CN105427273A - Abdominal fat segmentation method and device based on nuclear magnetic resonance image - Google Patents
Abdominal fat segmentation method and device based on nuclear magnetic resonance image Download PDFInfo
- Publication number
- CN105427273A CN105427273A CN201410482898.0A CN201410482898A CN105427273A CN 105427273 A CN105427273 A CN 105427273A CN 201410482898 A CN201410482898 A CN 201410482898A CN 105427273 A CN105427273 A CN 105427273A
- Authority
- CN
- China
- Prior art keywords
- image
- fat
- initial
- pixel
- stomach fat
- 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.)
- Granted
Links
Landscapes
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention discloses an abdominal fat segmentation method and device based on a nuclear magnetic resonance image so that accurate segmentation of visceral fat and subcutaneous fat in a nuclear magnetic resonance abdominal image is rapidly achieved. The method comprises preprocessing an initial nuclear magnetic resonance image, segmenting abdominal fat and a surrounding organization portion of the abdominal fat in a preprocessing image to obtain an initial image of the abdominal fat, segmenting an initial image of the abdominal fat into an outer contour image of the subcutaneous fat and an initial segmentation image of the visceral fat, performing level set evolution on the initial segmentation image of the visceral fat to obtain a final image of the visceral fat, performing subtraction operation on the final image of the visceral fat and a product of the outer contour image of the subcutaneous fat and the final image of the visceral fat, and taking an obtained result as a final image of the subcutaneous fat. Compared with the prior art, the method provided in the invention can rapidly and accurately segment the subcutaneous fat and the visceral fat.
Description
Technical field
The present invention relates to field of medical image processing, be specifically related to a kind of stomach fat dividing method based on nuclear magnetic resonance image and device.
Background technology
In modern society, abdominal obesity has become the healthy serious problems of impact the elderly day by day.Abdominal obesity can divide for the obesity too much caused by subcutaneous fat and the obesity too much caused by interior fat.It is less that subcutaneous fat crosses multipair healthy impact, and interior fat is crossed and can be caused the diseases such as metabolic disturbance, type II diabetes, hypertension and high fat of blood at most.Therefore, the distribution defining stomach fat effectively, is objectively extremely important.
Many medical imaging procedure have been applied in the abnormal fat detection of belly, such as, and Dual energy X ray absorptiometry, CT, ultrasonic and nuclear magnetic resonance (MagneticResonance, MR) etc.In all image modes, nuclear magnetic resonance is widely adopted because of its Non-ionizing radiation and to the characteristic of soft tissue high-resolution, wherein, fat more can be distinguished with body fluid by water fat isolated nuclei resonance effectively, has now become and has carried out for stomach fat the main image mode analyzed.
For the segmentation of subcutaneous fat, there are now automatic, semi-automatic and full-automatic three kinds of methods.Automatic method is mainly based on the nuclear magnetic resonance image of two dimension, and it has the high advantage of accuracy rate, but consuming time longer.Existing software is surrounded by NIHImage, SliceOmatic, Analyze, HippoFat and EasyVision, above software package all can realize manually, automatically and full automatic stomach fat split.But these software packages are all based on T1W nuclear magnetic resonance image, consuming time higher and very strong to the experience dependence of operator.
Under the basis of above-mentioned work, the existing segmentation of certain methods to stomach fat and subcutaneous fat is now attempted.Such as, the method increased by Threshold segmentation and region is realized jointly, or setting two initial boundaries realize the segmentation of two parts fat.But these two kinds of method manual interventions are excessively strong, and segmentation result is subject to the impact of Initial parameter sets.Also have and attempt from the angle of active appearance models, but, because everyone situation difference to some extent, the accuracy of abundant training set sample guarantee segmentation therefore must be collected.
Summary of the invention
The embodiment of the present invention provides a kind of stomach fat dividing method based on nuclear magnetic resonance image and device, to realize the accurate segmentation to interior fat and subcutaneous fat in nuclear magnetic resonance belly image fast.
The embodiment of the present invention provides a kind of stomach fat dividing method based on nuclear magnetic resonance image, and described method comprises:
Carry out pre-service to initial nuclear magnetic resonance image, described initial nuclear magnetic resonance image comprises stomach fat and described stomach fat surrounding tissue all-in-one-piece nuclear magnetic resonance image;
Split the stomach fat in described pretreatment image and described stomach fat surrounding tissue part, obtain the initial pictures of described stomach fat;
Be the outer profile image of subcutaneous fat and the initial segmentation image of interior fat by the initial image segmentation of described stomach fat;
Level set evolution is carried out to the initial segmentation image of described interior fat, obtains the final image of described interior fat;
The product of final image of the outer profile image of described subcutaneous fat and described interior fat and the final image of described interior fat are done additive operation, and acquired results is as the final image of described subcutaneous fat.
Another embodiment of the present invention provides a kind of stomach fat segmenting device based on nuclear magnetic resonance image, and described device comprises:
Pretreatment module, for carrying out pre-service to initial nuclear magnetic resonance image, described initial nuclear magnetic resonance image comprises stomach fat and described stomach fat surrounding tissue all-in-one-piece nuclear magnetic resonance image;
First segmentation module, for splitting stomach fat in described pretreatment image and described stomach fat surrounding tissue part, obtains the initial pictures of described stomach fat;
Second segmentation module, for by the initial image segmentation of described stomach fat being the outer profile image of subcutaneous fat and the initial segmentation image of interior fat;
Evolution module, for carrying out level set evolution to the initial segmentation image of described interior fat, obtains the final image of described interior fat;
Computing module, do additive operation for the product of the final image by the outer profile image of described subcutaneous fat and described interior fat and the final image of described interior fat, acquired results is as the final image of described subcutaneous fat.
From the invention described above embodiment, due to the stomach fat in the pre-service of incipient nucleus nuclear magnetic resonance image, pretreatment image and the described segmentation of stomach fat surrounding tissue part and the segmentation again etc. of stomach fat initial pictures, whole processing procedure is without the need to parameters input, automaticity is high, compared with prior art, method provided by the invention can be split subcutaneous fat and interior fat quickly and accurately.
Accompanying drawing explanation
Fig. 1 is the basic procedure schematic diagram of the stomach fat dividing method based on nuclear magnetic resonance image of the embodiment of the present invention;
Fig. 2 is the stomach fat segmenting device logical organization schematic diagram based on nuclear magnetic resonance image that the embodiment of the present invention provides;
Fig. 3 is the stomach fat segmenting device logical organization schematic diagram based on nuclear magnetic resonance image that another embodiment of the present invention provides;
Fig. 4 is the stomach fat segmenting device logical organization schematic diagram based on nuclear magnetic resonance image that another embodiment of the present invention provides;
Fig. 5 is the stomach fat segmenting device logical organization schematic diagram based on nuclear magnetic resonance image that another embodiment of the present invention provides;
Fig. 6 is the stomach fat segmenting device logical organization schematic diagram based on nuclear magnetic resonance image that another embodiment of the present invention provides.
Embodiment
The embodiment of the present invention provides a kind of stomach fat dividing method based on nuclear magnetic resonance image, comprise: carry out pre-service to initial nuclear magnetic resonance image, described initial nuclear magnetic resonance image comprises stomach fat and described stomach fat surrounding tissue all-in-one-piece nuclear magnetic resonance image; Split the stomach fat in described pretreatment image and described stomach fat surrounding tissue part, obtain the initial pictures of described stomach fat; Be the outer profile image of subcutaneous fat and the initial segmentation image of interior fat by the initial image segmentation of described stomach fat; Level set evolution is carried out to the initial segmentation image of described interior fat, obtains the final image of described interior fat; The product of final image of the outer profile image of described subcutaneous fat and described interior fat and the final image of described interior fat are done additive operation, and acquired results is as the final image of described subcutaneous fat.The embodiment of the present invention also provides accordingly based on the stomach fat segmenting device of nuclear magnetic resonance image.Below be described in detail respectively.
The basic procedure of the stomach fat dividing method based on nuclear magnetic resonance image of the embodiment of the present invention with reference to figure 1, can mainly comprise the steps that S101 is to step S105:
S101, carries out pre-service to initial nuclear magnetic resonance image, and described initial nuclear magnetic resonance image comprises stomach fat and described stomach fat surrounding tissue all-in-one-piece nuclear magnetic resonance image.
In embodiments of the present invention, carrying out pre-service to initial nuclear magnetic resonance image can be: by carrying out biased field rectification to described initial nuclear magnetic resonance image, obtains pretreatment image.Particularly, nuclear magnetic resonance (the MagneticResonance of clinical acquisition, MR) image can be equivalent to the MR image do not polluted by nonuniform field and be multiplied by the gray scale nonuniform field slowly changed in spatial domain, namely the uneven characteristic of gray scale can be similar to field by a smooth multiplicative and represents, be v (x, y)=u (x, y) f (x, y), wherein, (x, y) position that is pixel, v (x, y) for measuring the gradation of image obtained, u (x, y) is real gradation of image, f (x, y) is the biased field of unknown smooth change.By taking the logarithm to above formula, Fourier transform, be multiplied by identical frequency-domain function, spatial domain inverse transformation, both sides fetching number, the Recovery image obtaining biased field is: u (x, y)=v (x, y)/f (x, y), herein, v (x, y) is initial nuclear magnetic resonance image, u (x, y) be not by the MR image that nonuniform field pollutes, the gray scale nonuniform field that f (x, y) representation space slowly changes, available gauss low frequency filter realizes, it suppress the nonuniform field of multiplicative, obtain the MR image after correcting.
S102, splits the stomach fat in the pretreated pretreatment image of step S101 and described stomach fat surrounding tissue part, obtains the initial pictures of stomach fat.
Particularly, in embodiments of the present invention, split the stomach fat in the pretreated pretreatment image of step S101 and described stomach fat surrounding tissue part, the initial pictures obtaining stomach fat comprises the steps that S1021 is to step S1027:
S1021, random selecting two pixels from the pixel of pretreatment image every tomographic image, respectively as the center of each bunch in two bunches.
K-mean cluster (K-meansclustering) is the non-supervisory real-time clustering algorithm of one that MacQueen proposes, and Data Placement is by the basis of minimum error function predetermined class number K.In embodiments of the present invention, K can get 2, namely two bunches (Cluster).
S1022, calculates the Euclidean distance of remaining pixel to Liang Gecu center respectively, and remaining pixel is assigned to Euclidean distance shorter bunch.
Pixel is to the Euclidean distance at bunch center
wherein, (x
1, y
1) be the coordinate of pixel, (x
2, y
2) be the coordinate at bunch center.
S1023, according to cluster result, recalculates the center of in two bunches each bunch.
Circular is the arithmetical mean of all pixels dimension separately in getting bunch.
S1024, by pixel whole in every for pretreatment image tomographic image according to recalculating gained center cluster again.
S1025, repeats step S1022 to S1024, until cluster result no longer changes.
In embodiments of the present invention, so-called cluster result no longer changes and refers to, the result of this cluster is the same with the result of a front cluster.
S1026, is defined as stomach fat by part high for Liang Gecu center gray-scale value.
S1027, by all grey scale pixel value sets in stomach fat two bunches bunch, by grey scale pixel value resets all in another bunch in stomach fat two bunches, obtains the two-value template of stomach fat.
The initial image segmentation of stomach fat is the outer profile image of subcutaneous fat and the initial segmentation image of interior fat by S103.
Particularly, in embodiments of the present invention, be that the outer profile image of subcutaneous fat and the initial segmentation image of interior fat comprise the steps that S1031 is to step S1037 by the initial image segmentation of stomach fat:
S1031, detects the outline of stomach fat by Canny operator.
S1032, adopts horizontal scanning line to scan the two-value template of the stomach fat that previous embodiment step S1027 obtains from top to bottom.
S1034, calculates the intersection point of each sweep trace and stomach fat.
S1035, according to the coordinate figure on x-axis direction from small to large antinode sort, choose the coordinate of the maximum pixel of coordinate figure on the coordinate of the minimum pixel of coordinate figure on x-axis direction and x-axis direction.
S1036, all grey scale pixel values on first line segment and the second line segment are put 1, obtain the initial segmentation image of interior fat, wherein, first line segment is that the minimum pixel of the coordinate figure on x-axis direction is connected with the frontier point apart from its nearlyer side sweep trace the line segment formed, and the second line segment is that the maximum pixel of the coordinate figure on x-axis direction is connected with the frontier point apart from its nearlyer side sweep trace the line segment formed.
S1037, all grey scale pixel values pixel that pixel minimum for the coordinate figure on x-axis direction is maximum with the coordinate figure on x-axis direction be connected on the line segment of formation put 1, obtain the outer profile image of subcutaneous fat.
S104, carries out level set evolution to the initial segmentation image of interior fat, obtains the final image of described interior fat.
Particularly, in embodiments of the present invention, carry out level set evolution to the initial segmentation image of interior fat, the final image obtaining described interior fat comprises the steps that S1041 is to step S1043:
S1041, obtains gradient vector flow field by minimization of energy function.
Catch the external fields of force of image border from both direction while that gradient vector flow (GVFSnake) being a kind of, it is the gradient vector diffusion of the outline map f of image function, by bivector field f (x, y)=[u (x, y), v (x, y)] composition, gradient vector flow field is obtained by minimization of energy function:
S1042, using stomach fat border as a level set of two-dimentional continuous function, by constantly upgrading two-dimentional continuous function with the curve lain in described two-dimentional continuous function that develops.
In embodiments of the present invention, two-dimentional continuous function can be levelset function.If use ψ to represent two-dimentional continuous function (such as, levelset function), then upgrade two-dimentional continuous function and can represent with the following methods with the curve lain in described two-dimentional continuous function that develops:
Zero level set representations aim curve Γ (t) of two dimension continuous function (such as, levelset function), that is:
S1043, the image obtained when the level set of two-dimentional continuous function is zero is as the final image of interior fat.
In two dimension continuous function, the speed of curve evolvement is made up of two parts, that is, relevant with curvature of curve internal forces item and the external force item relevant with GVF.When the level set of two-dimentional continuous function (such as, levelset function) is zero, interior fat segmentation terminates.
S105, the outer profile image of subcutaneous fat and the product of the final image of interior fat and the final image of interior fat are done additive operation, and acquired results is as the final image of subcutaneous fat.
From the stomach fat dividing method based on nuclear magnetic resonance image that the invention described above embodiment provides, due to the stomach fat in the pre-service of incipient nucleus nuclear magnetic resonance image, pretreatment image and the described segmentation of stomach fat surrounding tissue part and the segmentation again etc. of stomach fat initial pictures, whole processing procedure is without the need to parameters input, automaticity is high, compared with prior art, method provided by the invention can be split subcutaneous fat and interior fat quickly and accurately.
Be described the stomach fat segmenting device based on nuclear magnetic resonance image of the embodiment of the present invention for performing the above-mentioned stomach fat dividing method based on nuclear magnetic resonance image below, its basic logical structure is with reference to accompanying drawing 2.For convenience of explanation, the stomach fat segmenting device based on nuclear magnetic resonance image of accompanying drawing 2 example illustrate only the part relevant to the embodiment of the present invention, mainly comprise pretreatment module 201, first and split module 202, second segmentation module 203, evolution module 204 and computing module 205, each module is described in detail as follows:
Pretreatment module 201, for carrying out pre-service to initial nuclear magnetic resonance image, described initial nuclear magnetic resonance image comprises stomach fat and described stomach fat surrounding tissue all-in-one-piece nuclear magnetic resonance image;
First segmentation module 202, for splitting stomach fat in described pretreatment image and described stomach fat surrounding tissue part, obtains the initial pictures of described stomach fat;
Second segmentation module 203, for by the initial image segmentation of described stomach fat being the outer profile image of subcutaneous fat and the initial segmentation image of interior fat;
Evolution module 204, for carrying out level set evolution to the initial segmentation image of described interior fat, obtains the final image of described interior fat;
Computing module 205, do additive operation for the product of the final image by the outer profile image of described subcutaneous fat and described interior fat and the final image of described interior fat, acquired results is as the final image of described subcutaneous fat.
It should be noted that, in the embodiment of the stomach fat segmenting device based on nuclear magnetic resonance image of above accompanying drawing 2 example, the division of each functional module only illustrates, can be as required in practical application, the facility of the such as configuration requirement of corresponding hardware or the realization of software is considered, and above-mentioned functions distribution is completed by different functional modules, inner structure by the described stomach fat segmenting device based on nuclear magnetic resonance image is divided into different functional modules, to complete all or part of function described above.And, in practical application, corresponding functional module in the present embodiment can be by corresponding hardware implementing, also can perform corresponding software by corresponding hardware to complete, such as, aforesaid pretreatment module can be have to perform aforementionedly to carry out pretreated hardware to initial nuclear magnetic resonance image, such as pretreater also can be general processor or other hardware devices that can perform corresponding computer program thus complete aforementioned function; For another example aforesaid computing module, can be have to perform aforementioned the outer profile image of subcutaneous fat and the product of the final image of interior fat and the final image of interior fat to be done additive operation, acquired results is as the hardware of the final image function of subcutaneous fat, such as arithmetical unit also can be general processor or other hardware devices (each embodiment that this instructions provides all can apply foregoing description principle) that can perform corresponding computer program thus complete aforementioned function.
The pretreatment module 201 of accompanying drawing 2 example can comprise correcting unit 301, as shown in Figure 3 the stomach fat segmenting device based on nuclear magnetic resonance image that provides of another embodiment of the present invention.Correcting unit 301, for by carrying out biased field rectification to described initial nuclear magnetic resonance image, obtains pretreatment image.
Particularly, nuclear magnetic resonance (the MagneticResonance of clinical acquisition, MR) image can be equivalent to the MR image do not polluted by nonuniform field and be multiplied by the gray scale nonuniform field slowly changed in spatial domain, namely the uneven characteristic of gray scale can be similar to field by a smooth multiplicative and represents, be v (x, y)=u (x, y) f (x, y), wherein, (x, y) position that is pixel, v (x, y) for measuring the gradation of image obtained, u (x, y) is real gradation of image, f (x, y) is the biased field of unknown smooth change.Correcting unit 301 by taking the logarithm to above formula, Fourier transform, be multiplied by identical frequency-domain function, spatial domain inverse transformation, both sides fetching number, the Recovery image obtaining biased field is: u (x, y)=v (x, y)/f (x, y), herein, v (x, y) is initial nuclear magnetic resonance image, u (x, y) be not by the MR image that nonuniform field pollutes, the gray scale nonuniform field that f (x, y) representation space slowly changes, available gauss low frequency filter realizes, it suppress the nonuniform field of multiplicative, obtain the MR image after correcting.
First segmentation module 202 of accompanying drawing 2 example can comprise a bunch center determining unit 401, allocation units 402, center re-computation unit 403, heavy cluster cell 404, stomach fat determining unit 405 and two-value template determining unit 406, the stomach fat segmenting device based on nuclear magnetic resonance image that theres is provided of another embodiment of the present invention as shown in Figure 4, wherein:
Bunch center determining unit 401, for two pixels of random selecting in the pixel from the every tomographic image of described pretreatment image, respectively as the center of each bunch in two bunches.
K-mean cluster (K-meansclustering) is the non-supervisory real-time clustering algorithm of one that MacQueen proposes, and Data Placement is by the basis of minimum error function predetermined class number K.In the present embodiment, K can get 2, namely two bunches (Cluster).
Allocation units 402, for calculating the Euclidean distance of remaining pixel to described Liang Gecu center respectively, and described remaining pixel is assigned to Euclidean distance shorter bunch.
Pixel is to the Euclidean distance at bunch center
wherein, (x
1, y
1) be the coordinate of pixel, (x
2, y
2) be the coordinate at bunch center.
Center re-computation unit 403, for according to cluster result, recalculates the center of in described two bunches each bunch.
Circular is, center re-computation unit 403 to get bunch in the arithmetical mean of all pixels dimension separately.
Heavy cluster cell 404, for recalculating gained center cluster again by pixel whole in every for described pretreatment image tomographic image according to described.
Allocation units 402, center re-computation unit 403 and heavy cluster cell 404 repeat, successively until cluster result no longer changes.
In the present embodiment, so-called cluster result no longer changes and refers to, the result of this cluster is the same with the result of a front cluster.
Stomach fat determining unit 405, for being defined as stomach fat by part high for described Liang Ge center gray-scale value.
Two-value template determining unit 406, for by all grey scale pixel value sets in described stomach fat two bunches bunch, by grey scale pixel value resets all in another bunch in described stomach fat two bunches, obtains the two-value template of described stomach fat.
Second segmentation module 203 of accompanying drawing 4 example can comprise detecting unit 5, scanning element 502, intersection point calculation unit 503, choose unit 504, first determining unit 505 and the second determining unit 506, the stomach fat segmenting device based on nuclear magnetic resonance image that theres is provided of another embodiment of the present invention as shown in Figure 5, wherein:
Detecting unit 501, for detecting the outline of described stomach fat by Canny operator.
Scanning element 502, for the two-value template adopting horizontal scanning line to scan described stomach fat from top to bottom.
Intersection point calculation unit 503, for calculating the intersection point of each sweep trace and described stomach fat.
Choose unit 504, for according to the coordinate figure on x-axis direction from small to large antinode sort, choose the coordinate of the maximum pixel of coordinate figure on the coordinate of the minimum pixel of coordinate figure on x-axis direction and x-axis direction.
First determining unit 505, for all grey scale pixel values on the first line segment and the second line segment are put 1, obtain the initial segmentation image of interior fat, described first line segment is that the minimum pixel of the coordinate figure on described x-axis direction is connected with the frontier point apart from its nearlyer side sweep trace the line segment formed, and described second line segment is that the maximum pixel of the coordinate figure on described x-axis direction is connected with the frontier point apart from its nearlyer side sweep trace the line segment formed.
Second determining unit 506, all grey scale pixel values be connected on the line segment of formation for the pixel that pixel minimum for the coordinate figure on x-axis direction is maximum with the coordinate figure on x-axis direction put 1, obtain the outer profile image of subcutaneous fat.
The evolution module 204 of accompanying drawing 2 example can comprise gradient vector flow field acquiring unit 601, function updating block 602 and the 3rd determining unit 603, the stomach fat segmenting device based on nuclear magnetic resonance image that theres is provided of another embodiment of the present invention as shown in Figure 6, wherein:
Gradient vector flow field acquiring unit 601, for obtaining gradient vector flow field by minimization of energy function.
Catch the external fields of force of image border from both direction while that gradient vector flow (GVFSnake) being a kind of, it is the gradient vector diffusion of the outline map f of image function, by bivector field f (x, y)=[u (x, y), v (x, y)] composition, gradient vector flow field is obtained by minimization of energy function:
Function updating block 602, for a level set using stomach fat border as two-dimentional continuous function, by constantly upgrading described two-dimentional continuous function with the curve lain in described two-dimentional continuous function that develops.
In the present embodiment, two-dimentional continuous function can be levelset function.If use ψ to represent two-dimentional continuous function (such as, levelset function), then upgrade two-dimentional continuous function and can represent with the following methods with the curve lain in described two-dimentional continuous function that develops:
Zero level set representations aim curve Γ (t) of two dimension continuous function (such as, levelset function), that is:
3rd determining unit 603, for the image that obtains when the level set of two-dimentional continuous function is zero final image as described interior fat.
In two dimension continuous function, the speed of curve evolvement is made up of two parts, that is, relevant with curvature of curve internal forces item and the external force item relevant with GVF.When the level set of two-dimentional continuous function (such as, levelset function) is zero, interior fat segmentation terminates.
It should be noted that, the content such as information interaction, implementation between each module/unit of said apparatus, due to the inventive method embodiment based on same design, its technique effect brought is identical with the inventive method embodiment, particular content see describing in the inventive method embodiment, can repeat no more herein.
One of ordinary skill in the art will appreciate that all or part of step in the various methods of above-described embodiment is that the hardware that can carry out instruction relevant by program has come, this program can be stored in a computer-readable recording medium, storage medium can comprise: ROM (read-only memory) (ROM, ReadOnlyMemory), random access memory (RAM, RandomAccessMemory), disk or CD etc.
The stomach fat dividing method based on nuclear magnetic resonance image provided the embodiment of the present invention above and device are described in detail, apply specific case herein to set forth principle of the present invention and embodiment, the explanation of above embodiment just understands method of the present invention and core concept thereof for helping; Meanwhile, for one of ordinary skill in the art, according to thought of the present invention, all will change in specific embodiments and applications, in sum, this description should not be construed as limitation of the present invention.
Claims (10)
1. based on a stomach fat dividing method for nuclear magnetic resonance image, it is characterized in that, described method comprises:
Carry out pre-service to initial nuclear magnetic resonance image, described initial nuclear magnetic resonance image comprises stomach fat and described stomach fat surrounding tissue all-in-one-piece nuclear magnetic resonance image;
Split the stomach fat in described pretreatment image and described stomach fat surrounding tissue part, obtain the initial pictures of described stomach fat;
Be the outer profile image of subcutaneous fat and the initial segmentation image of interior fat by the initial image segmentation of described stomach fat;
Level set evolution is carried out to the initial segmentation image of described interior fat, obtains the final image of described interior fat;
The product of final image of the outer profile image of described subcutaneous fat and described interior fat and the final image of described interior fat are done additive operation, and acquired results is as the final image of described subcutaneous fat.
2. method according to claim 1, is characterized in that, describedly carries out pre-service to initial nuclear magnetic resonance image, comprising:
By carrying out biased field rectification to described initial nuclear magnetic resonance image, obtain pretreatment image.
3. method according to claim 1, is characterized in that, the stomach fat in the described pretreatment image of described segmentation and described stomach fat surrounding tissue part, obtain the initial pictures of described stomach fat, comprise the steps that S1021 is to step S1027:
S1021, random selecting two pixels from the pixel of described pretreatment image every tomographic image, respectively as the center of each bunch in two bunches;
S1022, calculates the Euclidean distance of remaining pixel to described Liang Gecu center respectively, and described remaining pixel is assigned to Euclidean distance shorter bunch;
S1023, according to cluster result, recalculates the center of in described two bunches each bunch;
S1024, recalculates gained center cluster again by pixel whole in every for described pretreatment image tomographic image according to described;
S1025, repeats step S1022 to S1024, until cluster result no longer changes;
S1026, is defined as stomach fat by part high for described Liang Ge center gray-scale value;
S1027, by all grey scale pixel value sets in described stomach fat two bunches bunch, by grey scale pixel value resets all in another bunch in described stomach fat two bunches, obtains the two-value template of described stomach fat.
4. method according to claim 3, is characterized in that, described is the outer profile image of subcutaneous fat and the initial segmentation image of interior fat by the initial image segmentation of described stomach fat, comprises the steps that S1031 is to step S1037:
S1031, detects the outline of described stomach fat by Canny operator;
S1032, adopts horizontal scanning line to scan the two-value template of described stomach fat from top to bottom;
S1034, calculates the intersection point of each sweep trace and described stomach fat;
S1035, sorts to described intersection point from small to large according to the coordinate figure on x-axis direction, chooses the coordinate of the maximum pixel of coordinate figure on the coordinate of the minimum pixel of coordinate figure on x-axis direction and x-axis direction;
S1036, all grey scale pixel values on first line segment and the second line segment are put 1, obtain the initial segmentation image of interior fat, described first line segment is that the minimum pixel of the coordinate figure on described x-axis direction is connected with the frontier point apart from its nearlyer side sweep trace the line segment formed, and described second line segment is that the maximum pixel of the coordinate figure on described x-axis direction is connected with the frontier point apart from its nearlyer side sweep trace the line segment formed;
S1037, all grey scale pixel values pixel maximum with the coordinate figure on x-axis direction for pixel minimum for the coordinate figure on described x-axis direction be connected on the line segment of formation put 1, obtain the outer profile image of described subcutaneous fat.
5. method according to claim 1, is characterized in that, the described initial segmentation image to described interior fat carries out level set evolution, obtains the final image of described interior fat, comprising:
Gradient vector flow field is obtained by minimization of energy function;
Using stomach fat border as a level set of two-dimentional continuous function, by constantly upgrading described two-dimentional continuous function with the curve lain in described two-dimentional continuous function that develops;
The image obtained when the level set of two-dimentional continuous function is zero is as the final image of described interior fat.
6. based on a stomach fat segmenting device for nuclear magnetic resonance image, it is characterized in that, described device comprises:
Pretreatment module, for carrying out pre-service to initial nuclear magnetic resonance image, described initial nuclear magnetic resonance image comprises stomach fat and described stomach fat surrounding tissue all-in-one-piece nuclear magnetic resonance image;
First segmentation module, for splitting stomach fat in described pretreatment image and described stomach fat surrounding tissue part, obtains the initial pictures of described stomach fat;
Second segmentation module, for by the initial image segmentation of described stomach fat being the outer profile image of subcutaneous fat and the initial segmentation image of interior fat;
Evolution module, for carrying out level set evolution to the initial segmentation image of described interior fat, obtains the final image of described interior fat;
Computing module, do additive operation for the product of the final image by the outer profile image of described subcutaneous fat and described interior fat and the final image of described interior fat, acquired results is as the final image of described subcutaneous fat.
7. device according to claim 6, is characterized in that, described pretreatment module comprises:
Correcting unit, for by carrying out biased field rectification to described initial nuclear magnetic resonance image, obtains pretreatment image.
8. device according to claim 6, is characterized in that, described first segmentation module comprises:
Bunch center determining unit, for two pixels of random selecting in the pixel from the every tomographic image of described pretreatment image, respectively as the center of each bunch in two bunches;
Allocation units, for calculating the Euclidean distance of remaining pixel to described Liang Gecu center respectively, and described remaining pixel is assigned to Euclidean distance shorter bunch;
Center re-computation unit, for according to cluster result, recalculates the center of in described two bunches each bunch;
Heavy cluster cell, for recalculating gained center cluster again by pixel whole in every for described pretreatment image tomographic image according to described;
Allocation units, center re-computation unit and heavy cluster cell repeat, successively until cluster result no longer changes;
Stomach fat determining unit, for being defined as stomach fat by part high for described Liang Ge center gray-scale value;
Two-value template determining unit, for by all grey scale pixel value sets in described stomach fat two bunches bunch, by grey scale pixel value resets all in another bunch in described stomach fat two bunches, obtains the two-value template of described stomach fat.
9. device according to claim 8, is characterized in that, described second segmentation module, comprising:
Detecting unit, for detecting the outline of described stomach fat by Canny operator;
Scanning element, for the two-value template adopting horizontal scanning line to scan described stomach fat from top to bottom;
Intersection point calculation unit, for calculating the intersection point of each sweep trace and described stomach fat;
Choosing unit, for sorting to described intersection point from small to large according to the coordinate figure on x-axis direction, choosing the coordinate of the maximum pixel of coordinate figure on the coordinate of the minimum pixel of coordinate figure on x-axis direction and x-axis direction;
First determining unit, for all grey scale pixel values on the first line segment and the second line segment are put 1, obtain the initial segmentation image of interior fat, described first line segment is that the minimum pixel of the coordinate figure on described x-axis direction is connected with the frontier point apart from its nearlyer side sweep trace the line segment formed, and described second line segment is that the maximum pixel of the coordinate figure on described x-axis direction is connected with the frontier point apart from its nearlyer side sweep trace the line segment formed;
Second determining unit, all grey scale pixel values on the line segment formed for pixel maximum with the coordinate figure on x-axis direction for pixel minimum for the coordinate figure on described x-axis direction being connected put 1, obtain the outer profile image of described subcutaneous fat.
10. device according to claim 6, is characterized in that, described evolution module comprises:
Gradient vector flow field acquiring unit, for obtaining gradient vector flow field by minimization of energy function;
Function updating block, for a level set using stomach fat border as two-dimentional continuous function, by constantly upgrading described two-dimentional continuous function with the curve lain in described two-dimentional continuous function that develops;
3rd determining unit, for the image that obtains when the level set of two-dimentional continuous function is zero final image as described interior fat.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410482898.0A CN105427273B (en) | 2014-09-19 | 2014-09-19 | A kind of stomach fat dividing method and device based on nuclear magnetic resonance image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410482898.0A CN105427273B (en) | 2014-09-19 | 2014-09-19 | A kind of stomach fat dividing method and device based on nuclear magnetic resonance image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105427273A true CN105427273A (en) | 2016-03-23 |
| CN105427273B CN105427273B (en) | 2018-06-26 |
Family
ID=55505454
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410482898.0A Active CN105427273B (en) | 2014-09-19 | 2014-09-19 | A kind of stomach fat dividing method and device based on nuclear magnetic resonance image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105427273B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107705305A (en) * | 2017-10-20 | 2018-02-16 | 上海联影医疗科技有限公司 | A kind of method and device of Medical Image Processing |
| CN109523561A (en) * | 2018-12-19 | 2019-03-26 | 睿仁佳心(上海)科技有限公司 | Muscle and fat image dividing method in automatic abdomen |
| CN110179488A (en) * | 2019-06-11 | 2019-08-30 | 西北农林科技大学 | A method of it is formed using CT scan data detection pig living body trunk |
| CN110415246A (en) * | 2019-08-06 | 2019-11-05 | 东北大学 | A kind of analysis method of stomach fat ingredient |
| CN112168168A (en) * | 2020-10-14 | 2021-01-05 | 北京赛迈特锐医疗科技有限公司 | System and method for automatically and quantitatively evaluating whole body fat by using MR (magnetic resonance) technology |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1767788A (en) * | 2003-04-08 | 2006-05-03 | 株式会社日产医药 | Apparatus and method of supporting diagnostic imaging for medical use |
| CN103208114A (en) * | 2013-01-25 | 2013-07-17 | 西安电子科技大学 | Stomach adipose tissue extraction method based on interactive segmentation |
-
2014
- 2014-09-19 CN CN201410482898.0A patent/CN105427273B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1767788A (en) * | 2003-04-08 | 2006-05-03 | 株式会社日产医药 | Apparatus and method of supporting diagnostic imaging for medical use |
| CN103208114A (en) * | 2013-01-25 | 2013-07-17 | 西安电子科技大学 | Stomach adipose tissue extraction method based on interactive segmentation |
Non-Patent Citations (4)
| Title |
|---|
| POSITANO V 等: "An Accurate and Robust Method for Unsupervised Assessment of Abdominal Fat by MRI", 《JOURNAL OF MAGNETIC RESONANCE IMAGING》 * |
| 刘木华 等: "基于模糊C均值聚类的牛肉图像中脂肪和肌肉区域分割技术", 《农业工程学报》 * |
| 吴水才 等: "腿部磁共振图像自动分割算法研究", 《北京工业大学学报》 * |
| 梅媚 等: "基于CT图像的势函数聚类分割人体脂肪检测", 《天津大学学报》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107705305A (en) * | 2017-10-20 | 2018-02-16 | 上海联影医疗科技有限公司 | A kind of method and device of Medical Image Processing |
| CN109523561A (en) * | 2018-12-19 | 2019-03-26 | 睿仁佳心(上海)科技有限公司 | Muscle and fat image dividing method in automatic abdomen |
| CN109523561B (en) * | 2018-12-19 | 2021-04-02 | 睿佳(武汉)软件科技有限公司 | Automatic intra-abdominal muscle and fat image segmentation method |
| CN110179488A (en) * | 2019-06-11 | 2019-08-30 | 西北农林科技大学 | A method of it is formed using CT scan data detection pig living body trunk |
| CN110179488B (en) * | 2019-06-11 | 2023-06-02 | 西北农林科技大学 | Method for detecting pig living body carcass composition by using CT scanning data |
| CN110415246A (en) * | 2019-08-06 | 2019-11-05 | 东北大学 | A kind of analysis method of stomach fat ingredient |
| CN112168168A (en) * | 2020-10-14 | 2021-01-05 | 北京赛迈特锐医疗科技有限公司 | System and method for automatically and quantitatively evaluating whole body fat by using MR (magnetic resonance) technology |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105427273B (en) | 2018-06-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6329697B2 (en) | Method for automatic segmentation and quantification of body tissue | |
| US10039501B2 (en) | Computer-aided diagnosis (CAD) apparatus and method using consecutive medical images | |
| CN105427273A (en) | Abdominal fat segmentation method and device based on nuclear magnetic resonance image | |
| Nitzken et al. | Improving full-cardiac cycle strain estimation from tagged CMR by accurate modeling of 3D image appearance characteristics | |
| CN112885453A (en) | Method and system for identifying pathological changes in subsequent medical images | |
| US9773325B2 (en) | Medical imaging data processing apparatus and method | |
| US11854193B2 (en) | Validity evaluation device for cancer region detection | |
| CN109949288A (en) | Tumor type determines system, method and storage medium | |
| CN104881858A (en) | Method and device for extracting background parenchymal enhancement in breast | |
| Elnakib et al. | New automated Markov–Gibbs random field based framework for myocardial wall viability quantification on agent enhanced cardiac magnetic resonance images | |
| EP2498222A2 (en) | Method and system for regression-based 4D mitral valve segmentation from 2D+T magnetic resonance imaging slices | |
| Gambino et al. | Automatic skull stripping in MRI based on morphological filters and fuzzy c-means segmentation | |
| CN111127373A (en) | Blood vessel image extraction method and device based on local section analysis | |
| Zhang et al. | A prior knowledge-guided, deep learning-based semiautomatic segmentation for complex anatomy on magnetic resonance imaging | |
| CN116309551B (en) | Method, device, electronic equipment and readable medium for determining focus sampling area | |
| CN115661152B (en) | Target development condition analysis method based on model prediction | |
| US20150016701A1 (en) | Pulse sequence-based intensity normalization and contrast synthesis for magnetic resonance imaging | |
| Jain et al. | An adaptive hybrid technique for pancreas segmentation using CT image sequences | |
| Qi et al. | Blind image quality assessment for MRI with a deep three-dimensional content-adaptive hyper-network | |
| KR101126223B1 (en) | Liver segmentation method using MR images | |
| CN107563992B (en) | Method and device for detecting breast skin line | |
| CN114821049A (en) | Method, device and equipment for segmenting image and storage medium | |
| Woźniak et al. | Segmentation of 3D magnetic resonance brain vessel images based on level set approaches | |
| Campbell et al. | 3-dimensional throat region segmentation from MRI data based on fourier interpolation and 3-dimensional level set methods | |
| Dhore et al. | Chest x-ray segmentation using watershed and super pixel segmentation technique |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |