CN108961278B - Method and system for abdominal wall muscle segmentation based on image data - Google Patents
Method and system for abdominal wall muscle segmentation based on image dataDownload PDFInfo
- Publication number
- CN108961278B CN108961278BCN201810638182.3ACN201810638182ACN108961278BCN 108961278 BCN108961278 BCN 108961278BCN 201810638182 ACN201810638182 ACN 201810638182ACN 108961278 BCN108961278 BCN 108961278B
- Authority
- CN
- China
- Prior art keywords
- muscle
- abdominal wall
- image
- hernia
- contour
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
Images
Classifications
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/155—Segmentation; Edge detection involving morphological operators
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
Description
Technical Field
The invention relates to the technical field of image processing, in particular to an abdominal wall muscle segmentation method and an abdominal wall muscle segmentation system based on image data.
Background
In modern medical treatment, B-ultrasound, CT (Computed Tomography) and MRI (Nuclear Magnetic Resonance Imaging) are common image diagnosis technologies for diagnosing abdominal wall hernia diseases, and in the clinical application process, the diagnosis of abdominal wall hernia diseases by B-ultrasound, CT and MRI images has the following three defects:
the identification of the abdominal wall hernia defect abdominal wall muscle by B ultrasonic, CT and MRI images is unclear. Most of patients with abdominal hernia have lesions of abdominal wall muscles around the defect, the CT value of the tissues is reduced (the CT value of normal skeletal muscles is 45-75, while the CT value of abdominal wall muscles around the defect is often not in the 95% confidence interval), and partial patients have steatosis and the like. Therefore, two-dimensional images in clinical existing equipment (CT and MRI) examination are not clear, and even if partial muscle degeneration focus can be seen, the two-dimensional images are easy to ignore.
The measurement of the space size and volume of the abdominal hernia sac by B-ultrasonic, CT and MRI images is unclear. Clinically, when an abdominal hernia patient is subjected to surgical treatment, the spatial anatomical relationship between a first abdominal cavity (normal abdominal cavity) and a second abdominal cavity (hernia sac space) needs to be evaluated, the respective spatial volume and the volume occupied by contents need to be measured respectively, and then whether the abdominal cavity can completely accommodate all hernia contents is evaluated. Two-dimensional images such as CT, MRI and the like can not clearly evaluate the size of the respective space volume and the volume occupied by the contents.
The measurement of the position and the size of the hernia defect of the abdominal wall hernia by B ultrasonic, CT and MRI images is not clear. The position and the size of the hernia defect (small size: 0-4 cm, medium size: 4-8 cm, large size: 8-12 cm: large size: >12 cm) are displayed by B-ultrasonic or abdominal CT flat scanning, but the comprehensive conditions of abdominal wall hernia (the area of the hernia ring and the volume of the hernia sac) cannot be considered by the two-dimensional image, and the pathological changes of abdominal wall muscles around the defect are not considered.
In view of the above, it is obvious that the prior art has inconvenience and disadvantages in practical use, so that improvement is needed.
Disclosure of Invention
In view of the above-mentioned drawbacks, an object of the present invention is to provide a method and a system for abdominal wall muscle segmentation based on image data, so as to achieve automatic extraction and segmentation of an image area of abdominal hernia muscle and three-dimensional multi-angle visualization imaging of the abdominal hernia muscle area.
In order to achieve the above object, the present invention provides a method for abdominal wall muscle segmentation based on image data, comprising:
A. carrying out self-adaptive bilateral filtering pretreatment on the acquired image of the abdominal wall muscle;
B. setting a pre-segmentation abdominal wall hernia muscle image area for the preprocessed image of the abdominal wall hernia muscle according to the abdominal wall hernia muscle image characteristics, and extracting an initial abdominal wall hernia muscle segmentation area;
C. setting an image characteristic function of the abdominal wall hernia muscle, and extracting an abdominal wall hernia muscle edge contour in the initial abdominal wall hernia muscle partition area;
D. and optimizing the edge contour of the abdominal wall hernia muscle to obtain the final abdominal wall hernia muscle division area.
According to the method, the step A comprises the following steps:
a1, calculating a noise level spatial domain standard deviation Sigma S of the image by adopting iterative operation;
a2, calculating a value domain Gaussian function standard deviation Sigma R of the image;
and A3, according to the spatial domain standard deviation Sigma S and the value domain Gaussian function standard deviation Sigma R, performing bilateral filtering pretreatment on the image to obtain a pretreated image of the abdominal wall muscle.
According to the method, the step B comprises the following steps:
b1, according to the abdominal wall hernia muscle image characteristics, drawing the outline of a first abdominal wall hernia muscle image area by a first cross-section image and a last cross-section image in the preprocessed abdominal wall hernia muscle image;
b2, according to the abdominal wall hernia muscle image characteristics, drawing the outline of a second abdominal wall hernia muscle image area in the first and last coronary images in the preprocessed abdominal wall muscle image images;
and B3, generating a first segmentation result of the abdominal muscle image according to the sketched outlines of the first abdominal hernia muscle image area and the second abdominal hernia muscle image area.
According to the method, the step C comprises the following steps:
c1, determining an initial contour line according to the first segmentation result; the initial contour line comprises or does not comprise all the initial abdominal hernia muscle division areas;
c2, dividing the initial abdominal hernia muscle division area into an inner part and an outer part according to the initial contour line, and measuring the average gray value of the inner part and the outer part;
c3, calculating the image gray scale distance advantages of the areas of the inner part and the outer part;
and C4, according to the C-V level set model, performing iterative computation on the contours of the inner part and the outer part to form a boundary for dividing the initial abdominal wall hernia muscle division area, and obtaining the abdominal wall hernia muscle edge contour.
According to the method, the step D comprises the following steps:
d1, performing morphological treatment of free growth, corrosion, expansion and cavity filling on the obtained abdominal wall hernia muscle edge contour to keep the integrity and smoothness of the abdominal wall hernia muscle edge contour;
and D2, sequencing contour points of the fuzzy boundary along the fuzzy boundary to obtain the final abdominal wall hernia muscle segmentation area.
In order to achieve another object of the present invention, the present invention further provides a system for abdominal wall muscle segmentation based on image data, including:
the preprocessing module is used for carrying out self-adaptive bilateral filtering preprocessing on the acquired image of the abdominal wall muscle;
the first extraction module is used for setting a pre-segmentation abdominal wall hernia muscle image area for the preprocessed abdominal wall hernia muscle image according to the abdominal wall hernia muscle image characteristics and extracting an initial abdominal wall hernia muscle segmentation area;
the second extraction module is used for setting an image characteristic function of the abdominal wall hernia muscle and extracting an abdominal wall hernia muscle edge contour in the initial abdominal wall hernia muscle partition area;
and the optimization module is used for optimizing the edge contour of the abdominal wall hernia muscle to obtain the final abdominal wall hernia muscle division area.
According to the system, the preprocessing module comprises:
the first calculation submodule is used for calculating the noise level spatial domain standard deviation Sigma S of the image by adopting iterative operation;
the second calculation submodule is used for calculating a value domain Gaussian function standard deviation Sigma R of the image;
and the preprocessing submodule is used for performing bilateral filtering preprocessing on the image according to the spatial domain standard deviation Sigma S and the value domain Gaussian function standard deviation Sigma R to obtain a preprocessed image of the abdominal wall muscle.
According to the system, the first extraction module comprises:
the first drawing submodule is used for drawing the outline of a first abdominal hernia muscle image area according to the characteristics of the abdominal hernia muscle image and the first and last series cross-section images in the preprocessed abdominal wall hernia muscle image;
the second drawing submodule is used for drawing the outline of a second abdominal wall hernia muscle image area according to the abdominal wall hernia muscle image characteristics and the first and the last coronary position images in the preprocessed abdominal wall hernia muscle image;
and the generation submodule is used for generating a first segmentation result of the abdominal muscle image according to the sketched outlines of the first abdominal wall hernia muscle image area and the second abdominal wall hernia muscle image area.
According to the system, the second extraction module comprises:
the initial sub-module is used for determining an initial contour line according to the first segmentation result; the initial contour line contains or does not contain all the initial abdominal wall hernia muscle division areas;
the measuring submodule is used for dividing the initial abdominal wall hernia muscle division area into an inner part and an outer part according to the initial contour line and measuring the average gray value of the inner part and the outer part;
the third calculation submodule is used for calculating the image gray scale distance advantages of the areas of the inner part and the outer part;
and the fourth calculation submodule is used for performing iterative calculation on the contours of the inner part and the outer part according to the C-V level set model to form a boundary for dividing the initial abdominal wall hernia muscle division area and obtain the abdominal wall hernia muscle edge contour.
According to the system, the optimization module comprises:
the first optimization submodule is used for performing morphological processing of free growth, corrosion, expansion and cavity filling on the obtained abdominal wall hernia muscle edge contour so as to maintain the integrity and smoothness of the abdominal wall hernia muscle edge contour;
and the second optimization submodule is used for sequencing contour points of the fuzzy boundary along the fuzzy boundary in the abdominal wall hernia muscle edge contour to obtain a final abdominal wall hernia muscle segmentation area.
The acquired abdominal wall muscle image is subjected to self-adaptive bilateral filtering pretreatment to obtain a good image boundary, then a pre-segmentation abdominal wall hernia muscle image area is set for the pretreated abdominal wall muscle image based on the pretreated image according to the abdominal wall hernia muscle image characteristics, an initial abdominal wall hernia muscle segmentation area is extracted, and an interested target area is obtained; after the interested target area is extracted, setting an image characteristic function of the abdominal wall hernia muscle, and extracting an abdominal wall hernia muscle edge contour in the initial abdominal wall hernia muscle segmentation area to finally form a segmentation target boundary; and finally, performing morphological treatment on a target area, and performing optimization treatment on the abdominal wall hernia muscle edge contour to obtain a final abdominal wall hernia muscle division area. Therefore, the imaging characteristics of the abdominal hernia muscle are automatically segmented in the image data, the image area of the abdominal hernia muscle is automatically extracted and segmented, and the three-dimensional multi-angle visual imaging of the hernia muscle area is realized. The accurate segmentation result can visually display the anatomical structure of human viscera through three-dimensional visual reconstruction, and the stereoscopic vision can help hernia and abdominal wall surgeons in an all-around manner, and can be used for accurately positioning abdominal wall defects and the muscle lesion range of the abdominal wall around the abdominal wall defects.
Drawings
FIG. 1 is a schematic diagram of a system for abdominal wall muscle segmentation based on image data according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a system for abdominal wall muscle segmentation based on image data according to an embodiment of the present invention;
fig. 3 is a flowchart of a method for abdominal wall muscle segmentation based on image data according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, in an embodiment of the present invention, asystem 100 for abdominal wall muscle segmentation based on image data is provided, including:
thepreprocessing module 10 is used for performing adaptive bilateral filtering preprocessing on the acquired image of the abdominal wall muscle;
thefirst extraction module 20 is configured to set a pre-segmentation abdominal wall hernia muscle image area for the preprocessed abdominal wall hernia muscle image according to the abdominal wall hernia muscle image characteristics, and extract an initial abdominal wall hernia muscle segmentation area;
asecond extraction module 30, configured to set an image feature function of the abdominal wall hernia muscle, and extract an abdominal wall hernia muscle edge contour in the initial abdominal wall hernia muscle partition area;
and theoptimization module 40 is configured to perform optimization processing on the abdominal wall hernia muscle edge contour to obtain a final abdominal wall hernia muscle division area.
In this embodiment, first, in order to obtain a good image boundary, thepreprocessing module 10 performs adaptive bilateral filtering preprocessing on the acquired image of the abdominal wall muscle through thepreprocessing module 10; the image of the abdominal wall muscle can be acquired through B-ultrasound, CT and MRI. Then, based on the preprocessed image, thefirst extraction module 20 sets a pre-segmentation abdominal wall hernia muscle image area for the preprocessed image of the abdominal wall hernia muscle according to the image characteristics of the abdominal wall hernia muscle, and extracts an initial abdominal wall hernia muscle segmentation area; according to the abdominal wall hernia muscle image characteristics, interactive segmentation is carried out to manually set a pre-segmented abdominal wall hernia muscle image area so as to extract an interested target area. After the target area of interest is extracted, thesecond extraction module 30 sets an image feature function of the abdominal wall hernia muscle, and extracts an abdominal wall hernia muscle edge contour in the initial abdominal wall hernia muscle segmentation area to finally form a segmentation target boundary. Theoptimization module 40 applies target area morphological processing according to the obtained abdominal wall hernia muscle edge contour to perform optimization processing on the abdominal wall hernia muscle edge contour, so as to obtain a final abdominal wall hernia muscle division area. Therefore, the present embodiment provides a system for automatically segmenting abdominal muscles for abdominal hernia diseases based on image data, and thesystem 100 for automatically segmenting abdominal hernia muscles based on image data automatically segments the imaging characteristics of abdominal hernia muscles, automatically extracts the image areas of the abdominal hernia muscles, and realizes three-dimensional multi-angle visualization imaging of the hernia muscle areas.
Referring to fig. 2, in one embodiment of the present invention, thepreprocessing module 10 includes:
thefirst calculation submodule 11 is configured to calculate a noise level spatial domain standard deviation Sigma S of the image by using iterative operation;
thesecond calculation submodule 12 is configured to calculate a value domain gaussian function standard deviation Sigma R of the image;
and thepreprocessing submodule 13 is configured to perform bilateral filtering preprocessing on the image according to the spatial domain standard deviation Sigma S and the value domain gaussian function standard deviation Sigma R to obtain a preprocessed image of the abdominal wall muscle.
In this embodiment, thefirst computation submodule 11 computes a noise level spatial domain standard deviation Sigma S of the image by using iterative operation; the specific calculation process of thefirst calculation submodule 11 includes:
1. each pixel point of the image is taken as a left vertex, and a plurality of 7 multiplied by 7 areas can be divided;
2. taking the initially input image as an initial weak texture image, and calculating a noise level spatial domain Gaussian function standard deviation Sigma S (K);
3. carrying out mean comparison of a gradient covariance matrix on the initial weak texture image to obtain a new weak texture image area, and calculating a noise level spatial domain standard deviation Sigma S (K + 1);
4. judging Sigma S (K) and Sigma S (K + 1), if Sigma S (K) ≈ Sigma S (K + 1), stopping calculating and transferring to the step 5, otherwise, transferring to the step 6.
5. Output Sigma S = Sigma S (K + 1).
6. Let Sigma S (K) = Sigma S (K + 1), go tostep 2, continue the calculation, iterate 3 times in total.
Then, thesecond calculation submodule 12 calculates a value domain gaussian function standard deviation Sigma R of the image; the specific calculation process of thesecond calculation submodule 12 includes:
1. detecting each pixel value of the image by a boundary algorithm to obtain an edge pixel I(s);
2. calculating the sum Is of the pixels at the visible edge;
3. according to the formula: e = Is/N, (N Is the total number of pixel points), calculating the edge intensity e of the image;
4. according to the formula: sigma R = be, (b is a linear coefficient), and the value range gaussian function standard deviation Sigma R is calculated.
Finally, thepreprocessing submodule 13 performs bilateral filtering preprocessing on the image according to the spatial domain standard deviation Sigma S and the value domain gaussian function standard deviation Sigma R to obtain a preprocessed image of the abdominal wall muscle. Therefore, bilateral filtering pretreatment is carried out according to the space domain standard deviation Sigma S and the value domain Gaussian function standard deviation Sigma R which are measured in a self-adapting mode.
Referring to fig. 2, in one embodiment of the present invention, thefirst extraction module 20 includes:
thefirst drawing submodule 21 is used for drawing the outline of a first abdominal wall hernia muscle image area according to the abdominal wall hernia muscle image characteristics and the first and last series cross-sectional images in the preprocessed abdominal wall hernia muscle image;
the second delineation submodule 22 is used for delineating the contour of a second abdominal wall hernia muscle image area according to the abdominal wall hernia muscle image characteristics and the first and last serial coronal images in the preprocessed abdominal wall hernia muscle image;
and the generatingsubmodule 23 is configured to generate a first segmentation result of the abdominal muscle image according to the outlined outlines of the first and second abdominal hernia muscle image areas.
In this embodiment, based on the preprocessed image images, according to the abdominal wall hernia muscle image features, the pre-segmented abdominal wall hernia muscle image areas are set interactively, automatically or manually, and the target area of interest is extracted. The predetermined search setting or the manual setting may be performed by thefirst delineation sub-module 21, the second delineation sub-module 22, and the generation sub-module 23 described above.
Referring to fig. 2, in one embodiment of the present invention, thesecond extraction module 30 includes:
aninitial submodule 31, configured to determine an initial contour line according to the first segmentation result; the initial contour line contains or does not contain all the initial abdominal wall hernia muscle division areas;
thedetermination submodule 32 is configured to divide the initial abdominal hernia muscle division area into an inner part and an outer part according to the initial contour line, and determine average gray values of the inner part and the outer part;
athird computing submodule 33, configured to compute image gray scale distance dominance of the regions of the inner and outer portions;
and thefourth calculation submodule 34 is configured to perform iterative calculation on the contours of the inner part and the outer part according to the C-V level set model to form a boundary for dividing the initial abdominal wall hernia muscle division area, so as to obtain the abdominal wall hernia muscle edge contour.
In this embodiment, the initial sub-module 31 defines an initial contour line according to the contour line of the first segmentation result based on the image of the first segmentation result, themeasurement sub-module 32 divides the image into an inner region and an outer region according to the initial contour line of the image, and divides the image into two parts C1 and C2 according to the average gray level of the image of the inner region and the outer region; thethird calculation submodule 33 performs level set evolution according to the contour length, the inner and outer area and the inner and outer area gray scale distance capability, and thefourth calculation submodule 34 performs level set evolution according to the C-V level set model, and moves the contour line to gradually approach the boundary between the inner part and the outer part under the action of the internal force and the external force, so as to finally form a segmentation target boundary. The C-V level set model is a new active contour model proposed by Chan and Vese in 2001, and a curve is evolved by minimizing an energy function based on a simplified Mumford-Shah model by using a level set thought.
Referring to fig. 2, in one embodiment of the present invention, theoptimization module 40 includes:
thefirst optimization submodule 41 is used for performing morphological processing of free growth, corrosion, expansion and cavity filling on the obtained abdominal wall hernia muscle edge contour so as to maintain the integrity and smoothness of the abdominal wall hernia muscle edge contour;
and thesecond optimization submodule 42 is configured to sort contour points of the fuzzy boundary along the fuzzy boundary of the contour of the abdominal wall hernia muscle edge to obtain a final abdominal wall hernia muscle segmentation area.
In this embodiment, according to the obtained abdominal wall hernia muscle edge contour, a target area morphological treatment is applied, and an initial contour optimization treatment is performed to finally segment the target, so as to obtain the abdominal wall hernia muscle area. The specific optimization process of theoptimization module 40 is as follows:
the first step is as follows: morphological processing of the target area; the morphological processing of the target area includes free growth, erosion, expansion, void filling, maintaining contour integrity and smoothness. Thefirst optimization submodule 41 implements the specific processing content:
1. filling holes through closed operation, and expanding the outline to make up for missing boundaries;
2. boundary noise is eliminated by an on operation, and spike noise is eliminated by narrowing a contour.
The second step, optimizing the initial contour;
in order to handle the blurred boundary of the target region in the current-layer medical image, the contour points need to be sorted along the boundary. The contour point sorting based on polar coordinates can obtain a good sorting result. The implementation is realized by thesecond optimization submodule 42, which specifically comprises the following steps:
1. and acquiring the center point of the contour. And establishing a two-dimensional coordinate system according to the image plane, wherein the top point of the upper left corner of the image is the origin, the horizontal direction to the right is the positive direction of the X axis, and the vertical direction to the down is the positive direction of the Y axis. And calculating a central point according to the outline points obtained by rough segmentation.
2. And establishing a polar coordinate system. The central point of the contour is taken as a pole 0 in the image plane, a ray is led to be a pole axis along the positive direction of the x axis of the coordinate system of the image plane, and the counterclockwise direction is taken as the positive direction of the angle.
3. And (4) sorting the contour points. Most medical images have a relatively regular tissue structure, so there are two main cases for the ordering of contour points: firstly, when the contour central point of the target area of the medical image is positioned in the contour, namely the target shape is a convex shape, the polar angle of each contour point is different, the contour points can be sequenced only according to the polar angle, and the contour points are stored in a contour queue according to the polar angle in a size from human to human; secondly, when the center point of the outline is positioned outside the outline, namely the target shape is a concave shape, the outline points are firstly sorted from small to large according to polar angles, then the condition that the polar angles are equal is discussed, and after all the points are traversed, the outline points in the stack are sequentially popped up and stored in an outline queue. Therefore, the image area of the abdominal hernia muscle is finally extracted and segmented, and the three-dimensional multi-angle visual imaging of the hernia muscle area is realized.
Thesystem 100 for abdominal wall muscle segmentation based on image data provided in the above embodiments belongs to the field of medical image three-dimensional visualization processing and computer medical auxiliary diagnosis and treatment systems, and in particular relates to an abdominal wall muscle segmentation, identification and three-dimensional visualization imaging technology for abdominal wall hernia diseases in medical images by using a computer artificial intelligence technology.
In order to achieve another object of the present invention, the present invention further provides a method for performing abdominal wall muscle segmentation based on image data by using thesystem 100 for performing multiple abdominal wall muscle segmentation based on image data.
Referring to fig. 3, in an embodiment of the present invention, a method for abdominal wall muscle segmentation based on image data is provided, including:
step S301, thepreprocessing module 10 performs adaptive bilateral filtering preprocessing on the acquired image of the abdominal wall muscle; so as to obtain good image boundaries of the abdominal wall muscles;
step S302, thefirst extraction module 20 sets a pre-segmentation abdominal wall hernia muscle image area for the preprocessed abdominal wall hernia muscle image according to the abdominal wall hernia muscle image characteristics, and extracts an initial abdominal wall hernia muscle segmentation area; according to the abdominal wall hernia muscle image characteristics, performing interactive segmentation presetting or manual setting on a pre-segmented abdominal wall hernia muscle image area to extract an interested target area;
step S303, thesecond extraction module 30 sets an image feature function of the abdominal wall hernia muscle, and extracts an abdominal wall hernia muscle edge contour in the initial abdominal wall hernia muscle partition area; extracting an abdominal wall hernia muscle edge contour in the initial abdominal wall hernia muscle segmentation area by setting an image characteristic function of the abdominal wall hernia muscle, and finally forming a segmentation target boundary;
step S304, theoptimization module 40 performs optimization processing on the abdominal wall hernia muscle edge contour to obtain the final abdominal wall hernia muscle division area. And according to the obtained abdominal wall hernia muscle edge contour, applying target area morphological processing, performing initial contour optimization processing, and finally segmenting a target to obtain the abdominal wall hernia muscle area. In addition, in order to process the fuzzy boundary of the target area in the current-layer medical image, the contour points are sorted along the boundary, and the sorting of the contour points based on polar coordinates can obtain a good sorting result so as to obtain the final abdominal wall hernia muscle segmentation area.
The abdominal wall muscle segmentation method based on the image data can realize the digitization and the intellectualization of physical and physiological parameters of a human body, and carries out dynamic medical image processing, accurate three-dimensional reconstruction and quantitative analysis research, thereby providing a more vivid brand-new technical means for the diagnosis and treatment of diseases, and the education and training of doctors. Therefore, the automatic division result of abdominal hernia muscle is of great significance for objectively and quantitatively evaluating the conditions of abdominal hernia (transverse diameter of hernia ring, area of hernia ring, volume of hernia sac) and anterior abdominal wall muscles (rectus abdominis, external oblique muscles, internal oblique muscles and transverse abdominal muscles). The accurate segmentation result can visually display the anatomical structure of human viscera through three-dimensional visual reconstruction, and the stereoscopic vision can help hernia and abdominal wall surgeons in an all-around manner, and can be used for accurately positioning abdominal wall defects and the muscle lesion range of the abdominal wall around the abdominal wall defects.
In one embodiment of the present invention, the step S301 includes:
thefirst calculation submodule 11 calculates a noise level spatial domain standard deviation Sigma S of the image by iterative operation;
thesecond calculation submodule 12 calculates a value domain gaussian function standard deviation Sigma R of the image;
in this embodiment, thepreprocessing submodule 13 performs bilateral filtering preprocessing on the image according to the spatial domain standard deviation Sigma S and the value domain gaussian function standard deviation Sigma R to obtain a preprocessed image of the abdominal wall muscle.
Thefirst calculation submodule 11 calculates a noise level spatial domain standard deviation Sigma S of the image by iterative operation; specifically, the method comprises the following steps:
1. each pixel point of the image is taken as a left vertex, and a plurality of 7 multiplied by 7 areas can be divided;
2. taking the initially input image as an initial weak texture image, and calculating a noise level spatial domain Gaussian function standard deviation Sigma S (K);
3. carrying out mean comparison of a gradient covariance matrix on the initial weak texture image to obtain a new weak texture image area, and calculating a noise level spatial domain standard deviation Sigma S (K + 1);
4. judging Sigma S (K) and Sigma S (K + 1), if Sigma S (K) ≈ Sigma S (K + 1), stopping calculating and transferring to the step 5, otherwise, transferring to the step 6.
5. Output Sigma S = Sigma S (K + 1).
6. Let Sigma S (K) = Sigma S (K + 1), go tostep 2, continue the calculation, iterate 3 times in total.
Then, thesecond calculation submodule 12 calculates a value domain gaussian function standard deviation Sigma R of the image; the specific calculation process of thesecond calculation submodule 12 includes:
1. detecting each pixel value of the image by a boundary algorithm to obtain an edge pixel I(s);
2. calculating the sum Is of the pixels at the visible edge;
3. according to the formula: e = Is/N, (N Is the total number of pixel points), calculating the edge intensity e of the image;
4. according to the formula: sigma R = be, (b is a linear coefficient), and the value range gaussian function standard deviation Sigma R is calculated.
And finally, thepreprocessing submodule 13 performs bilateral filtering preprocessing on the image according to the spatial domain standard deviation Sigma S and the value domain gaussian function standard deviation Sigma R to obtain a preprocessed image of the abdominal wall muscle. Therefore, bilateral filtering pretreatment is carried out according to the space domain standard deviation Sigma S and the value domain Gaussian function standard deviation Sigma R which are measured in a self-adapting mode.
In an embodiment of the present invention, the step S302 includes:
thefirst drawing submodule 21 draws the outline of a first abdominal wall hernia muscle image area according to the abdominal wall hernia muscle image characteristics and the first and last series cross-sectional images in the preprocessed abdominal wall hernia muscle image;
the second delineation submodule 22 delineates the contour of a second abdominal wall hernia muscle image area according to the abdominal wall hernia muscle image characteristics and the first and last coronary images in the preprocessed abdominal wall hernia muscle image;
thegeneration submodule 23 generates a first segmentation result of the abdominal muscle image according to the outlined first and second abdominal hernia muscle image area outlines. By adopting the steps, the pre-segmented abdominal wall hernia muscle image area can be set automatically or manually through interactive segmentation based on the preprocessed image and according to the abdominal wall hernia muscle image characteristics, and the interested target area is extracted.
In an embodiment of the present invention, the step S303 includes:
theinitial submodule 31 determines an initial contour line according to the first segmentation result; the initial contour line contains or does not contain all the initial abdominal wall hernia muscle division areas;
thedetermination submodule 32 divides the initial abdominal hernia muscle division area into an inner part and an outer part according to the initial contour line, and determines average gray values of the inner part and the outer part;
thethird computing submodule 33 computes the image gray scale distance dominance of the regions of the inner and outer parts;
and thefourth calculation submodule 34 performs iterative calculation on the contours of the inner part and the outer part according to the C-V level set model to form a boundary for dividing the initial abdominal wall hernia muscle division area, so as to obtain the abdominal wall hernia muscle edge contour.
In the embodiment, the image based on the first segmentation result is defined as an initial contour line according to the contour line of the first segmentation, the image is divided into an inner area and an outer area according to the initial contour line of the image, and the image is divided into two parts C1 and C2 according to the average gray level of the image of the inner area and the outer area; according to the contour length, the area of the inner region and the outer region and the gray distance capability of the inner region and the outer region, level set evolution is carried out according to a C-V level set model, the contour line of the movable wheel gradually approaches to the boundary of the inner region and the outer region under the action of internal force and external force, and finally a segmentation target boundary is formed. The iterative computation of thefourth computation submodule 34 includes the following steps:
1. calculating the average gray value of the inner area and the outer area;
2. calculating the distance maintenance of gray values of the inner area and the outer area;
3. gradually propagating and calculating the movable contour lines of the two areas;
4. the movable contour is searched inwards on the outer side, and the movable contour is searched outwards on the inner side;
5. calculating a suppressed active contour line;
6. judging whether the movable contour line is stable or not; if the contour line is stable, calculating to obtain the contour line, otherwise, turning to thestep 2 to continue calculating.
7. A stable contour is obtained.
In an embodiment of the present invention, the step S304 includes:
thefirst optimization submodule 41 performs morphological processing of free growth, erosion, expansion and cavity filling on the obtained abdominal wall hernia muscle edge contour, and keeps the integrity and smoothness of the abdominal wall hernia muscle edge contour;
thefirst optimization submodule 41 sequences contour points of the fuzzy boundary along the fuzzy boundary of the fuzzy boundary in the abdominal wall hernia muscle edge contour to obtain a final abdominal wall hernia muscle segmentation area.
In this embodiment, according to the obtained abdominal wall hernia muscle edge contour, a target area morphological treatment is applied, and an initial contour optimization treatment is performed to finally segment the target, so as to obtain the abdominal wall hernia muscle area. Specifically, thefirst optimization submodule 41 utilizes morphological processing of the target area; the morphological processing of the target area includes free growth, erosion, expansion, void filling, maintaining contour integrity and smoothness. The specific processing content is as follows: filling holes through closed operation, and expanding the outline to make up for missing boundaries; boundary noise is eliminated by an on operation, and spike noise is eliminated by narrowing a contour. Thefirst optimization submodule 41 performs an optimization process of the initial contour. This is to process the fuzzy boundary of the target region in the current layer medical image, and the contour points need to be sorted along the boundary. The contour point sorting based on polar coordinates can obtain a good sorting result. The method comprises the following specific steps:
1. and acquiring the center point of the contour. And establishing a two-dimensional coordinate system according to the image plane, wherein the vertex at the upper left corner of the image is the origin, the horizontal right direction is the positive direction of the X axis, and the vertical downward direction is the positive direction of the Y axis. And calculating a central point according to the outline points obtained by rough segmentation.
2. And establishing a polar coordinate system. The central point of the contour is taken as a pole 0 in the image plane, a ray is led to be a pole axis along the positive direction of the x axis of the coordinate system of the image plane, and the counterclockwise direction is taken as the positive direction of the angle.
3. And (4) sorting the contour points. Most medical images have a relatively regular tissue structure, so there are two main cases for the ordering of contour points: firstly, when the contour central point of the target area of the medical image is positioned in the contour, namely the target shape is a convex shape, the polar angle of each contour point is different, the contour points can be sequenced only according to the polar angle, and the contour points are stored in a contour queue according to the polar angle in a size from human to human; secondly, when the center point of the outline is positioned outside the outline, namely the target shape is a concave shape, the outline points are firstly sorted from small to large according to polar angles, then the condition that the polar angles are equal is discussed, and after all the points are traversed, the outline points in the stack are sequentially popped up and stored in an outline queue.
Therefore, the method for dividing abdominal wall muscles based on image data provided by the invention has important significance for realizing automatic division of abdominal hernia muscles and objectively and quantitatively evaluating the conditions of abdominal wall hernias (hernia ring transverse diameter, hernia ring area and hernia sac volume) and the conditions of anterior abdominal wall muscles (abdominal rectus muscle, abdominal external oblique muscle, abdominal internal oblique muscle and abdominal transverse muscle). The accurate segmentation result can visually display the anatomical structure of human viscera through three-dimensional visual reconstruction, and the stereoscopic vision can help hernia and abdominal wall surgeons in an all-around manner, and can be used for accurately positioning abdominal wall defects and the muscle lesion range of the abdominal wall around the abdominal wall defects.
In summary, the acquired abdominal wall muscle image is subjected to adaptive bilateral filtering preprocessing to obtain a good image boundary, and then, on the basis of the preprocessed image, according to the abdominal wall hernia muscle image characteristics, a pre-segmentation abdominal wall hernia muscle image area is set for the preprocessed abdominal wall muscle image, and an initial abdominal wall hernia muscle segmentation area is extracted to obtain an interested target area; after the interested target area is extracted, setting an image characteristic function of the abdominal wall hernia muscle, and extracting an abdominal wall hernia muscle edge contour in the initial abdominal wall hernia muscle segmentation area to finally form a segmentation target boundary; and finally, performing morphological treatment on a target area, and performing optimization treatment on the abdominal wall hernia muscle edge contour to obtain a final abdominal wall hernia muscle division area. Therefore, the imaging characteristics of the abdominal hernia muscle are automatically segmented in the image data, the image area of the abdominal hernia muscle is automatically extracted and segmented, and the three-dimensional multi-angle visual imaging of the hernia muscle area is realized. The accurate segmentation result can visually display the anatomical structure of human viscera through three-dimensional visual reconstruction, and the stereoscopic vision can help hernia and abdominal wall surgeons in an all-around manner, and can be used for accurately positioning abdominal wall defects and the muscle lesion range of the abdominal wall around the abdominal wall defects.
The present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof, and it should be understood that various changes and modifications can be effected therein by one skilled in the art without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (2)
1. An abdominal wall muscle segmentation method based on image data is characterized by comprising the following steps:
A. carrying out self-adaptive bilateral filtering pretreatment on the acquired image of the abdominal wall muscle;
B. setting a pre-segmentation abdominal wall hernia muscle image area for the preprocessed abdominal wall hernia muscle image according to the abdominal wall hernia muscle image characteristics, and extracting an initial abdominal wall hernia muscle segmentation area;
C. setting an image characteristic function of the abdominal wall hernia muscle, and extracting an abdominal wall hernia muscle edge contour in the initial abdominal wall hernia muscle partition area;
D. optimizing the edge contour of the abdominal wall hernia muscle to obtain a final abdominal wall hernia muscle division area;
the step A comprises the following steps:
a1, calculating a noise level spatial domain standard deviation Sigma S of the image by adopting iterative operation;
a2, calculating a value domain Gaussian function standard deviation Sigma R of the image;
a3, according to the spatial domain standard deviation Sigma S and the value domain Gaussian function standard deviation Sigma R, performing bilateral filtering pretreatment on the image to obtain a pretreated image of the abdominal wall muscle;
the step B comprises the following steps:
b1, according to the abdominal wall hernia muscle image characteristics, drawing the outline of a first abdominal wall hernia muscle image area by a first cross-section image and a last cross-section image in the preprocessed abdominal wall hernia muscle image;
b2, according to the abdominal wall hernia muscle image characteristics, drawing the outline of a second abdominal wall hernia muscle image area in the first and last coronary images in the preprocessed abdominal wall muscle image images;
b3, generating a first segmentation result of the abdominal muscle image according to the sketched outlines of the first abdominal hernia muscle image area and the second abdominal hernia muscle image area;
the step C comprises the following steps:
c1, determining an initial contour line according to the first segmentation result; the initial contour line contains or does not contain all the initial abdominal wall hernia muscle division areas;
c2, dividing the initial abdominal hernia muscle division area into an inner part and an outer part according to the initial contour line, and measuring the average gray value of the inner part and the outer part;
c3, calculating the image gray scale distance advantages of the areas of the inner part and the outer part;
c4, according to the C-V level set model, performing iterative computation on the contours of the inner part and the outer part to form a boundary for segmenting the initial abdominal wall hernia muscle segmentation area and obtain an abdominal wall hernia muscle edge contour;
the step D comprises the following steps:
d1, performing morphological treatment of free growth, corrosion, expansion and cavity filling on the obtained abdominal wall hernia muscle edge contour to keep the integrity and smoothness of the abdominal wall hernia muscle edge contour;
d2, sorting contour points of the fuzzy boundary along the fuzzy boundary in the contour of the abdominal wall hernia muscle edge to obtain a final abdominal wall hernia muscle segmentation area;
the optimization processing comprises sorting contour points in a fuzzy boundary of a target area in the image of the abdominal wall hernia muscle by adopting a polar angle sorting mode in a polar coordinate, and specifically comprises the following steps:
(1) Obtaining the center point of the contour: establishing a two-dimensional coordinate system according to an image plane, wherein the top point of the upper left corner of the image is an original point, the horizontal right direction is the positive direction of an X axis, the vertical downward direction is the positive direction of a Y axis, and a central point is calculated according to contour points obtained by rough segmentation;
(2) Establishing a polar coordinate system: taking the central point of the contour in the image plane as a pole 0, simultaneously leading a ray along the positive direction of the x axis of the coordinate system of the image plane as a pole axis, and setting the anticlockwise direction as the positive direction of the angle;
(3) Sorting of contour points: when the contour central point of the target area of the medical image is positioned in the contour, the target shape is a convex shape, the polar angles of all contour points are different, the contour points can be sequenced according to the polar angles, and the contour points are stored in a contour queue according to the polar angles in a size from human size to human size; when the center point of the outline is positioned outside the outline, the target shape is a concave shape, the outline points are firstly sorted from small to large according to polar angles, then the condition that the polar angles are equal is discussed, and after all the points are traversed, the outline points in the stack are sequentially popped up and stored in an outline queue.
2. A system for implementing the method for abdominal wall muscle segmentation based on image data of claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810638182.3ACN108961278B (en) | 2018-06-20 | 2018-06-20 | Method and system for abdominal wall muscle segmentation based on image data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810638182.3ACN108961278B (en) | 2018-06-20 | 2018-06-20 | Method and system for abdominal wall muscle segmentation based on image data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108961278A CN108961278A (en) | 2018-12-07 |
| CN108961278Btrue CN108961278B (en) | 2023-01-31 |
Family
ID=64490371
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810638182.3AActiveCN108961278B (en) | 2018-06-20 | 2018-06-20 | Method and system for abdominal wall muscle segmentation based on image data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108961278B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110111386B (en)* | 2019-05-10 | 2023-01-24 | 深圳大学 | A method and system for automatically measuring structural angles in muscle tissue |
| CN111583286B (en)* | 2020-04-09 | 2023-01-20 | 天津大学 | Abdomen MRI (magnetic resonance imaging) image contour extraction method based on Flow-XDoG operator |
| CN113409309B (en)* | 2021-07-16 | 2023-10-27 | 北京积水潭医院 | Muscle CT image delineation method, system, electronic equipment and machine storage medium |
| CN116309385B (en)* | 2023-02-27 | 2023-10-10 | 之江实验室 | Abdominal fat and muscle tissue measurement method and system based on weakly supervised learning |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002222410A (en)* | 2001-01-25 | 2002-08-09 | Hitachi Medical Corp | Diagnostic imaging device |
| CN101887583A (en)* | 2010-06-24 | 2010-11-17 | 东软集团股份有限公司 | Method and device for extracting brain tissue image |
| CN104809723A (en)* | 2015-04-13 | 2015-07-29 | 北京工业大学 | Three-dimensional liver CT (computed tomography) image automatically segmenting method based on hyper voxels and graph cut algorithm |
| CN106447682A (en)* | 2016-08-29 | 2017-02-22 | 天津大学 | Automatic segmentation method for breast MRI focus based on Inter-frame correlation |
| CN106570856A (en)* | 2016-08-31 | 2017-04-19 | 天津大学 | Common carotid artery intima-media thickness measuring device and method combining level set segmentation and dynamic programming |
| CN106890009A (en)* | 2017-03-17 | 2017-06-27 | 深圳大学 | A kind of method for automatic measurement and device of skeletal muscle Volume Changes |
| CN107481252A (en)* | 2017-08-24 | 2017-12-15 | 上海术理智能科技有限公司 | Dividing method, device, medium and the electronic equipment of medical image |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009111580A2 (en)* | 2008-03-04 | 2009-09-11 | Tomotherapy Incorporated | Method and system for improved image segmentation |
| CN103886564A (en)* | 2013-11-29 | 2014-06-25 | 沈阳东软医疗系统有限公司 | PET heart image myocardial contour line segmentation method and device |
- 2018
- 2018-06-20CNCN201810638182.3Apatent/CN108961278B/enactiveActive
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002222410A (en)* | 2001-01-25 | 2002-08-09 | Hitachi Medical Corp | Diagnostic imaging device |
| CN101887583A (en)* | 2010-06-24 | 2010-11-17 | 东软集团股份有限公司 | Method and device for extracting brain tissue image |
| CN104809723A (en)* | 2015-04-13 | 2015-07-29 | 北京工业大学 | Three-dimensional liver CT (computed tomography) image automatically segmenting method based on hyper voxels and graph cut algorithm |
| CN106447682A (en)* | 2016-08-29 | 2017-02-22 | 天津大学 | Automatic segmentation method for breast MRI focus based on Inter-frame correlation |
| CN106570856A (en)* | 2016-08-31 | 2017-04-19 | 天津大学 | Common carotid artery intima-media thickness measuring device and method combining level set segmentation and dynamic programming |
| CN106890009A (en)* | 2017-03-17 | 2017-06-27 | 深圳大学 | A kind of method for automatic measurement and device of skeletal muscle Volume Changes |
| CN107481252A (en)* | 2017-08-24 | 2017-12-15 | 上海术理智能科技有限公司 | Dividing method, device, medium and the electronic equipment of medical image |
Non-Patent Citations (1)
| Title |
|---|
| "医学图像分割算法研究及在腿部肌肉组织分割中的应用";王亚琼;《中国优秀硕士学位论文全文数据库信息科技辑》;20130315(第03期);第2-5章* |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108961278A (en) | 2018-12-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107230206B (en) | Multi-mode data-based 3D pulmonary nodule segmentation method for hyper-voxel sequence lung image | |
| CN107563998B (en) | Method for processing heart image in medical image | |
| US8787648B2 (en) | CT surrogate by auto-segmentation of magnetic resonance images | |
| CN110363765A (en) | Osteoporosis parameter automatic measurement method based on CT image | |
| CN108961278B (en) | Method and system for abdominal wall muscle segmentation based on image data | |
| CN102411795B (en) | Combined display method of multimode image of coronary artery | |
| CN109064476B (en) | CT chest radiography lung tissue image segmentation method based on level set | |
| CN107945169B (en) | A Coronary Artery Imaging Analysis Method | |
| US20150023578A1 (en) | Device and method for determining border of target region of medical images | |
| CN107403201A (en) | Tumour radiotherapy target area and jeopardize that organ is intelligent, automation delineation method | |
| CN103440665A (en) | Automatic segmentation method of knee joint cartilage image | |
| CN106997594B (en) | Method and device for locating eye tissue | |
| CN114972272B (en) | A Grad-CAM-based method for segmenting COVID-19 lesions | |
| CN111105385A (en) | Method for processing human body joint data provided based on tomography technology | |
| Giannini et al. | A fully automatic algorithm for segmentation of the breasts in DCE-MR images | |
| CN116650115A (en) | Orthopedic surgery navigation registration method based on UWB mark points | |
| CN112562058B (en) | Method for quickly establishing intracranial vascular simulation three-dimensional model based on transfer learning | |
| Mori et al. | Automated nomenclature of bronchial branches extracted from CT images and its application to biopsy path planning in virtual bronchoscopy | |
| Zhu et al. | 3D automatic MRI level set segmentation of inner ear based on statistical shape models prior | |
| CN109544673B (en) | A three-dimensional imaging method and system for medical imaging cartilage segmentation | |
| Zhang et al. | Segmenting the thoracic, abdominal and pelvic musculature on CT scans combining atlas-based model and active contour model | |
| CN114418984A (en) | Human tissue symmetry detection and analysis method based on ultrasound | |
| Eskildsen et al. | Quantitative comparison of two cortical surface extraction methods using MRI phantoms | |
| Feng et al. | Automatic fetal weight estimation using 3D ultrasonography | |
| Reinhardt et al. | Intrathoracic airway wall detection using graph search and scanner PSF information |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |