CN101984916B - Blood vessel diameter measuring method based on digital image processing technology - Google Patents
Blood vessel diameter measuring method based on digital image processing technology Download PDFInfo
- Publication number
- CN101984916B CN101984916B CN2010105472250A CN201010547225A CN101984916B CN 101984916 B CN101984916 B CN 101984916B CN 2010105472250 A CN2010105472250 A CN 2010105472250A CN 201010547225 A CN201010547225 A CN 201010547225A CN 101984916 B CN101984916 B CN 101984916B
- Authority
- CN
- China
- Prior art keywords
- image
- point
- blood vessel
- pixel
- blood vessels
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention provides a blood vessel diameter measuring method based on digital image processing technology. The method includes that: (1) Gauss matched filtering method is utilized to enhance a blood vessel image; (2) the image is normalized and binarized; (3) according to related theory of mathematical morphology, the image is refined, and skeleton is extracted; (4) straight line fitting method is used for solving slope of a section of blood vessel, pixel number of blood vessel is detected along the direction vertical to the blood vessel, and then the blood vessel diameter is solved according to dot pitch. The invention is used for measuring medical image blood vessel diameter, computing efficiency is high, requirement on image quality is low, and computation is accurate.
Description
Technical field
Involved in the present invention is the measuring method of medical image blood vessels caliber, particularly based on the blood vessels caliber method for automatic measurement of digital image processing techniques.
Background technology
In decades, aspect medical science, mainly adopt sciagraphy, take the photograph the sheet method, ophthalmoscope surveys to survey as method and microscope and carry out blood vessels caliber as means such as methods and measure, mostly this type diagnosis is to adopt the mechanism of machinery and optical bond.The artificial subjective error of existing these methods is bigger, can not solve the more definite relation in caliber border technically at all and get rid of the subjective random error of quantization operation.In recent years; Introducing along with Flame Image Process and computer technology; Up to the present some combining screen Display Techniques have appearred; As the somewhere measured zone is calibrated the method that similar " slide calliper rule " of observed pattern are measured with cursor, this technology is mixed with artificial subjective factors naturally, and effect is more coarse.Press for the stronger caliber quantization method of a kind of objectivity clinically as supplementary means, thus computer image processing technology be applied to Calibration become development today trend with research focus.
The blood vessels caliber Measurement Study mainly concentrates on the blood-vessel image processing method both at home and abroad.Classify as technical elements such as pretreatment, cutting techniques, measuring technique.The main purpose that blood-vessel image is cut apart is to extract doctor's interesting areas.External now existing many experts have proposed multiple dividing method to the characteristics of blood-vessel image; These methods are broadly divided into two big types: the first kind belongs to well-regulated method, comprises blood vessel method for tracing, matched filtering method, local auto-adaptive threshold method, topological self adaptation distortion size method, based on the method for mathematical morphology; Second type belongs to the method that supervision is arranged, and comprises pixel classification method based on neutral net, based on blood vessel segmentation method of ridge etc.In addition, there is the scholar to be directed against the similar Gaussian curve of cross section intensity profile of blood vessel, proposed measuring method based on the Gaussian curve model.Like the blood vessel width measure based on two dimensional model of propositions such as J.Lowell, the application of propositions such as L.Gang is revised second order Gauss wave filter amplitude method and is detected and measure blood vessel; Wang Shuzhen etc. measure blood vessel through blood-vessel image being implemented processing methods such as amplification, binaryzation, mathematical morphology processing, refinement, rim detection.
Summary of the invention
The object of the present invention is to provide the higher blood vessels caliber measuring method of a kind of precision based on digital image processing techniques.
The objective of the invention is to realize like this: 1, blood-vessel image is strengthened with the Gauss matched filtering method; 2, with image normalization, binaryzation; 3, according to the correlation theory of mathematical morphology,, extract skeleton with image thinning; 4, obtain the slope of one section blood vessel with the method for fitting a straight line, and detect the number of pixels of blood vessel, try to achieve blood vessels caliber according to a distance again along perpendicular direction.
The invention has the beneficial effects as follows: 1) computational efficiency is high; 2) less demanding to picture quality; 3) calculate accurately.
Description of drawings
Fig. 1 is a template of judging intersecting blood vessels point;
Fig. 2 a is an original image;
Fig. 2 b is the scale picture;
Fig. 2 c. is a measurement result, is measured zone in the rectangle frame, the vessel segment for measuring such as 1S, 2S, 3S, and colored line segment is a measurement point.
The specific embodiment
For example the present invention is done more detailed description below:
The input blood-vessel image is used for the specified measurement zone with upper and lower, left and right four parameters, and these four values have been indicated the distance (with pixel is unit) of measured zone apart from four borders, original image upper and lower, left and right respectively.When last/following parameter value less than 0 or during greater than picture altitude, its value of program acquiescence is 0 or picture altitude; When a left side/right parameter value less than 0 or during greater than picture traverse, its value of program acquiescence is 0 or picture traverse.
The concrete performing step of blood vessels caliber algorithm proposed by the invention is following:
To the similar Gaussian curve of vessel cross-sections intensity profile, therefore adopt the Gauss matched filtering method that original image is carried out matched filtering, extracting vasculature part, and suppress non-vasculature part, the axis that has also strengthened blood vessel simultaneously.
Make the element of two-dimensional filtering nuclear
Postrotational filtering core element does
θ
i(0≤θ
i≤π) i coupling nuclear of expression towards (because blood vessel move towards difference, generally need 12 Gauss's template of different directions (Δ θ=15 °) original image is carried out convolution and could ideally extract blood vessel during as output valve) with peak value (peak response value).
If pixel (x, the coordinate figure that y) after rotation transformation, obtains for (u, v).With coordinate figure (u, v) in the filtering core calculating formula below the substitution,
k
i(x，y)＝exp(-u
2/2σ
2)
i＝0，1，...，11
Wherein, N={ (u, v) || | u|≤3 σ, | v|≤L/2} is the set of the pixel in the vessel segment.Get vessel segment length L=7, vessel radius is 3 σ, σ=1.So just, can set up 12 filtering cores.Because the data that calculate are the numerical value less than 1, make the practical application filtering core be k ' (x, y)=(k (and x, y)-m
i), m wherein
iAverage for this template.
Step 2, the normalization of image.
Original image carries out normalization after Gauss matched filtering strengthens; Purpose be when eliminate gathering because of the excessive influence that brings to subsequent treatment of gradation of image difference, the gray scale spacing of image is drawn back or made intensity profile even, thereby increase contrast; Make image detail clear, strengthen image.Normalized formula is the maximum gradation value of y=I/ image, and wherein y is a normalization image afterwards, and I is an original image.
Step 3, the binaryzation of image.
At first adopt maximum variance between clusters to ask for binary-state threshold; Then with calculating the shared ratio of given object (blood vessel) in the resulting bianry image of this threshold value, if this ratio less than empirical value 0.35, then increases threshold value; Again produce bianry image, till the bianry image that generation meets the demands.
Step 4, refinement.
Utilize morphological method to carry out the skeletonizing computing.
Step 5 is confirmed the cross point of blood vessel with the method for pattern match.
Write down the coordinate of each pixel on the skeleton line, obtain the spatial positional information of vascular skeleton.Confirm the cross point of blood vessel with the method for pattern match.According to intersecting blood vessels towards difference, preset the direction template of 16 kinds of different intersecting blood vessels points, as shown in Figure 1.
Step 6 is followed the tracks of the pixel on the skeleton line with 8-neighborhood scanning algorithm.
Each intersecting blood vessels point is circulated and makes marks, be used for judging whether this point is treated, and judge the type of this point.Each intersecting blood vessels point is judged in 8 pixels of its neighborhood that if having only a puncta vasculosa that does not make marks, then this point is counted as more following on this vessel segment, does the marked continued and handles; If two puncta vasculosas that do not make marks of surpassing are arranged, then current point is designated as branch.After the puncta vasculosa that does not make marks done marked, as the starting point of fresh blood pipeline section.More following to current point on the detected vessel segment then judged 8 pixels of its neighborhood again, circulation according to this, and till another branch point that detects blood vessel, the processing one section blood vessel that is over so just.
Step 7 is measured the physical distance between the pixel.
Import a width of cloth scale picture.The characteristic that scale on the scale has the gray scale sudden change is the edge, and the present invention has adopted the sobel operator that image is carried out rim detection, can effectively locate the edge and suppress noise.At first need calibrate scale high scale zone relatively clearly, so that obtain result more accurately.Image is carried out floor projection and upright projection, get the half the point of gray value greater than maximum gradation value, marking also with rectangle, intercepting goes out this zone.Then to this zone binaryzation; Calculate the number that goes up each row pixel in the rectangle, the number of obtaining pulse in this rectangle again is graduated number, and each goes total number of pixel divided by graduated number and average with the rectangular area at last; It is the mean number of pixel between adjacent scale; The reuse minimum scale is divided by the mean number of pixel between adjacent scale, and the gained result is the physical distance between the pixel, promptly puts distance.
Step 8, Calibration.
At first, be placed on the point on the detected vessel segment in the one-dimension array, fit to straight line according to these points then, and obtain this collinear slope.Next, be the skeleton line that step-length follow the tracks of to detect this blood vessel with 6 pixels, whenever at a distance from 6 pixel detection to point be current process points; Edge and this skeleton line match perpendicular direction of straight slope of coming out; Detect to both sides, up to detecting background, promptly till the vessel boundary.Then, adding up the number of pixels of both direction mutually is exactly the number of pixels that blood vessels caliber occupies.At last, apart from obtaining the corresponding length of these number of pixels, be the blood vessels caliber of this section blood vessel according to institute's invocation point in the step 7.Repeat above process, all measure until all vessel segments and finish.
Fig. 2 is the blood vessels caliber result who adopts said method to measure, and the concrete width of each measurement point is (0.1mm of unit):
Section 1:214.29,214.29,214.29,214.29,200.00,207.14,214.29,214.29,207.14,214.29
Section 2:178.57,142.86,142.86,142.86,142.86,157.14,150.00,150.00,150.00,142.86,142.86,142.86,142.86,150.00,157.14,150.00,150.00,142.86,150.00,150.00,150.00,207.14
Section 3:235.71,207.14,214.29,228.57,228.57,228.57.
Claims (1)
1. blood vessels caliber measuring method based on digital image processing techniques, its concrete steps are:
The first step, figure image intensifying: get vessel segment length L=7, vessel radius is 3 σ, σ=1, establish pixel (x, the coordinate figure that y) after rotation transformation, obtains for (u, v), with coordinate figure (u, v) substitution filtering core calculating formula k
i(x, y)=exp (u
2/ 2 σ
2), wherein, i=0,1 ..., 11, and (u v) satisfies u≤3 σ, | v|≤L/2;
Make the practical application filtering core be k' (x, y)=(k (and x, y)-mi), m wherein
iBe k
i(x, y) average; To original image carry out convolution and be enhanced as output valve with peak value after image I;
In second step, normalization: making normalization image afterwards is the maximum gradation value of y=I/ original image, and wherein I is the image after strengthening;
The 3rd step; Binaryzation: at first adopt maximum variance between clusters to ask for binary-state threshold and generate bianry image; Investigate vasculature part that threshold value tries to achieve shared ratio in bianry image then, if ratio less than empirical value 0.35, then increases threshold value; Again produce bianry image, till the bianry image that generation meets the demands;
The 4th step, refinement: utilize morphological method to carry out the skeletonizing computing, obtain the vascular skeleton image;
The 5th goes on foot, and confirms the cross point of blood vessel with the method for pattern match: at first, write down the coordinate of each pixel on the skeleton line, obtain the spatial positional information of vascular skeleton; Then, according to intersecting blood vessels towards difference, confirm the cross point of blood vessel with the method for pattern match;
In the 6th step, with 8-neighborhood scanning algorithm the pixel on the skeleton line is followed the tracks of: each intersecting blood vessels point is circulated and makes marks, be used for judging whether this intersecting blood vessels point is treated, and judge the type of this intersecting blood vessels point; Each intersecting blood vessels point is judged in 8 pixels of its neighborhood that if having only a puncta vasculosa that does not make marks, then this puncta vasculosa is counted as more following on the vessel segment, does the marked continued and handles; If two puncta vasculosas that do not make marks of surpassing are arranged; Then current intersecting blood vessels point is designated as branch, the puncta vasculosa that does not make marks is done marked after, as the starting point of fresh blood pipeline section; More following on the detected fresh blood pipeline section then as current point; Judge 8 pixels of its neighborhood again, till another branch point that detects blood vessel, so just handle the one section blood vessel that is over;
In the 7th step, measure the physical distance between the pixel: import a width of cloth scale picture, adopt the sobel operator that scale map is looked like to carry out rim detection; Then, scale map is looked like to carry out floor projection and upright projection, get the half the point of gray value, mark also intercepting with rectangle and go out the half the zone of gray value greater than maximum gradation value greater than maximum gradation value; Then, to cutting the rectangular area binaryzation of going out, calculate the number of each row pixel in the rectangle, the number of obtaining pulse in this rectangle again is graduated number; At last; With total number of each row pixel of rectangular area divided by graduated number and average, the mean number of pixel between promptly adjacent scale, the reuse minimum scale is divided by the mean number of pixel between adjacent scale; The gained result is the physical distance between the pixel, promptly puts distance;
The 8th step, Calibration: be placed on the point on the detected vessel segment in the one-dimension array,, and obtain this collinear slope according to the some match straight line in the one-dimension array; Then; With 6 pixels is the skeleton line that step-length follow the tracks of to detect this blood vessel: will be whenever at a distance from 6 pixel detection to point be current process points; Edge and this skeleton line match perpendicular direction of straight slope of coming out; Detect to both sides, up to detecting background, adding up the number of pixels of the both direction of current process points mutually is exactly the number of pixels that blood vessels caliber occupies; At last, apart from obtaining the corresponding length of number of pixels that blood vessels caliber occupies, be the blood vessels caliber of this section blood vessel according to institute's invocation point in the 7th step; Repeated for the 8th step, all measure until all vessel segments and finish.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105472250A CN101984916B (en) | 2010-11-17 | 2010-11-17 | Blood vessel diameter measuring method based on digital image processing technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105472250A CN101984916B (en) | 2010-11-17 | 2010-11-17 | Blood vessel diameter measuring method based on digital image processing technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101984916A CN101984916A (en) | 2011-03-16 |
| CN101984916B true CN101984916B (en) | 2012-10-31 |
Family
ID=43709354
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2010105472250A Expired - Fee Related CN101984916B (en) | 2010-11-17 | 2010-11-17 | Blood vessel diameter measuring method based on digital image processing technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101984916B (en) |
Families Citing this family (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102764124B (en) * | 2012-07-09 | 2013-12-25 | 华东师范大学 | Magnetic resonance imaging-based perforator flap blood vessel positioning and measurement method |
| CN102908120B (en) * | 2012-10-09 | 2014-09-17 | 北京大恒图像视觉有限公司 | Eye fundus image registration method, eye fundus image optic disk nerve and vessel measuring method and eye fundus image matching method |
| CN105701833B (en) * | 2016-02-22 | 2018-11-20 | 西南交通大学 | Alimentary canal capsule endoscope video hookworm image computer automatic testing method |
| CN105962881A (en) * | 2016-07-26 | 2016-09-28 | 西安交通大学第附属医院 | Blood vessel recognition method and device |
| CN107194928B (en) * | 2017-06-15 | 2019-12-24 | 华中科技大学同济医学院附属协和医院 | Vision-based automatic extraction method for vein blood sampling needle pricking points |
| KR102469720B1 (en) * | 2017-10-31 | 2022-11-23 | 삼성전자주식회사 | Electronic device and method for determining hyperemia grade of eye using the same |
| CN108198211A (en) * | 2017-11-20 | 2018-06-22 | 海纳医信（北京）软件科技有限责任公司 | The processing method and processing device of eye fundus image, storage medium, processor |
| CN110279393B (en) * | 2018-03-19 | 2023-04-28 | 洋华光电股份有限公司 | Microvascular detection device and method |
| CN109886973B (en) * | 2019-01-25 | 2021-01-08 | 杭州晟视科技有限公司 | Blood vessel extraction method and device and computer readable storage medium |
| CN109886938B (en) * | 2019-01-29 | 2023-07-18 | 深圳市科曼医疗设备有限公司 | Automatic measuring method for blood vessel diameter of ultrasonic image |
| CN109829942B (en) * | 2019-02-21 | 2023-04-28 | 韶关学院 | Automatic quantification method for retinal vessel diameter of fundus image |
| CN112617789A (en) * | 2020-07-28 | 2021-04-09 | 上海大学 | Laser speckle blood flow imaging method and system |
| CN112288794B (en) * | 2020-09-04 | 2021-09-07 | 深圳硅基智能科技有限公司 | Method and device for measuring blood vessel diameter of fundus image |
| CN114323303B (en) * | 2021-12-31 | 2023-08-29 | 深圳技术大学 | Body temperature measuring method, device, infrared thermometer and storage medium |
| CN116542966B (en) * | 2023-06-28 | 2023-09-08 | 贵州医科大学附属医院 | Intelligent bone age analysis method for children endocrine abnormality detection |
| CN117576096A (en) * | 2024-01-16 | 2024-02-20 | 成都泰盟软件有限公司 | Method and device for automatically measuring vessel diameter based on image recognition |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101140659A (en) * | 2007-09-29 | 2008-03-12 | 华中科技大学 | Method for dividing vas data in digital vas angiography image |
-
2010
- 2010-11-17 CN CN2010105472250A patent/CN101984916B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101140659A (en) * | 2007-09-29 | 2008-03-12 | 华中科技大学 | Method for dividing vas data in digital vas angiography image |
Non-Patent Citations (2)
| Title |
|---|
| 徐智.心血管造影图像的二维信息处理及其三维重建研究.《天津大学博士学位论文》.2003,正文第21-49、55、77-86、93页. * |
| 王淑珍等.一种血管管径的精确测量方法.《中国医学物理学杂志》.1999,第16卷(第3期),第146-147、160页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101984916A (en) | 2011-03-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101984916B (en) | Blood vessel diameter measuring method based on digital image processing technology | |
| CN106846344B (en) | A kind of image segmentation optimal identification method based on the complete degree in edge | |
| CN107248161A (en) | Retinal vessel extracting method is supervised in a kind of having for multiple features fusion | |
| CN103914827B (en) | The visible detection method of weather strip for automobile profile defects | |
| CN102999886B (en) | Image Edge Detector and scale grating grid precision detection system | |
| CN104794721B (en) | A kind of quick optic disk localization method based on multiple dimensioned spot detection | |
| CN101706843A (en) | Interactive film Interpretation method of mammary gland CR image | |
| US20180025495A1 (en) | Systems and methods of choroidal neovascularization detection using optical coherence tomography angiography | |
| CN103994786B (en) | Image detecting method for arc ruler lines of pointer instrument scale | |
| CN104363815A (en) | Image processing device, image processing method, and image processing program | |
| Ofir et al. | On detection of faint edges in noisy images | |
| CN114757950B (en) | Ultrasonic image processing method, device and computer readable storage medium | |
| CN103226829A (en) | Image edge detection method based on edge enhancement operator | |
| CN108550145A (en) | A kind of SAR image method for evaluating quality and device | |
| CN103177421B (en) | Noise reduction processing method of ultrasound medical image | |
| CN107909588A (en) | Partition system under MRI cortex based on three-dimensional full convolutional neural networks | |
| CN105556567B (en) | Method and system for vertebral location detection | |
| US20160210740A1 (en) | Method and system for spine position detection | |
| CN104715459B (en) | Blood-vessel image Enhancement Method | |
| CN106600615B (en) | A kind of Edge-Detection Algorithm evaluation system and method | |
| AU2014259527A1 (en) | Robust segmentation of retinal pigment epithelium layer | |
| CN102509273B (en) | Tumor segmentation method based on homogeneous pieces and fuzzy measure of breast ultrasound image | |
| CN103914845A (en) | Method for acquiring initial contour in ultrasonic image segmentation based on active contour model | |
| CN105930811A (en) | Palm texture feature detection method based on image processing | |
| CN106372593B (en) | Optic disk area positioning method based on vascular convergence |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20121031 Termination date: 20181117 |