CN103946868A - Processing method and system for medical images - Google Patents

Processing method and system for medical images Download PDF

Info

Publication number
CN103946868A
CN103946868A CN201380003067.4A CN201380003067A CN103946868A CN 103946868 A CN103946868 A CN 103946868A CN 201380003067 A CN201380003067 A CN 201380003067A CN 103946868 A CN103946868 A CN 103946868A
Authority
CN
China
Prior art keywords
pixel
medical image
spectral intensity
intensity value
popping
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.)
Pending
Application number
CN201380003067.4A
Other languages
Chinese (zh)
Inventor
黄勃
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Publication of CN103946868A publication Critical patent/CN103946868A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20092—Interactive image processing based on input by user
    • G06T2207/20101—Interactive definition of point of interest, landmark or seed
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing

Abstract

The present invention discloses a processing method and system for medical images of a tongue, which comprises following steps. The first step: manually pick up more than one prix from a required medical image; the second step: enlarge the picked prixes; the third step: draw together the common areas of each processed and enlarged prix. The invention overcomes the problem seen in current method that only coating on the tongue can be separated from tongue nature and fails to half-automatically draws the tongue parts out.

Description

A kind of medical image processing method and system
Technical field
The present invention relates to a kind of image treatment method, especially a kind of medical image processing method.
Background technology
In computing machine tongue iconography, in order to impart knowledge to students and the needs of scientific research, often need by the tongue fur extracting section of the tongue spectral intensity value of the same race in tongue digitized video out, use for subsequent treatment, but in tongue image, tongue fur part is often distributed with the tongue fur of multiple spectral intensity value, and border between different tongue furs is very fuzzy, be difficult to clearly make a distinction, present stage only has the method that tongue fur and tongue nature part are made a distinction, and does not but have semi-automatic by the tongue fur extracting section of spectral intensity value of the same race method out.
Summary of the invention
The object of the invention is to propose a kind of medical image processing method, it can solve in computing machine tongue iconography, need to be by the tongue fur extracting section of the spectral intensity value of the same race in tongue digitized video problem out.
In order to achieve the above object, the technical solution adopted in the present invention is as follows:
A kind of medical image processing method, is characterized in that comprising the following steps:
The first step, manually chooses 1 above pixel in tongue digitized video desired zone at random;
Second step, carries out pixel to selected each pixel and expands processing;
The 3rd step, extracts each pixel and expands common region after treatment as selected areas through pixel.
Preferably, described pixel expands to be processed, and comprises the following steps: each pixel of manually choosing is at random stacked,
The 1st step, stack top pixel is popped;
The 2nd step, is designated " choosing " by the zone bit of the pixel of popping;
The 3rd step, check in order the neighborhood territory pixel with the pixel of popping, if wherein certain pixel meets boundary condition and its zone bit is not designated " choosing ", stacked this pixel, and be " choosing " this pixel logo, otherwise continue to search next neighborhood territory pixel, all inspecteds of the neighborhood territory pixel of pixel until this is popped;
The 4th step, if stack non-NULL repeats the 1st step, if stack is empty, the pixel that all zone bits are designated " choosing " is that pixel expands region after treatment.
Preferably, described boundary condition be pop the spectral intensity value of pixel and the spectral intensity value of its neighborhood territory pixel squared difference and be less than respectively boundary threshold,
(P’ h-P h) 2＜Thres h&&(P’ s-P s) 2＜Thres s&&(P’ v-P v) 2＜Thres v，
P ' wherein h, P h, Thres hfor medical image is at the spectral intensity value of the pixel of popping of HSV spectral intensity value space h component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel, P ' wherein s, P s, Thres sfor medical image is at the spectral intensity value of the pixel of popping of HSV spectral intensity value space s component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel, P ' wherein v, P v, Thres vfor medical image is in the spectral intensity value of the pixel of popping of HSV spectral intensity value space v component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel.
Preferably, described boundary threshold Thres xthe computing method of (the h component that x is medical image or s component or v component) are as follows:
A step, obtains the peaked gray level Lv that this medical image divides the relative entropy of spirogram full figure, and this medical image of Lv place correspondence divides the histogrammic value Hv of spirogram, and wherein relative entropy is defined as
Entropy = Σ i = 1 N p i ln p i q i + Σ i = 1 N q i ln q i p i ,
P 1, p 2..., p nand q 1, q 2..., q nit is the probability distribution of tongue fur and tongue nature pixel;
B step, calculates respectively this medical image and divides the histogrammic Zuo Feng of spirogram and gray level La, the Lb at right peak place, and this medical image of La, Lb place correspondence divides histogrammic value Ha, the Hb of spirogram;
C step, if (Ha-Hv)/(Lv-La) > (Hb-Hv)/(Lb-Lv),
Thres x＝E*(Hb-Hv)/(Lb-Lv)，
Otherwise,
Thres x＝E*(Ha-Hv)/(Lv-La)。
Another object of the embodiment of the present invention is to provide a kind of medical image processing system, it is characterized in that comprising:
Multiple spot pixel is chosen module, for manually choosing at random 1 above pixel of tongue digitized video desired zone;
Pixel expands processing module, expands processing for selected each pixel being carried out to pixel;
Common region extraction module, expands common region after treatment as selected areas for extracting each pixel through pixel.
Preferably, described pixel expands processing module, comprising:
The stacked module of selected pixels, stacked for each pixel of manually choosing at random;
The stack top pixel module of popping, for popping stack top pixel;
The pixel logo module of popping, for being designated " choosing " by the zone bit of the pixel of popping;
Neighborhood territory pixel processing module, for checking in order and the neighborhood territory pixel of the pixel of popping, if wherein certain pixel meets boundary condition and its zone bit is not designated " choosing ", stacked this pixel, and be " choosing " this pixel logo, otherwise continue to search next neighborhood territory pixel, all inspecteds of the neighborhood territory pixel of pixel until this is popped;
Stack non-NULL judge module, for judging that whether stack is empty, if stack non-NULL jumps to the stack top pixel module of popping, if stack be sky, jumps to end module;
Finishing module, is that pixel expands region after treatment for all zone bits being designated to the pixel of " choosing ".
The present invention has following beneficial effect:
This medical image processing method can extract the tongue fur of the spectral intensity value of the same race in computing machine tongue iconography accurately, is conducive to subsequent treatment and uses.
Brief description of the drawings
Fig. 1 is the functional-block diagram of medical image processing method of the present invention;
Fig. 2 is the functional-block diagram that the pixel of medical image processing method of the present invention expands treatment step;
Fig. 3 is the functional-block diagram of the boundary threshold computing method of medical image processing method of the present invention;
Fig. 4 is the structural representation of medical image processing system of the present invention;
Fig. 5 is that the pixel of medical image processing system of the present invention expands the structural representation of processing module.
Embodiment
Below, by reference to the accompanying drawings and embodiment, the present invention is described further.
In computing machine tongue iconography, in order to impart knowledge to students and the needs of scientific research, often need by the tongue fur extracting section of the tongue spectral intensity value of the same race in tongue digitized video out, use for subsequent treatment, but in tongue image, tongue fur part is often distributed with the tongue fur of multiple spectral intensity value, and border between different tongue furs is very fuzzy, be difficult to clearly make a distinction, present stage only has the method that tongue fur and tongue nature part are made a distinction, and does not but have semi-automatic by the tongue fur extracting section of spectral intensity value of the same race method out.
By the tongue fur extracting section of spectral intensity value of the same race method out, be in fact equivalent to carry out two-value Image Segmentation one time.So-called two-value image, just refer to institute on image gray-scale value a little only use two kinds may, be that " 0 " is just " 255 ", namely whole image presents obvious black and white effect.In order to obtain desirable two-value image, generally adopt Threshold sementation, it object and background are had strong contrast image cut apart effective especially, the not overlapping region of boundary definition that its calculates simple and sealing for total energy, is communicated with.The pixel that all gray scales are more than or equal to threshold value is judged as and belongs to object, and gray-scale value represents with " 255 ", otherwise these pixels are excluded beyond object area, and gray-scale value be " 0 ", expression background.So the border of object just becomes the set of the point of some inside like this, and these points all have at least an adjoint point not belong to this object.If interested object has the gray-scale value of uniformity in inside, and it is in a homogeneous background with another one gray-scale value, uses threshold method can obtain reasonable effect.If object, with the difference (such as texture difference) not on gray-scale value of background, can be converted to this character the difference of gray scale, then utilize thresholding technology to cut apart this image.In order to make to cut apart robust more, applicability is stronger, and system should be selected threshold value automatically.Image Segmentation algorithm based on knowledge such as object, environment and application domains has more ubiquity and adaptability than the algorithm based on fixed threshold.These knowledge comprise: corresponding to the size of the image greyscale characteristic of object, object, object quantity of dissimilar object etc. in shared ratio, image in image.
But in tongue image, tongue fur part is often distributed with the tongue fur of multiple spectral intensity value, and border between different tongue furs is very fuzzy, be difficult to clearly make a distinction, therefore we consider the result of two-value Image Segmentation repeatedly to integrate, and get its common region, as extracting result, its result is more accurate, avoids introducing unnecessary noise to subsequent step.
Therefore, as shown in Figure 1, be a kind of medical image processing method realization flow that the embodiment of the present invention provides, details are as follows:
The first step, manually chooses 1 above pixel in tongue digitized video desired zone at random; Tongue fur part is often distributed with the tongue fur of multiple spectral intensity value, and border between different tongue furs is very fuzzy, be difficult to clearly make a distinction, for each pixel, follow-up second step and the 3rd step are processed the region of extracting and are had nothing in common with each other, but its public part is basicly stable, are also the regions that needs extraction, so need manually to choose at random multiple pixels in tongue digitized video desired zone in the first step, in order to subsequent step application.
Second step, carries out pixel to selected each pixel and expands processing;
Wherein, pixel expands to be processed, and comprises the following steps: each pixel of manually choosing is at random stacked,
The 1st step, stack top pixel is popped;
The 2nd step, is designated " choosing " by the zone bit of the pixel of popping;
The 3rd step, check in order the neighborhood territory pixel with the pixel of popping, if wherein certain pixel meets boundary condition and its zone bit is not designated " choosing ", stacked this pixel, and be " choosing " this pixel logo, otherwise continue to search next neighborhood territory pixel, all inspecteds of the neighborhood territory pixel of pixel until this is popped;
The 4th step, if stack non-NULL repeats the 1st step, if stack is empty, the pixel that all zone bits are designated " choosing " is that pixel expands region after treatment.
Such method, has reduced the stack-incoming operation of popping of classic method, has improved the efficiency of method.
Here boundary condition be pop the spectral intensity value of pixel and the spectral intensity value of its neighborhood territory pixel squared difference and be less than respectively boundary threshold,
(P’ h-P h) 2＜Thres h&&(P’ s-P s) 2＜Thres s&&(P’ v-P v) 2＜Thres v，
P ' wherein h, P h, Thres hfor medical image is at the spectral intensity value of the pixel of popping of HSV spectral intensity value space h component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel, P ' wherein s, P s, Thres sfor medical image is at the spectral intensity value of the pixel of popping of HSV spectral intensity value space s component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel, P ' wherein v, P v, Thres vfor medical image is in the spectral intensity value of the pixel of popping of HSV spectral intensity value space v component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel.
HSV model was founded by Ai Erweilei Smith in 1978, and it is a kind of nonlinear transformation of primaries pattern.
Form and aspect (H) are the base attributes of color, are exactly usual said color names, as red, yellow etc.Saturation degree (S) refers to the purity of color, and higher color is purer, and low change is grey gradually, gets the numerical value of 0-100%.Lightness (V), gets 0-100%.
HSV is the point in cylindrical-coordinate system color description, the central shaft value of this cylinder is the white to top and the grey that is from the black of bottom in the middle of their, around the angle of this axle corresponding to " form and aspect ", arrive the distance of this axle corresponding to " saturation degree ", and along the height of this axle corresponding to " brightness ", " tone " or " lightness ".
HSV model is generally used in computer graphics application.Must select a color to be applied in the various applied environments of specific graphic elements user, often use HSV colour wheel.Therein, form and aspect are expressed as annulus; Can with one independently triangle represent saturation degree and lightness.Typically, this leg-of-mutton Z-axis instruction saturation degree, and transverse axis represents lightness.In this manner, select color can first in annulus, select form and aspect, in saturation degree and the lightness selecting to want from triangle.
The another kind of visual method of HSV model is cone.In this expression, form and aspect are represented as the angle around circular cone central shaft, and saturation degree is represented as the distance from the center of circle of the xsect of circular cone to this point, and lightness is represented as from the center of circle of the xsect of circular cone distance to the limit.Some expression has been used hexagonal pyramid body.This method is more suitable for showing this HSV color space in a single object; But due to its three-dimensional nature, it is not suitable for selecting color in two-dimension computer interface.
HSV color space can also be expressed as the right cylinder that is similar to above-mentioned cone, and form and aspect change along cylindrical excircle, and saturation degree is along the change of distance in the center of circle from xsect, and lightness is along xsect to bottom surface with the distance of end face and changing.This expressing possibility is considered to the more accurate mathematical model of HSV color space; But the saturation degree that can distinguish in practice and the number of levels of form and aspect are along with lightness approaches black and reduces.In addition computing machine is typically stored rgb value by limited accuracy rating; This has retrained precision, adds the restriction of mankind's Color perception, cone is illustrated in most cases more practical.
HSV color space meets the sensation of human eye to color, and people's vision is far better than the sensitivity to color to the sensitivity of brightness, so HSV color space more meets human vision property than rgb color space, is convenient to processing and the identification of color;
In hsv color space, 3 coordinates are independently, the variation that people can the each color component of independent perception;
What hsv color space formed is a uniform color model, adopts linear scale, and between color, sensorial distance is directly proportional to the Euclidean distance of putting in hsv color volume coordinate.
Therefore we adopt the boundary threshold condition of HSV color space to limit the border that pixel expansion is processed.
Can also regard in essence a kind of thresholding method as because extract homochromy tongue fur region again, thresholding method is divided into global threshold method and local thresholding method.So-called Local threshold segmentation method is that raw video is divided into less image, and each sub-image is chosen to corresponding threshold value.After Threshold segmentation, the boundary between adjacent sub-image may produce the uncontinuity of gray level, therefore needs to get rid of by smoothing technique.The conventional method of local threshold method has gray scale difference histogram method, differential histogram method.Although Local threshold segmentation method can be improved segmentation effect, there are several shortcomings:
(1) size of every width sub-image can not be too little, otherwise the result counting is meaningless.
(2) be arbitrarily cutting apart of every width image, if there is a width sub-image just in time to drop on target area or background area, and according to statistics, it cut apart, and perhaps can produce worse result.
(3) local threshold method all will be added up each width sub-image, and speed is slow, is difficult to adapt to the requirement of real-time.
Global threshold dividing method is applied often in image processing, and it adopts fixing Threshold segmentation image in view picture image.Classical threshold value is chosen taking grey level histogram as handling object.According to the difference of threshold value system of selection, can be divided into the method such as Modal Method, iterative threshold value selection.These methods are all to determine the threshold value cut apart taking the histogram of image as research object.Also have in addition inter-class variance thresholding method, Two-dimensional maximum-entropy split plot design, Fuzzy Threshold split plot design, co-occurrence matrix split plot design, region-growing method etc.
For fairly simple image, can suppose that object and background are respectively in different gray levels, image is polluted by zero-mean Gaussian noise, so the intensity profile curve approximation of image is thought to be formed by stacking by two normal distyribution functions, the histogram of image will there will be the peak value of two separation.For such image, segmentation threshold can select the corresponding gray-scale value of trough between histogrammic two crests as the threshold value of cutting apart.This dividing method inevitably there will be by mistake and cuts apart, and the pixel that makes a part originally belong to background is judged as object, and it is background that the one part of pixel that belongs to object can be mistaken as equally.Can prove, in the time that the size of object equates with background, select like this threshold value can make to miss a point probability and reach minimum.In most of the cases, because near the histogram of image pixel trough is very sparse, therefore this method on image to cut apart impact little.This method can be generalized to many object images with different gray averages.
Therefore selected the peaked gray level of the relative entropy of obtaining this medical image full figure here, and wherein relative entropy is defined as
Entropy = Σ i = 1 N p i ln p i q i + Σ i = 1 N q i ln q i p i ,
This is a kind of iterative threshold value selection algorithm, Image Segmentation is become to tongue fur region, tongue nature region two parts, to each gray level in image ask for respectively tongue fur region, tongue nature region two parts relative entropy and, choose make tongue fur region, tongue nature region two parts relative entropy with maximum gray level as the threshold value of cutting apart image.
For given image, because most pixel belongs to target area or background, and the gray level of target and background intra-zone pixel is more even, the gray scale of pixel and the gray level of its neighboring mean value are more or less the same, so the two-dimensional histogram that image is corresponding mainly concentrates on i, near the diagonal line of j plane, and present on the whole state bimodal and a paddy, two peaks correspond respectively to target and background.The shortcoming of classic method is the half-tone information of only having considered pixel, does not consider the spatial information of pixel, so segmentation effect is undesirable in the time that the signal to noise ratio (S/N ratio) of image reduces.Certainly, the gray scale of pixel is the most basic feature, but it is more responsive to noise ratio, for this reason, in the time cutting apart image, can consider again the area information of image, the segment space information that area grayscale feature has comprised image, and will be lower than a gray feature to the sensitivity of noise.And this method has solved this problem just.
Therefore we after deliberation, wherein said boundary threshold Thres xthe computing method of (the h component that x is medical image or s component or v component) are as follows:
A step, obtains the peaked gray level Lv that this medical image divides the relative entropy of spirogram full figure, and this medical image of Lv place correspondence divides the histogrammic value Hv of spirogram, and wherein relative entropy is defined as
Entropy = Σ i = 1 N p i ln p i q i + Σ i = 1 N q i ln q i p i ,
P 1, p 2..., p nand q 1, q 2..., q nit is the probability distribution of tongue fur and tongue nature pixel;
B step, calculates respectively this medical image and divides the histogrammic Zuo Feng of spirogram and gray level La, the Lb at right peak place, and this medical image of La, Lb place correspondence divides histogrammic value Ha, the Hb of spirogram; Wherein histogrammic Zuo Feng and right peak corresponding tongue fur and tongue nature classification respectively.
C step, if (Ha-Hv)/(Lv-La) > (Hb-Hv)/(Lb-Lv),
Thres x＝E*(Hb-Hv)/(Lb-Lv)，
Otherwise,
Thres x＝E*(Ha-Hv)/(Lv-La)。Here E is an empirical constant.
The 3rd step, extracts each pixel and expands common region after treatment as selected areas through pixel.
By above-mentioned method, can accurately the tongue fur of the spectral intensity value of the same race in computing machine tongue iconography be extracted, be conducive to subsequent treatment and use.We test 1210 computing machine tongue images, and three doctors are responsible for artificial demarcation.They are the tcm clinical practice experts who has 20 years professional experiences, and are all the members of relevant authoritative academic institution of China.Because the perception of spectral intensity value is a kind of common impression, and the relation of tongue picture spectral intensity value and disease has the common medical science regularity of distribution, so their suggestion height is consistent.Through test, the accuracy rate of cutting apart is up to 95.29%.
As shown in Figure 4, the architecture principle of a kind of medical image processing system providing for the embodiment of the present invention, details are as follows, comprising:
Multiple spot pixel is chosen module, for manually choosing at random 1 above pixel of tongue digitized video desired zone;
Pixel expands processing module, expands processing for selected each pixel being carried out to pixel;
Common region extraction module, expands common region after treatment as selected areas for extracting each pixel through pixel.
As shown in Figure 5, the pixel of a kind of medical image processing system providing for the embodiment of the present invention expands the architecture principle of processing module, and details are as follows, comprising:
The stacked module of selected pixels, stacked for each pixel of manually choosing at random;
The stack top pixel module of popping, for popping stack top pixel;
The pixel logo module of popping, for being designated " choosing " by the zone bit of the pixel of popping;
Neighborhood territory pixel processing module, for checking in order and the neighborhood territory pixel of the pixel of popping, if wherein certain pixel meets boundary condition and its zone bit is not designated " choosing ", stacked this pixel, and be " choosing " this pixel logo, otherwise continue to search next neighborhood territory pixel, all inspecteds of the neighborhood territory pixel of pixel until this is popped;
Stack non-NULL judge module, for judging that whether stack is empty, if stack non-NULL jumps to the stack top pixel module of popping, if stack be sky, jumps to end module;
Finishing module, is that pixel expands region after treatment for all zone bits being designated to the pixel of " choosing ".
Preferably, described boundary condition be pop the spectral intensity value of pixel and the spectral intensity value of its neighborhood territory pixel squared difference and be less than respectively boundary threshold, (P ' h-P h) 2< Thres haMP.AMp.Amp & (P ' s-P s) 2< Thres saMP.AMp.Amp & (P ' v-P v) 2< Thres v, P ' wherein h, P h, Thres hfor medical image is at the spectral intensity value of the pixel of popping of HSV spectral intensity value space h component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel, P ' wherein s, P s, Thres sfor medical image is at the spectral intensity value of the pixel of popping of HSV spectral intensity value space s component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel, P ' wherein v, P v, Thres vfor medical image is in the spectral intensity value of the pixel of popping of HSV spectral intensity value space v component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel.
Preferably, described boundary threshold Thres xthe computing method of (the h component that x is medical image or s component or v component) are as follows:
A step, obtains the peaked gray level Lv that this medical image divides the relative entropy of spirogram full figure, and this medical image of Lv place correspondence divides the histogrammic value Hv of spirogram, and wherein relative entropy is defined as
Entropy = &Sigma; i = 1 N p i ln p i q i + &Sigma; i = 1 N q i ln q i p i ,
P 1, p 2..., p nand q 1, q 2..., q nit is the probability distribution of tongue fur and tongue nature pixel;
B step, calculates respectively this medical image and divides the histogrammic Zuo Feng of spirogram and gray level La, the Lb at right peak place, and this medical image of La, Lb place correspondence divides histogrammic value Ha, the Hb of spirogram;
C step, if (Ha-Hv)/(Lv-La) > (Hb-Hv)/(Lb-Lv),
Thres x＝E*(Hb-Hv)/(Lb-Lv)，
Otherwise,
Thres x＝E*(Ha-Hv)/(Lv-La)。
For a person skilled in the art, can be according to technical scheme described above and design, make other various corresponding changes and distortion, and these all changes and distortion all should belong to the protection domain of the claims in the present invention within.

Claims (6)

1. a medical image processing method, is characterized in that comprising the following steps:
The first step, manually chooses 1 above pixel in tongue digitized video desired zone at random;
Second step, carries out pixel to selected each pixel and expands processing;
The 3rd step, extracts each pixel and expands common region after treatment as selected areas through pixel.
2. medical image processing method as claimed in claim 1, is characterized in that, described pixel expands to be processed, and comprises the following steps:
The 1st step, each pixel of manually choosing is at random stacked;
The 2nd step, stack top pixel is popped;
The 3rd step, is designated " choosing " by the zone bit of the pixel of popping;
The 4th step, check in order the neighborhood territory pixel with the pixel of popping, if wherein certain pixel meets boundary condition and its zone bit is not designated " choosing ", stacked this pixel, and be " choosing " this pixel logo, otherwise continue to search next neighborhood territory pixel, all inspecteds of the neighborhood territory pixel of pixel until this is popped;
The 5th step, if stack non-NULL repeats the 2nd step, if stack is empty, the pixel that all zone bits are designated " choosing " is that pixel expands region after treatment.
3. medical image processing method as claimed in claim 1, is characterized in that, described boundary condition be pop the spectral intensity value of pixel and the spectral intensity value of its neighborhood territory pixel squared difference and be less than respectively boundary threshold,
(P’ h-P h) 2＜Thres h&&(P’ s-P s) 2＜Thres s&&(P’ v-P v) 2＜Thres v，
P ' wherein h, P h, Thres hfor medical image is at the spectral intensity value of the pixel of popping of HSV spectral intensity value space h component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel, P ' wherein s, P s, Thres sfor medical image is at the spectral intensity value of the pixel of popping of HSV spectral intensity value space s component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel, P ' wherein v, P v, Thres vfor medical image is in the spectral intensity value of the pixel of popping of HSV spectral intensity value space v component, spectral intensity value and the corresponding boundary threshold of its neighborhood territory pixel.
4. medical image processing method as claimed in claim 1, is characterized in that, described boundary threshold Thres xthe computing method of (the h component that x is medical image or s component or v component) are as follows:
A step, obtains the peaked gray level Lv that this medical image divides the relative entropy of spirogram full figure, and this medical image of Lv place correspondence divides the histogrammic value Hv of spirogram, and wherein relative entropy is defined as
Entropy = &Sigma; i = 1 N p i ln p i q i + &Sigma; i = 1 N q i ln q i p i ,
P 1, p 2..., p nand q 1, q 2..., q nit is the probability distribution of tongue fur and tongue nature pixel;
B step, calculates respectively this medical image and divides the histogrammic Zuo Feng of spirogram and gray level La, the Lb at right peak place, and this medical image of La, Lb place correspondence divides histogrammic value Ha, the Hb of spirogram;
C step, if (Ha-Hv)/(Lv-La) > (Hb-Hv)/(Lb-Lv),
Thres x＝E*(Hb-Hv)/(Lb-Lv)，
Otherwise,
Thres x＝E*(Ha-Hv)/(Lv-La)。
5. a medical image processing system, is characterized in that comprising:
Multiple spot pixel is chosen module, for manually choosing at random 1 above pixel of tongue digitized video desired zone;
Pixel expands processing module, expands processing for selected each pixel being carried out to pixel;
Common region extraction module, expands common region after treatment as selected areas for extracting each pixel through pixel.
6. medical image processing system as claimed in claim 5, is characterized in that, described pixel expands processing module, comprising:
The stacked module of selected pixels, stacked for each pixel of manually choosing at random;
The stack top pixel module of popping, for popping stack top pixel;
The pixel logo module of popping, for being designated " choosing " by the zone bit of the pixel of popping;
Neighborhood territory pixel processing module, for checking in order and the neighborhood territory pixel of the pixel of popping, if wherein certain pixel meets boundary condition and its zone bit is not designated " choosing ", stacked this pixel, and be " choosing " this pixel logo, otherwise continue to search next neighborhood territory pixel, all inspecteds of the neighborhood territory pixel of pixel until this is popped;
Stack non-NULL judge module, for judging that whether stack is empty, if stack non-NULL jumps to the stack top pixel module of popping, if stack be sky, jumps to end module;
Finishing module, is that pixel expands region after treatment for all zone bits being designated to the pixel of " choosing ".
CN201380003067.4A 2013-05-13 2013-05-13 Processing method and system for medical images Pending CN103946868A (en)

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2013/075527 WO2014183246A1 (en) 2013-05-13 2013-05-13 Medical image processing method and system

Publications (1)

Publication Number Publication Date
CN103946868A true CN103946868A (en) 2014-07-23

Family

ID=51193112

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201380003067.4A Pending CN103946868A (en) 2013-05-13 2013-05-13 Processing method and system for medical images

Country Status (2)

Country Link
CN (1) CN103946868A (en)
WO (1) WO2014183246A1 (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105869151A (en) * 2016-03-24 2016-08-17 天津大学 Tongue segmentation and tongue coating and tongue body separation method
CN109978895A (en) * 2019-03-29 2019-07-05 北京峰云视觉技术有限公司 A kind of tongue body image partition method and device
CN115953392A (en) * 2023-03-09 2023-04-11 四川博瑞客信息技术有限公司 Tongue body coating quality evaluation method based on artificial intelligence

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105232050B (en) * 2015-09-18 2018-10-19 天津中医药大学 A kind of noninvasive detection device of human body thickness of the tongue coating
CN107610087B (en) * 2017-05-15 2020-04-28 华南理工大学 Tongue coating automatic segmentation method based on deep learning
CN110570515A (en) * 2019-09-03 2019-12-13 天津工业大学 method for carrying out human skeleton three-dimensional modeling by utilizing CT (computed tomography) image
CN113724248B (en) * 2021-09-16 2024-03-26 北京航空航天大学 Medical image needle track detection method and device
CN114913422B (en) * 2022-05-23 2024-03-15 桂林理工大学 Karst mountain area soil erosion and water loss field monitoring device

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101082980A (en) * 2006-06-01 2007-12-05 上海吉量软件科技有限公司 Intersection point matched pair stuffing algorithm
CN101101668A (en) * 2007-07-13 2008-01-09 杭州电子科技大学 Medical image cutting method based on oscillatory network
WO2009038376A1 (en) * 2007-09-21 2009-03-26 Korea Institute Of Oriental Medicine Extraction method of tongue region using graph-based approach and geometric properties
CN102509312A (en) * 2011-09-20 2012-06-20 哈尔滨工业大学 Color range space of human body digital tongue image color and extraction method thereof
CN102737376A (en) * 2012-03-31 2012-10-17 常熟市支塘镇新盛技术咨询服务有限公司 Improved region growing method applied to coronary artery angiography image segmentation
CN102982784A (en) * 2011-09-06 2013-03-20 联咏科技股份有限公司 Image adjusting method

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101082980A (en) * 2006-06-01 2007-12-05 上海吉量软件科技有限公司 Intersection point matched pair stuffing algorithm
CN101101668A (en) * 2007-07-13 2008-01-09 杭州电子科技大学 Medical image cutting method based on oscillatory network
WO2009038376A1 (en) * 2007-09-21 2009-03-26 Korea Institute Of Oriental Medicine Extraction method of tongue region using graph-based approach and geometric properties
CN102982784A (en) * 2011-09-06 2013-03-20 联咏科技股份有限公司 Image adjusting method
CN102509312A (en) * 2011-09-20 2012-06-20 哈尔滨工业大学 Color range space of human body digital tongue image color and extraction method thereof
CN102737376A (en) * 2012-03-31 2012-10-17 常熟市支塘镇新盛技术咨询服务有限公司 Improved region growing method applied to coronary artery angiography image segmentation

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105869151A (en) * 2016-03-24 2016-08-17 天津大学 Tongue segmentation and tongue coating and tongue body separation method
CN105869151B (en) * 2016-03-24 2018-12-25 天津大学 Tongue segmentation and tongue fur tongue nature separation method
CN109978895A (en) * 2019-03-29 2019-07-05 北京峰云视觉技术有限公司 A kind of tongue body image partition method and device
CN109978895B (en) * 2019-03-29 2021-08-13 北京峰云视觉技术有限公司 Tongue image segmentation method and device
CN115953392A (en) * 2023-03-09 2023-04-11 四川博瑞客信息技术有限公司 Tongue body coating quality evaluation method based on artificial intelligence

Also Published As

Publication number Publication date
WO2014183246A1 (en) 2014-11-20

Similar Documents

Publication Publication Date Title
CN103946868A (en) Processing method and system for medical images
CN103208001B (en) In conjunction with shape-adaptive neighborhood and the remote sensing image processing method of texture feature extraction
CN102063623B (en) Method for extracting image region of interest by combining bottom-up and top-down ways
Yang et al. Boundary detection using double-opponency and spatial sparseness constraint
CN103186904B (en) Picture contour extraction method and device
CN104182763B (en) A kind of floristics identifying system based on flower feature
CN108830280A (en) A kind of small target detecting method based on region nomination
CN103914699A (en) Automatic lip gloss image enhancement method based on color space
CN107403183A (en) The intelligent scissor method that conformity goal is detected and image segmentation is integrated
CN108229458A (en) A kind of intelligent flame recognition methods based on motion detection and multi-feature extraction
CN107016403B (en) A method of completed region of the city threshold value is extracted based on nighttime light data
CN107392130A (en) Classification of Multispectral Images method based on threshold adaptive and convolutional neural networks
CN102567731A (en) Extraction method for region of interest
Casanova et al. Texture analysis using fractal descriptors estimated by the mutual interference of color channels
CN107657619A (en) A kind of low-light (level) Forest fire image dividing method
CN104102928B (en) A kind of Classifying Method in Remote Sensing Image based on texture primitive
CN105447274B (en) A method of seashore wetland drawing being carried out to intermediate resolution remote sensing images using object oriented classification technology
CN103793930A (en) Pencil drawing image generation method and device
CN104392233B (en) A kind of image saliency map extracting method based on region
CN107122713B (en) Analog property detection method based on deep learning
CN101853513B (en) Time and space significance visual attention method based on entropy
CN106373086A (en) Augmented reality-based ring 2D online real-time trying method and system
CN107154044A (en) A kind of dividing method of Chinese meal food image
CN108491786A (en) A kind of method for detecting human face based on hierarchical network and Cluster merging
CN104408733A (en) Object random walk-based visual saliency detection method and system for remote sensing image

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20140723

WD01 Invention patent application deemed withdrawn after publication