CN104794715A - Auxiliary system for information extraction of ophthalmic slit lamp images and diagnosis of cataract - Google Patents
Auxiliary system for information extraction of ophthalmic slit lamp images and diagnosis of cataract Download PDFInfo
- Publication number
- CN104794715A CN104794715A CN201510195780.4A CN201510195780A CN104794715A CN 104794715 A CN104794715 A CN 104794715A CN 201510195780 A CN201510195780 A CN 201510195780A CN 104794715 A CN104794715 A CN 104794715A
- Authority
- CN
- China
- Prior art keywords
- circle
- center
- cataract
- information
- processing module
- 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
Abstract
The invention discloses an auxiliary system for information extraction of ophthalmic slit lamp images and diagnosis of cataract. The system comprises a database, a patient information and image displaying module and an image information processing module; the database is connected with the patient information and image displaying module; the patient information and image displaying module is connected with the image information processing module; the image information processing module is connected with the database; the database stores patient information and raw and processed images; the database inputs the stored information into the patient information and image displaying module; the patient information and image displaying module transmits the information to the image information processing module; the image information processing module processes the information; the degree of cataract is calculated with a statistical model; the image information processing module transmits the processed information to the database. The system has the advantages that feature information is automatically extracted from the ophthalmic slit lamp images and the information is further made use to assist in diagnosing the cataract.
Description
Technical field
The invention belongs to technical field of medical equipment, relate to Medical Imaging, image processing algorithm etc., specifically a kind of diagnosis ancillary statistics of cataract eye illness and magic magiscan.
Background technology
Medical Imaging is a new branch of science in area of medical diagnostics, at present its in clinical application widely, diagnosis for disease provides larger science and foundation intuitively, clinical symptom, chemical examination etc. can be coordinated better, for the final state of an illness of diagnosing exactly plays irreplaceable effect.Also be applied to treatment aspect preferably, such as: the medical images such as CT, X-ray, B ultrasonic, eye slit-lamp figure, because medical image can reflect the state of an illness of patient intuitively, thus substantially increase the accuracy of diagnosis simultaneously.Further, diagnosis can be made more scientific from the information extraction image and quantitative test, combine with the clinical experience of doctor, doctor can be made the state of an illness of patient and diagnose more accurately.
When doctor diagnoses cataract eye illness by eye slit-lamp figure, usually to depend on the degree deciding cataract eye illness in figure along the brightness in some unique points of axis.This can face two problems:
One, from eye slit-lamp figure, this axis is found out exactly and these unique points are most important.And on eye slit-lamp figure, have the region of a likeness in form " keyhole ", bring very large difficulty to utilizing image processing process axis, location and unique point.
Two, the human interpretation of doctor to eye slit-lamp figure is very time-consuming, also easily occurs that same doctor is in the situation inconsistent to cataract eye illness degree diagnostic result of different time, or the situation that different doctor is inconsistent to cataract eye illness degree diagnostic result.
Summary of the invention
For solving the problems referred to above that prior art exists, the invention discloses the information extraction of a kind of slit-lamp figure and the diagnosis aid system of cataract eye illness, this system relate to the image treating of robust, from the automatic characteristic information extraction of eye slit-lamp figure, and these information are utilized to break the diagnosis of auxiliary cataract eye illness further.The axis of eye slit-lamp figure is located with model's Caro RANSAC algorithm (RANSAC), extract the brightness of Important Characteristic Points in figure, and filter out most important several Variable Factors further by the method for variables choice in statistics, simulate cataract eye illness Degree Model.To the image of all tests, this algorithm correctly can find middle bobbin or tell the figure of imaging errors.Compared with the diagnostic result of existing doctor expert, the accuracy rate of the present invention to the judgement of cataract eye illness degree is higher, and it can reach 95.8%.
The present invention takes following technical scheme: the information extraction of eye slit-lamp figure and the diagnosis aid system of cataract eye illness, comprise database, patient information and image display, image information processing module, database and patient information and image display are connected, patient information and image display and image information processing module are connected, and image information processing module and database are connected; Databases storage patient information, original and process after image, database is by deposited information input patient information and image display, patient information and image display again by information transmission to image information processing module, image information processing module processes information, and calculating cataract eye illness degree value with statistical model, the information after process is sent to database by image information processing module.
Preferably, image information processing module is handled as follows successively to information: one, read in a slit-lamp figure; Two, image boundary is extracted; Three, the axis of eye slit-lamp figure is located by Monte Carlo and random sampling unification algorism; Four, cataract eye illness degree is simulated with statistical model is auxiliary.
Preferably, image information processing module is extracted image boundary and is adopted Canny boundary detection, finds out the border of front cortex and arcus senilis.
Preferably, image information processing module adopts the axis of Monte Carlo and random sampling unification algorism location eye slit-lamp figure:
A. the center of circle is determined with Monte-Carlo step
A) utilize stratified sampling on boundary line, select five points randomly, it is right that these five points can form 20 points, and find out 5 strings the longest;
B) vertical line of each string in a) step is drawn;
C) 10, the intersection point of previous step 5 vertical lines is obtained;
That d) asks previous step 10 intersection points on average obtains the center of circle;
B. with axis, random sampling unification algorism location
A) center of circle of arcus senilis is determined
1) adopt the Monte-Carlo step described in steps A to determine center of circle method, repeat 1000 times, obtain a center of circle at every turn, on average these 1000 centers of circle obtain a center of circle, are designated as C1;
2) again adopt the Monte-Carlo step described in steps A to determine center of circle method, repeat 1000 times, and retain from the center of circle close to C1, the center of circle be on average retained obtains another center of circle, is designated as C2;
B) center of circle of front cortex is determined
The same with the center of circle of location arcus senilis, also look for the center of circle of front cortex in two steps, unlike the 2nd) in the circulation each time of step 1000 times, obtain average after the center of circle, this center of circle and some C2 are linked up and obtains a line, calculate this line and horizontal angle;
C) by the angle of next " ballot " axis of random sampling unification algorism, it is located
Above-mentionedly obtain 1000 lines through C2 point, to the angle of cut of each line computation itself and parallel lines, calculate the histogram of these 1000 angles of cut, select that line of the angle of cut that frequency is the highest as final axis.
Preferably, image information processing module statistical model is auxiliary simulates cataract eye illness degree: adopt the aging eye illness score-system of AREDS, wherein 6 grades altogether, and value 1 shows do not have cataract, and value 6 shows the most serious cataract, after finding axis, 7 unique points along this line drawing also obtain their brightness value, arcus senilis and corneal bow respectively, front cortex and anterior cortex, front lenticle and anteriorlentil, cheuch and sulcus, rear lenticle and posterior lentil, rear thin layer and posterior lamella, rear cortex and post cortex, add several composite factor in a model, altogether define 10 variablees, open one's eyes slit-lamp figure as " training " data set again with 63, most important several Variable Factors is screened by the method for variables choice in statistics, and simulate cataract eye illness Degree Model: the brightness ratio-0.4654 between the brightness+1.40517*anterior lentil of cataract eye illness degree value=0.03077*sulcus and posterior lentil.
When doctor diagnoses cataract by eye slit-lamp figure, usually to depend in figure and decide the cataractous order of severity along the characteristic point luminance of axis.Therefore, from eye slit-lamp figure, this axis is found out exactly most important with these unique points.The present invention utilizes the method for random sampling and statistics " ballot " to remove the impact of noise on figure, locates this axis better.
In addition, the human interpretation of eye slit-lamp figure is compared and wastes time and energy, also easily occur that same doctor is in the situation inconsistent to cataract eye illness degree diagnostic result of different time, or the situation that different doctor is inconsistent to cataract eye illness degree diagnostic result.Therefore, after accurately locating axis and be extracted in these unique points brightness, the present invention provides a quantitative equation further by statistics, effectively improves the accurate rate that doctor diagnoses cataract eye illness, and speed sooner and more objective.
Accompanying drawing explanation
Fig. 1 determines the center of circle by Monte Carlo method.
Fig. 2 comes axis, " ballot " location by RANSAC algorithm.
Fig. 3 is the eye slit-lamp figure that accurately located axis.
Fig. 4 is the histogram that the axis angle " ballot " corresponding to Fig. 3 counts.
Fig. 5 is the eye slit-lamp figure having imaging errors and accurately can not locate axis.
Fig. 6 is the histogram that the axis angle " ballot " corresponding to Fig. 5 counts.
Fig. 7 is the feature mark poiX in a slit-lamp figure.
Fig. 8 is present system module frame chart.
Embodiment
Below in conjunction with accompanying drawing, the preferred embodiment of the present invention is elaborated.
See Fig. 8, the information extraction of the present embodiment eye slit-lamp figure and the diagnosis aid system of cataract eye illness comprise database, patient information and image display, Image Information Processing and diagnostic module, database and patient information and image display are connected, patient information and image display and Image Information Processing and diagnostic module are connected, and Image Information Processing and diagnostic module and database are connected, databases storage patient information, original and process after image and the information such as diagnosis index and result, information is inputted patient information and image display by it, patient information and image display again by information transmission to Image Information Processing and diagnostic module, Image Information Processing and diagnostic module pass through Boundary Extraction, determine the center of circle with the sampling of model's Caro and obtain the brightness of unique point on axis with axis, RANSAC algorithm location etc., and calculate cataract eye illness degree value with statistical model, doctor draws final diagnosis conclusion accordingly, the final diagnosis conclusion of the image after process and doctor is sent to database by Image Information Processing and diagnostic module.
Image Information Processing and the concrete treatment scheme of diagnostic module to image information of the information extraction of the present embodiment eye slit-lamp figure and the diagnosis aid system of cataract eye illness are as follows:
1. read in a slit-lamp figure.
2. extract the border of image.
Use Canny boundary detection, find out the border of front cortex and arcus senilis.Matching can be carried out with the camber line on circle substantially in the border of front cortex and arcus senilis.
3. the axis of eye slit-lamp figure is located with model's Caro and RANSAC algorithm (RANSAC).
C. the center of circle is determined with the sampling of model's Caro.
A) utilize stratified sampling on boundary line, select five points randomly, from 20 some centerings that these five points can be formed, find out 5 strings the longest.
B) vertical line of each string is above drawn.
C) intersection point of previous step 5 vertical lines is obtained.If five points of random choosing just drop on a circle, these 5 vertical lines can intersect at a point.But under normal circumstances, 10 intersection points can be obtained.
That d) asks these 10 intersection points on average obtains the center of circle.
See Fig. 1.
D. axis is located with RANSAC algorithm.
A) center of circle of arcus senilis is determined.
The border of arcus senilis is more clear, hardly by the impact of " Key Hole ", so picture noise is smaller.Determine the center of circle of arcus senilis in two steps:
1) by " determining the center of circle with the sampling of the model's Caro " method described in step, repeat 1000 times, a center of circle can be obtained at every turn.These 1000 centers of circle average obtain a center of circle, are designated as C1.
2) because in step 1) in, that carries out equal weight to 1000 centers of circle on average obtains C1, if when having a center of circle deviation ratio is larger due to picture noise, the position of C1 is just accurate not.In step 2, then with describe in step " with model's Caro sampling determine the center of circle " method, repeat 1000 times.But, only retain those specifically from the closer center of circle of C1.Those centers of circle be retained average obtain a center of circle, are designated as C2.
B) center of circle of front cortex is determined.
The same with the center of circle of location arcus senilis, also look for the center of circle of front cortex in two steps.Unlike the 2nd) in the circulation each time of step 1000 times, obtain average after the center of circle, this center of circle is linked up with some C2 and obtains a line.Calculate this line and horizontal angle.
C) by the angle of next " ballot " axis of RANSAC algorithm, it is located.
From two steps above, obtain 1000 lines through a) mid point C2, to the angle of cut of each line computation itself and water line.Calculate the histogram of these 1000 angles of cut, frequency is higher means the angle of cut of random sample more " approval " its correspondence.The statistic sampling " ballot " of Here it is indication.The present invention selects that line of the angle of cut that frequency is the highest as final axis.
See Fig. 2-6.
4. the diagnosis of cataract eye illness is assisted with statistical model.
Current doctor, to the diagnosis of cataract eye illness degree, is by human interpretation's eye slit-lamp figure, utilizes the brightness of some unique points on figure compared with the palpebral fissure gap lamp standard drawing of a set of expert's preliminary election, finally obtains a value to describe cataract eye illness degree.The present invention adopts the aging associated eye conditions score-system of AREDS, wherein 6 grades altogether.Value 1 shows do not have cataract, and is worth 6 and shows the most serious cataract.After have found axis, according to the suggestion of cataract eye illness expert, 7 unique points along this line drawing also obtain their brightness value, be arcus senilis (corneal bow), front cortex (anterior cortex), front lenticle (anterior lentil), cheuch (sulcus), rear lenticle (posterior lentil), rear thin layer (posterior lamella) and rear cortex (post cortex) respectively, these put concrete position as shown in Figure 7.
In doctor's Artificial Diagnosis process, be reaction from anterior lentil to posterior lamella along the brightness of axis change be an important indicator of cataract eye illness degree, therefore, the brightness ratio of several composite factor such as between anterior lentil and posterior lentil has been added in a model.Altogether define 10 variablees, then slit-lamp figure is opened one's eyes as " training " data set with 63, filter out most important several Variable Factors by the method for variables choice in statistics, and simulate cataract eye illness Degree Model: the brightness ratio-0.4654 between the brightness+1.40517*anterior lentil of cataract eye illness degree value=0.03077*Sulcus and posterior lentil.
With the 150 slit-lamp figure that open one's eyes, model is tested subsequently.To the image of all tests, algorithm of the present invention all correctly have found middle bobbin or tells the figure of imaging errors.Compared with the diagnostic result of existing doctor expert, model of the present invention to the rate of accuracy reached of the judgement of cataract eye illness degree to 95.8%.
Basic terms
Slit-lamp: full name " slit-lamp microscope " is the most used a kind of optical device of ophthalmology.Clearly can to observe before eyelid, conjunctiva, sclera, cornea, anterior chamber, iris, pupil, crystalline lens and vitreum 1/3 by slit-lamp microscope, the position of pathology, character, size and the degree of depth thereof can be determined.
Corneal bow: arcus senilis.
Anterior corte: front cortex.
Posterior corte: rear cortex.
Those of ordinary skill in the art will be appreciated that; it is of the present invention that above embodiment is only used to explanation; and not as limitation of the invention, as long as within the scope of the invention, all will protection scope of the present invention be fallen into the change of above embodiment, modification.
Claims (5)
1. the information extraction of a slit-lamp figure and the diagnosis aid system of cataract eye illness, it is characterized in that comprising database, patient information and image display, image information processing module, database and patient information and image display are connected, patient information and image display and image information processing module are connected, and image information processing module and database are connected; Databases storage patient information, original and process after image, database is by deposited information input patient information and image display, patient information and image display again by information transmission to image information processing module, image information processing module processes information, and calculating cataract eye illness degree value with statistical model, the information after process is sent to database by image information processing module.
2. the information extraction of as claimed in claim 1 slit-lamp figure and the diagnosis aid system of cataract eye illness, is characterized in that: described image information processing module is handled as follows successively to information: one, read in a slit-lamp figure; Two, image boundary is extracted; Three, the axis of eye slit-lamp figure is located by Monte Carlo and random sampling unification algorism; Four, cataract eye illness degree is simulated with statistical model is auxiliary.
3. the information extraction of as claimed in claim 2 slit-lamp figure and the diagnosis aid system of cataract eye illness, is characterized in that: described image information processing module is extracted image boundary and adopted Canny boundary detection, finds out the border of front cortex and arcus senilis.
4. the information extraction of as claimed in claim 3 slit-lamp figure and the diagnosis aid system of cataract eye illness, is characterized in that: described image information processing module adopts the axis of Monte Carlo and random sampling unification algorism location eye slit-lamp figure:
A. the center of circle is determined with Monte-Carlo step
A) utilize stratified sampling on boundary line, select five points randomly, it is right that these five points can form 20 points, and find out 5 strings the longest;
B) vertical line of each string in a) step is drawn;
C) 10, the intersection point of previous step 5 vertical lines is obtained;
That d) asks previous step 10 intersection points on average obtains the center of circle;
B. with axis, random sampling unification algorism location
A) center of circle of arcus senilis is determined
1) adopt the Monte-Carlo step described in steps A to determine center of circle method, repeat 1000 times, obtain a center of circle at every turn, on average these 1000 centers of circle obtain a center of circle, are designated as C1;
2) again adopt the Monte-Carlo step described in steps A to determine center of circle method, repeat 1000 times, and retain from the center of circle close to C1, the center of circle be on average retained obtains another center of circle, is designated as C2;
B) center of circle of front cortex is determined
The same with the center of circle of location arcus senilis, also look for the center of circle of front cortex in two steps, unlike the 2nd) in the circulation each time of step 1000 times, obtain average after the center of circle, this center of circle and some C2 are linked up and obtains a line, calculate this line and horizontal angle;
C) by the angle of next " ballot " axis of random sampling unification algorism, it is located
Above-mentionedly obtain 1000 lines through C2 point, to the angle of cut of each line computation itself and parallel lines, calculate the histogram of these 1000 angles of cut, select that line of the angle of cut that frequency is the highest as final axis.
5. the information extraction of as claimed in claim 4 slit-lamp figure and the diagnosis aid system of cataract eye illness, it is characterized in that: described image information processing module statistical model is auxiliary simulates cataract eye illness degree: adopt the aging eye illness score-system of AREDS, wherein 6 grades altogether, value 1 shows do not have cataract, and value 6 shows the most serious cataract, after finding axis, 7 unique points along this line drawing also obtain their brightness value, arcus senilis and corneal bow respectively, front cortex and anterior cortex, front lenticle and anterior lentil, cheuch and sulcus, rear lenticle and posterior lentil, rear thin layer and posteriorlamella, rear cortex and post cortex, add several composite factor in a model, altogether define 10 variablees, open one's eyes slit-lamp figure as " training " data set again with 63, most important several Variable Factors is screened by the method for variables choice in statistics, and simulate cataract eye illness Degree Model: the brightness ratio-0.4654 between the brightness+1.40517*anterior lentil of cataract eye illness degree value=0.03077*sulcus and posterior lentil.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510195780.4A CN104794715A (en) | 2015-04-22 | 2015-04-22 | Auxiliary system for information extraction of ophthalmic slit lamp images and diagnosis of cataract |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510195780.4A CN104794715A (en) | 2015-04-22 | 2015-04-22 | Auxiliary system for information extraction of ophthalmic slit lamp images and diagnosis of cataract |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104794715A true CN104794715A (en) | 2015-07-22 |
Family
ID=53559495
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510195780.4A Pending CN104794715A (en) | 2015-04-22 | 2015-04-22 | Auxiliary system for information extraction of ophthalmic slit lamp images and diagnosis of cataract |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104794715A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080161781A1 (en) * | 2005-02-19 | 2008-07-03 | Mcardle George J | Apparatus and Processes For Preventing or Delaying Onset or Progression of Age-Related Cataract |
| CN101584574A (en) * | 2008-05-19 | 2009-11-25 | 复旦大学附属眼耳鼻喉科医院 | Lens image analysis method |
| CN102984997A (en) * | 2009-08-24 | 2013-03-20 | 新加坡保健服务集团有限公司 | A Method and system of determining a grade of nuclear cataract |
-
2015
- 2015-04-22 CN CN201510195780.4A patent/CN104794715A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080161781A1 (en) * | 2005-02-19 | 2008-07-03 | Mcardle George J | Apparatus and Processes For Preventing or Delaying Onset or Progression of Age-Related Cataract |
| CN101584574A (en) * | 2008-05-19 | 2009-11-25 | 复旦大学附属眼耳鼻喉科医院 | Lens image analysis method |
| CN102984997A (en) * | 2009-08-24 | 2013-03-20 | 新加坡保健服务集团有限公司 | A Method and system of determining a grade of nuclear cataract |
Non-Patent Citations (2)
| Title |
|---|
| SHAOHUA FAN ET AL: "An Automatic System for Classification of Nuclear Sclerosis from Slit-Lamp Photographs", 《LECTURE NOTES IN COMPUTER SCIENCE(LNCS)》 * |
| 鞠颖 等: "基于裂隙灯显微图像的眼前节特征提取的新方法", 《中国生物医学工程学报》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7370969B2 (en) | Corneal topography analysis system | |
| KR102198395B1 (en) | Method and System for Early Diagnosis of Glaucoma and Displaying suspicious Area | |
| CN107330449A (en) | A kind of BDR sign detection method and device | |
| CN110428421A (en) | Macula lutea image region segmentation method and apparatus | |
| CN109961848B (en) | Macular image classification method and device | |
| CN110188613A (en) | Image classification method and equipment | |
| CN104102899B (en) | Retinal vessel recognition methods and device | |
| US11062444B2 (en) | Artificial intelligence cataract analysis system | |
| CN112101424B (en) | Method, device and equipment for generating retinopathy identification model | |
| CN106214120A (en) | A kind of methods for screening of glaucoma | |
| US20210244272A1 (en) | Anterior eye disease diagnostic system and diagnostic method using same | |
| CN109036556A (en) | A method of keratoconus case is diagnosed based on machine learning | |
| CN113066066A (en) | Retinal abnormality analysis method and device | |
| Patil et al. | An efficient method of detecting exudates in diabetic retinopathy: Using texture edge features | |
| CN106446805A (en) | Segmentation method and system for optic cup in eye ground photo | |
| CN111179258A (en) | Artificial intelligence method and system for identifying retinal hemorrhage image | |
| Miao et al. | Classification of Diabetic Retinopathy Based on Multiscale Hybrid Attention Mechanism and Residual Algorithm | |
| CN110598652B (en) | Fundus data prediction method and device | |
| CN112006651A (en) | Cataract surgery auxiliary diagnosis system and method thereof | |
| CN116452571A (en) | Image recognition method based on deep neural network | |
| CN114445666A (en) | Deep learning-based method and system for classifying left eye, right eye and visual field positions of fundus images | |
| CN104794715A (en) | Auxiliary system for information extraction of ophthalmic slit lamp images and diagnosis of cataract | |
| Purwanithami et al. | Hemorrhage diabetic retinopathy detection based on fundus image using neural network and FCM segmentation | |
| Aurangzeb et al. | Systematic development of AI-Enabled diagnostic systems for glaucoma and diabetic retinopathy | |
| CN113470815A (en) | AI technology-based cataract patient vision impairment degree evaluation system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate 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: 20150722 |