CN105844617A - Brain parenchyma segmentation realization based on improved threshold segmentation algorithm - Google Patents
Brain parenchyma segmentation realization based on improved threshold segmentation algorithm Download PDFInfo
- Publication number
- CN105844617A CN105844617A CN201610154185.0A CN201610154185A CN105844617A CN 105844617 A CN105844617 A CN 105844617A CN 201610154185 A CN201610154185 A CN 201610154185A CN 105844617 A CN105844617 A CN 105844617A
- Authority
- CN
- China
- Prior art keywords
- segmentation
- image
- brain
- threshold
- value
- 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
Links
Classifications
-
- 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
- G06T2207/30016—Brain
Abstract
In the invention, through an improved threshold segmentation algorithm, segmentation of a grey matter and a white matter of a brain image is realized, which is good for auxiliary diagnosis of brain atrophy of a patient. The method is characterized by firstly, using an adaptive filter to process an original image, reducing a noise of the original image and enhancing a contrast ratio of the image; then, for the processed image, using a region growing algorithm to carry out brain decollement on the image, rejecting a non-parenchymal portion; and then using an iterative threshold method to calculate a threshold and using a near weighted value algorithm to carry out binarization segmentation on the image so that an obtained result is a segmentation result of a brain white matter; finally, based on a result of the previous step, carrying out segmentation so that the obtained result is a segmentation result of a brain grey matter.
Description
Technical field
The invention belongs to computer field of medical image processing, more specifically, relate to a kind of based on changing
The Threshold segmentation the entered dividing method to brain essence, in order to assist the diagnosis to patient's brain atrophy.
Background technology
Brain atrophy drastically influence the healthy of middle-aged and elderly people and quality of life.Statistics shows, existing
The mortality rate of all kinds of encephalatrophies more and more higher.Therefore the diagnoses and treatment to brain atrophy becomes increasingly
Important.And the segmentation to brain essence is that the brain atrophy state of an illness to diagnosis patient has very important reference
Meaning.But, there is many and have a strong impact on the noise of segmentation in medical image itself, and alba and brain are grey
The gray value difference of matter itself is not that these reasons the biggest etc. cause the segmentation of brain essence to become the most difficult,
Currently without the brain essence in a good partitioning algorithm energy well divided ownership medical image.So it is clinical
Mostly to be manually divided into master, but artificial segmentation is wasted time and energy, and has increased the weight of the work load of medical worker.
The most automatically being partitioned into brain essence, being that computer field of medical image processing one is very important grinds
Study carefully problem.
Clinical brain essence is split to be manually divided into master, but although artificial segmentation accurately, excessively takes
Thing, becomes the most meaningful so automatically splitting with computer to replace manually splitting.Currently in medical science
The method of image segmentation is the most, mainly has Threshold segmentation, region growing, fuzzy partition, feature extraction to divide
Class segmentation etc..Every kind of method has respective advantage and defect, and this causes not having a kind of good partitioning algorithm
Can divided ownership medical image the most accurately, and also neither one standard goes to pass judgment on the quality of segmentation result.
In Threshold segmentation, the difficult point of segmentation is choosing of threshold value, fact proved, the selection of threshold value appropriate with
The no effect to segmentation plays conclusive effect.Common threshold segmentation method have bimodal, iterative method, big
Tianjin method (OTSU method), Kirsh operator etc., the fundamental difference of these algorithms is the side choosing threshold value
Method is different.The principle of Two-peak method and simple: it considers that image is made up of foreground and background, at intensity histogram
On figure, front and back two scapes all form peak, are exactly the threshold value place of image at the lowest trough between bimodal.From
From the point of view of the effect of segmentation, when the contrast of current background is more strong, segmentation effect is preferable；The most invalid.
Iterative method is based on the thought approached, iterative computation.The image effect of the Threshold segmentation of iteration gained is good.
Threshold value based on iteration can distinguish the main region place of the foreground and background of image, but trickle at image
Place does not also have good discrimination.But it is surprising that to some particular image, the change of small data
But can cause the huge change of segmentation effect, both data simply vary slightly, but segmentation effect contrast
Greatly.Da-Jin algorithm is proposed in 1979 by big Tianjin, is the segmentation of prospect and background to image Image, note t
Threshold value, prospect is counted and accounted for image scaled is w0, and average gray is u0；Background is counted and accounted for image scaled is w1,
Average gray is u1.The grand mean gray scale of image is: u=w0*u0+w1*u1.From minimum gradation value to maximum
Gray value traversal t, when t makes value g=w0* (u0-u) 2+w1* (u1-u) 2 maximum, t is the optimal threshold of segmentation
Value.Da-Jin algorithm can be understood as follows: this formula is actually inter-class variance value, the prospect that threshold value t is partitioned into
Entire image is constituted with background two parts, and prospect value u0, probability is w0, background value u1, generally
Rate is w1, and grand mean is u, i.e. obtains this formula according to the definition of variance.Because variance is intensity profile uniformity
A kind of tolerance, variance yields is the biggest, illustrates that two parts difference of pie graph picture is the biggest, when partial target mistake is divided into background
Or part background mistake is divided into target that two parts difference all can be caused to diminish, therefore make the segmentation that inter-class variance is maximum
Mean that misclassification probability is minimum.Its thought of Kirsh operator is: each pixel i to digital picture, it is considered to it
The gray value of eight adjoint points, deduct the weighted sum of remaining five adjoint points with the weighted sum of wherein three consecutive points
Obtain difference, make three adjoint points constantly shift around this pixel, take the maximum of these eight differences as Kirsh
Operator.That is: setting Si as three adjoint point sums, Ti is five adjoint point sums, then Kirsh operator definitions is
K (i)=max{1, max (5Si-3Ti) } as taken threshold value THk, then as K (i) > THk time, pixel i is step limit
Edge point.
Summary of the invention
It is an object of the invention to design a kind of brain essence bag that can be preferably partitioned in medical image
The method including grey matter and white matter.During utilizing threshold value to split, examine when processing each pixel
The size of the gray value and threshold value of considering this pixel and pixel about compares, only more than or
Just think that equal to threshold value this pixel belongs to part to be split, otherwise it is assumed that this pixel belongs to background parts,
Gray value zero setting.
Preferably realize above-mentioned segmentation purpose, need image is carried out pretreatment, mainly include following in
Hold: first with sef-adapting filter, image is carried out pretreatment, mainly remove in medical science brain image and exist
Noise, and strengthen the contrast of image, after being beneficial to before Threshold segmentation.Then region growing is utilized to calculate
Method carries out brain stripping to image, removes non-brain substantial portion such as such as skin, bone etc..Subsequently utilize threshold value
Partitioning algorithm carries out the segmentation of alba to image, finally again with Threshold Segmentation Algorithm, image is carried out brain
The segmentation of grey matter.
Know-why is as it is shown in figure 1, concrete techniqueflow is as follows:
Step one: process original image first with sef-adapting filter, reduces the noise of original image and increases
The contrast of strong image；
Step 2: the image recycling algorithm of region growing processing step one carries out brain stripping to image
From, reject non-brain substantial portion；
Step 3: utilize iteration method to ask for threshold value, then utilizes and closes on weighted value algorithm and enter image
Row binarization segmentation, the result obtained is the segmentation result of alba；
Step 4: the most once split in the result of step 3, the result obtained is ectocinereal
Segmentation result；
Accompanying drawing explanation
Fig. 1 is that the present invention realizes the technology of the dividing method to brain essence based on the Threshold Segmentation Algorithm improved
Conceptual scheme.
Detailed description of the invention
Below in conjunction with the accompanying drawings the detailed description of the invention of the present invention is described, in order to those skilled in the art
Member is more fully understood that the present invention.Requiring particular attention is that, in the following description, may desalination and
Ignore the content introduction of the known function relevant with the present invention and design.
In the present embodiment, the present invention mainly includes following link to brain essence dividing method: 1. image
Pretreatment, 2. brain image are peeled off, are 3. asked for threshold value, 4. Threshold segmentation.
Image semantic classification uses sef-adapting filter, and brain image is peeled off and used algorithm of region growing, threshold value
Asking for using iterative method to ask for, its step is as follows: 1 obtains maximum gradation value and the minimum gradation value of image, point
It is not designated as ZMAX and ZMIN, makes initial threshold T0=(ZMAX+ZMIN)/2；2 will according to threshold value TK
Image is foreground and background, obtains both average gray value ZO and ZB respectively；3 obtain new threshold value
TK+1=(ZO+ZB)/2；If 4 TK=TK+1, then gained is threshold value；Otherwise turn 2, iterative computation.
The process of Threshold segmentation then uses the weighted value algorithm that closes on of innovation below will be to closing on weighted value algorithm
Explain.Each pixel, during segmentation, is being split by general Threshold Segmentation Algorithm
When, it is that the size of gray value by this point directly compares with the size of threshold value, according to the knot compared
Fruit determines whether belong to part to be split or belong to background parts.So result of segmentation is antinoise
Effect can be the poorest, because being this point of simple consideration splitting each pixel when
Gray value size, and whether and to leave it unattended be noise.Based on this reason, close on weighted value algorithm pair
The process of segmentation carries out some and improves, and certain pixel is designated as A, (it is entered the when of j) segmentation by i
One function of row processes
I.e. F (i, j)=(3f1(i,j)+2f2(i,j)+f3(i,j)+p(i,j))/16
Wherein f1(i, j) represent pixel (i, ground floor partial pixel point gray value j) and i.e.
f1(i, j)=p (i+1, j)+p (i+1, j+1)+p (i, j+1), in like manner f2(i, j) represents the sum of second layer partial pixel point gray value,
f3(i j) represents the sum of third layer partial pixel point gray value.
By result F that obtains, (i, if j) comparing F with the size of threshold value TK, (i, j) >=TK then judges A
Point is alba part, and its gray value is set to 100.The most then judge A point as background (grey matter) part,
Its gray value is set to 0.After all of pixel binary conversion treatment, the result obtained is exactly alba
The segmentation result of segmentation.Ectocinereal segmentation is just the same, it is simply that is negated by its Rule of judgment and can (note
Gray value is greater than zero.Because image background gray value is zero).Because after brain is peeled off, brain image only ash
Matter, white matter part, the supplementary set of white matter is grey matter.
The present invention a kind of based on the Threshold Segmentation Algorithm improved realize the dividing method of brain essence is had with
Lower feature:
The present invention proposes a set of new brain essence to be carried out dividing method, can carry out brain essence accurately
Segmentation.The weighted value algorithm that closes on of innovation carries out some improvement can effectively reduce noise to the process of segmentation
The impact that segmentation result is produced.The result of brain essence segmentation can assist doctor to patient's brain atrophy state of an illness
Diagnosis.
Although detailed description of the invention illustrative to the present invention is described above, in order to this technology is led
Artisans understand that the present invention, it should be apparent that the invention is not restricted to the scope of detailed description of the invention, right
From the point of view of those skilled in the art, as long as various change limits in appended claim and determines
The spirit and scope of the present invention in, these changes are apparent from, all utilize present inventive concept send out
Bright creation is all at the row of protection.
Claims (2)
1. the present invention is a kind of dividing method realized based on Threshold Segmentation Algorithm brain essence, mainly wraps
Include herein below: first image is carried out pretreatment, then utilize the Threshold Segmentation Algorithm of a kind of improvement to realize
Segmentation (including grey matter, white matter) to brain essence.
Technical scheme is as follows:
Step one: process image first with sef-adapting filter, to eliminate present in image
The noise of impact segmentation, and strengthen the contrast of grey matter, white matter, be conducive to improving the accuracy of segmentation.
Step 2: the picture utilizing step one to handle well carries out brain image stripping, removes beyond brain essence
Part, such as skin, bone etc..Brain image is peeled off algorithm and is used algorithm of region growing.
Step 3: utilize iterative method selected threshold, according to threshold value and combine the pixel around each pixel
Image is split by the gray value of point, and the result of segmentation is alba.
Step 4: re-use iterative method selected threshold, according to threshold value and combine around each pixel
Image is split by the gray value of pixel, and this time the result of segmentation is ectocinerea.
A kind of segmentation realized based on Threshold Segmentation Algorithm brain essence the most according to claim 1
Method, it is characterised in that during using Threshold segmentation, when each pixel is processed, and
It not simple comparing with threshold value, but combine the gray scale of the gray value of self and the pixel of its periphery
Value compares again.
The process splitting brain essence that is mainly characterized by of the present invention carries out some improvement.To improve
The accuracy of segmentation and anti-interference.Specifically include that (1) utilizes sef-adapting filter that image is carried out pre-place
Reason, (2) utilize algorithm of region growing that filtered image is carried out brain stripping.(3) iterative method is utilized to choose
Threshold value, carries out the segmentation of ectocinerea and alba respectively.
Utilizing iterative method to choose and take threshold value, its step is as follows:
1. obtain maximum gradation value and the minimum gradation value of image, be designated as ZMAX and ZMIN respectively,
Make initial threshold T0=(ZMAX+ZMIN)/2；
2. it is foreground and background according to threshold value TK by Image, obtains both average gray respectively
Value ZO and ZB；
3. obtain new threshold value TK+1=(ZO+ZB)/2；
4. if TK=TK+1, then gained is threshold value；Otherwise turn 2, iterative computation.
When utilizing algorithm of region growing that image carries out brain stripping, brain essence is chosen a seed points,
Choose a least threshold value of gray value, utilize the feature of algorithm of region growing just can split easily
Go out brain substantial portion.
In the cutting procedure of hindbrain essence, when processing certain pixel, (i, j), we use F
(i, j) processes,
F (i, j)=F (i, j)=(3f1(i,j)+2f2(i,j)+f3(i,j)+p(i,j))/16
Wherein f1(i, j) represent pixel (i, ground floor partial pixel point gray value j) and i.e.
f1(i, j)=p (i+1, j)+p (i+1, j+1)+p (i, j+1), in like manner f2(i, j) represents the sum of second layer partial pixel point gray value,
f3(i j) represents the sum of third layer partial pixel point gray value.
When F (i, j) >=TK P in season (and i, j)=100, otherwise P (i, j)=0.Wherein P (i, j) be pixel (i, j)
Gray value, TK is threshold value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610154185.0A CN105844617A (en) | 2016-03-17 | 2016-03-17 | Brain parenchyma segmentation realization based on improved threshold segmentation algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610154185.0A CN105844617A (en) | 2016-03-17 | 2016-03-17 | Brain parenchyma segmentation realization based on improved threshold segmentation algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105844617A true CN105844617A (en) | 2016-08-10 |
Family
ID=56587271
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610154185.0A Pending CN105844617A (en) | 2016-03-17 | 2016-03-17 | Brain parenchyma segmentation realization based on improved threshold segmentation algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105844617A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107103612A (en) * | 2017-03-28 | 2017-08-29 | 深圳博脑医疗科技有限公司 | Automate the quantitative calculation method of subregion brain atrophy |
| CN108765447A (en) * | 2018-04-26 | 2018-11-06 | 深圳博脑医疗科技有限公司 | A kind of image partition method, image segmentation device and electronic equipment |
| CN112150481A (en) * | 2020-09-17 | 2020-12-29 | 南京信息工程大学 | Powdery mildew image segmentation method |
| WO2021027240A1 (en) * | 2019-08-09 | 2021-02-18 | 上海依智医疗技术有限公司 | Brain atrophy identification method, and device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101756712A (en) * | 2008-11-19 | 2010-06-30 | 李祥 | Research of volume measuring method for CT (Computer Tomography) image intracranial hematoma |
| US8200013B2 (en) * | 2004-06-11 | 2012-06-12 | Universitat Bremen | Method and device for segmenting a digital cell image |
| CN103942780A (en) * | 2014-03-27 | 2014-07-23 | 北京工业大学 | Fuzzy-connectedness-algorithm-based segmentation method of thalamus and substructures of thalamus |
| CN104143190A (en) * | 2014-07-24 | 2014-11-12 | 东软集团股份有限公司 | Method and system for partitioning construction in CT image |
| CN105005998A (en) * | 2015-08-05 | 2015-10-28 | 大连理工大学 | Cerebrovascular image segmentation method based on multi-angle serialized image space feature point set |
-
2016
- 2016-03-17 CN CN201610154185.0A patent/CN105844617A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8200013B2 (en) * | 2004-06-11 | 2012-06-12 | Universitat Bremen | Method and device for segmenting a digital cell image |
| CN101756712A (en) * | 2008-11-19 | 2010-06-30 | 李祥 | Research of volume measuring method for CT (Computer Tomography) image intracranial hematoma |
| CN103942780A (en) * | 2014-03-27 | 2014-07-23 | 北京工业大学 | Fuzzy-connectedness-algorithm-based segmentation method of thalamus and substructures of thalamus |
| CN104143190A (en) * | 2014-07-24 | 2014-11-12 | 东软集团股份有限公司 | Method and system for partitioning construction in CT image |
| CN105005998A (en) * | 2015-08-05 | 2015-10-28 | 大连理工大学 | Cerebrovascular image segmentation method based on multi-angle serialized image space feature point set |
Non-Patent Citations (6)
| Title |
|---|
| QIUQIUSWEET: "阈值分割算法", 《HTTP://BLOG.SINA.COM.CN/S.BLOG_AB584CAC01014YG5.HTML》 * |
| 杨清滨: "超声血管图像滤波和分割方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 蔡道霖 等: "颅脑CT图像分割方法研究", 《临床医学工程》 * |
| 陈亮亮: "MRI大脑图像灰质与白质的分割", 《北京生物医学工程》 * |
| 陈通: "基于区域生长算法的MR 脑组织图像半自动提取方法", 《科技资讯》 * |
| 魏巍 等: "三维最大Renyi熵的灰度图像阈值分割算法", 《吉林大学学报（工学版）》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107103612A (en) * | 2017-03-28 | 2017-08-29 | 深圳博脑医疗科技有限公司 | Automate the quantitative calculation method of subregion brain atrophy |
| WO2018176985A1 (en) * | 2017-03-28 | 2018-10-04 | 深圳博脑医疗科技有限公司 | Quantitative calculation method for level of brain atrophy based on automatic segmentation |
| CN107103612B (en) * | 2017-03-28 | 2018-12-07 | 深圳博脑医疗科技有限公司 | Automate the quantitative calculation method of subregion brain atrophy |
| CN108765447A (en) * | 2018-04-26 | 2018-11-06 | 深圳博脑医疗科技有限公司 | A kind of image partition method, image segmentation device and electronic equipment |
| CN108765447B (en) * | 2018-04-26 | 2021-02-12 | 深圳博脑医疗科技有限公司 | Image segmentation method, image segmentation device and electronic equipment |
| WO2021027240A1 (en) * | 2019-08-09 | 2021-02-18 | 上海依智医疗技术有限公司 | Brain atrophy identification method, and device |
| CN112150481A (en) * | 2020-09-17 | 2020-12-29 | 南京信息工程大学 | Powdery mildew image segmentation method |
| CN112150481B (en) * | 2020-09-17 | 2023-07-18 | 南京信息工程大学 | Powdery mildew image segmentation method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | A composite model of wound segmentation based on traditional methods and deep neural networks | |
| Sreedhar et al. | A comparative study of melanoma skin cancer detection in traditional and current image processing techniques | |
| CN105844617A (en) | Brain parenchyma segmentation realization based on improved threshold segmentation algorithm | |
| Ignatious et al. | Computer aided lung cancer detection system | |
| CN105761260B (en) | A kind of skin image affected part dividing method | |
| Santy et al. | Segmentation methods for computer aided melanoma detection | |
| Indraswari et al. | Teeth segmentation on dental panoramic radiographs using decimation-free directional filter bank thresholding and multistage adaptive thresholding | |
| CN105719275A (en) | Parallel combination image defect segmentation method | |
| CN110738637B (en) | Automatic classification system for breast cancer pathological sections | |
| CN111882561A (en) | Cancer cell identification and diagnosis system | |
| CN103020953A (en) | Segmenting method of fingerprint image | |
| Lobo et al. | Classification and segmentation techniques for detection of lung cancer from CT images | |
| CN104182965B (en) | The method of chest muscle is split in a kind of galactophore image | |
| CN103971367A (en) | Hydrologic data image segmenting method | |
| CN110473215A (en) | A kind of image partition method for overhead distribution monitoring scene | |
| CN108765431B (en) | Image segmentation method and application thereof in medical field | |
| CN106780535A (en) | A kind of gray level image processing method | |
| CN112381084B (en) | Automatic contour recognition method for tomographic image | |
| CN101404062A (en) | Automatic screening method for digital galactophore image based on decision tree | |
| Jassim | Hybrid image segmentation using discerner cluster in FCM and histogram thresholding | |
| Vijayakumari et al. | Dental cyst diagnosis using texture analysis | |
| Dhamodharan et al. | Background preserved and feature-oriented contrast improvement using weighted cumulative distribution function for digital mammograms | |
| Singh et al. | Segmentation and characterization of breast tumor in mammograms | |
| Lee et al. | Enhancement of blood vessels in retinal imaging using the nonsubsampled contourlet transform | |
| CN112102350A (en) | Secondary image segmentation method based on Otsu and Tsallis entropy |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160810 |