is counted, and the point corresponding to the maximum value is determined as a trusted boundary point.
The method according to claim 1, wherein the step of cropping the fetal volume data according to the trusted boundary point to obtain an ultrasound three-dimensional fetal facial contour image comprises:

Making a cropping template according to the trusted boundary point;

The fetal volume data is cropped according to the cutting module to obtain an ultrasound three-dimensional fetal facial contour image.
The method according to claim 1, wherein the step of detecting the multi-frame slice of the fetal volume data in a predetermined direction to acquire the target area of each frame slice comprises:

Detecting, by using a preset classifier, the target area of each frame slice of the fetal volume data in a predetermined direction;

The target area and its corresponding slice are saved.
An ultrasonic three-dimensional fetal facial contour image processing system, comprising:

a target area detecting module, configured to detect a multi-frame slice of the fetal body data in a predetermined direction to obtain a target area of each frame slice, wherein the target area includes a fetal head area;

a trusted boundary point acquiring module, configured to filter out a slice including a target area, and perform face boundary detection on the selected slice to obtain a trusted boundary point;

And a cropping module, configured to crop the fetal volume data according to the trusted boundary point to obtain an ultrasound three-dimensional fetal facial contour image.
The system of claim 9 wherein said system further comprises:

And a correction module, configured to correct a position of the current frame slice target area when a difference between a position of the current frame slice target area and a position of the adjacent slice target area exceeds a predetermined threshold.