WO2011108995A1 - Analyse automatique d'images de la chambre antérieure d'un œil - Google Patents

Analyse automatique d'images de la chambre antérieure d'un œil Download PDF

Info

Publication number
WO2011108995A1
WO2011108995A1 PCT/SG2011/000088 SG2011000088W WO2011108995A1 WO 2011108995 A1 WO2011108995 A1 WO 2011108995A1 SG 2011000088 W SG2011000088 W SG 2011000088W WO 2011108995 A1 WO2011108995 A1 WO 2011108995A1
Authority
WO
WIPO (PCT)
Prior art keywords
eye
image
angle
arcs
edges
Prior art date
Application number
PCT/SG2011/000088
Other languages
English (en)
Inventor
Jun Cheng
Beng Hai Lee
Jiang Liu
Tin Aung
Baskaran Mani
Tien Yin Wong
Original Assignee
Agency For Science, Technology And Research
Singapore Health Services Pte Ltd
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 Agency For Science, Technology And Research, Singapore Health Services Pte Ltd filed Critical Agency For Science, Technology And Research
Publication of WO2011108995A1 publication Critical patent/WO2011108995A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • 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
    • G06T2207/20048—Transform domain processing
    • G06T2207/20061—Hough transform
    • 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
    • G06T2207/30041—Eye; Retina; Ophthalmic

Definitions

  • the present invention relates to methods and devices for automatically analysing an eye image to characterize the geometry of the region including the junction of the iris and cornea.
  • Glaucoma is an optic nerve disease resulting in progressive, irreversible loss of vision. It is often associated with increased pressure of fluid in the eye. Glaucoma can be roughly divided into open angle glaucoma and closed angle glaucoma, based on the "iridocorneal angle", which means the acute angle between the iris and cornea at the periphery of the anterior chamber of the eye.
  • Fig. 1 is a cross-sectional view of an eye
  • Fig. 2 shows an enlarged view of a portion of Fig. 1 : a side portion of the eye's anterior chamber.
  • the iris located between the cornea and the lens, controls the amount of light entering the eye.
  • the iris, cornea, and lens are bathed in a liquid called aqueous humor, which is continually produced by nearby tissues. It moves out of the eye via a drainage canal called the trabecular meshwork, which, as illustrated in Fig. 2, lies at the intersection of the iris and the cornea. Blocking of the canal leads to the increased fluid pressure.
  • open angle glaucoma When the iridocorneal angle is open, a glaucoma is called open angle glaucoma. Open angle glaucoma accounts for more than 90% of glaucoma patients. It is usually chronic and progresses gradually. Open angle glaucoma is a leading cause of blindness. It is particularly dangerous as it can progress gradually and go unnoticeable for years. Thus, regular eye examination is suggested to detect the disease before it becomes a serious problem. So far, open angle glaucoma cannot be cured, but it can be controlled by reducing the pressure in the eye using eyedrops and/or oral medications. The eyedrops work by enhancing the drainage of aqueous humor or by reducing the production of the fluid in the eye. Normally, patents need to have these medications continued for the rest of the patient's life.
  • closed angle glaucoma When the iridocorneal angle is narrow or even closed, it is called closed angle glaucoma. Closed angle glaucoma affects less than 10 percent of patients with glaucoma. It is often inherited and occurs more commonly in farsighted elderly people. A closed-angle glaucoma attack is usually acute, occurring when the drainage area is blocked. Closed angle glaucoma must be treated quickly; otherwise, it may result in blindness within hours to days of the attack.
  • the therapy for each type of glaucoma is specific, it is important to determine the type of the glaucoma. Thus, it is essential to visualize the iridocorneal angle to make a correct diagnosis of the disease.
  • the conventional process for measuring the iridoconeal angle is "gonioscopy", an eye examination looking at the front part of the eye between the cornea and the iris.
  • a "gonionlens” or “gonioscope” such as truncated cone with a reflective inner surface, is placed in contact with the eye (resting on the lower eyelid and/or the sclera).
  • the inner surface of the cone functions as a mirror, to reflect the light from the iridocorneal angle into the direction of the observer, where it can be collected and magnified.
  • Gonioscopy is considered as the most practical method to assess the angle, but it is subjective and difficult to quantify.
  • gonioscopy is not easy to perform as patients often blink their eyes, squeeze their eyelids together, etc. The success of gonioscopy therefore requires considerable photographic skill & effort as well as a full knowledge of the angle structures.
  • gonioscopy The purposed of gonioscopy is not simply to determine whether the iridocorneal angle is open or closed.
  • the main aims of gonioscopy are to evaluate the functional status of the angle, the degree of closure, and the risk of future closure.
  • the following structures are to be identified: Ciliary body band: The most posterior structure appears pink to dull-brown to slate- grey in colour. Its width depends on the position of the iris insertion; tending to be narrower in hypermetropic eyes and wider in myopic eyes.
  • the angle recess represents the dipping of the iris as it inserts into the ciliary body.
  • Scleral spur This is the most anterior projection of the sclera and the site of attachment of the longitudinal muscle of the ciliary body. Gonioscopically, the scleral spur is situated just posterior to the trabeculum and appears as a narrow dense, often shiny, whitish band. It is the most important land-mark because it has a relatively consistent appearance in different eyes.
  • Trabecular meshwork This extends from the scleral spur to Schwalbe's line.
  • the posterior functional pigmented part lies adjacent to the scleral spur and has a greyish-blue translucent appearance.
  • the anterior non-function part lies adjacent to Schwalbe's line and has a whitish colour. Trabecular pigmentation is rare prior to puberty.
  • Schwalbe's line This is the most anterior structure and appears as an opaque line. Anatomically it represents the peripheral termination of Descemet's membrane and the anterior limit of the trabeculum.
  • Gonioscopy is presently essential for glaucoma diagnosis. Without gonipscopy, it is difficult for doctors to get familiar with the variety of normal and abnormal findings that may be present.
  • the Shaffer grading system (D Minckler, P Foster and PT Hung, Angle Closure Glaucoma Classification and Racial Variation, Asian Journal of Ophthalmology, vol. 3, no. 3, 4, 2001 ) is a widely used system comparing iridocorneal angles. It records the angle in degrees of arc subtended by the inner surface of the trabecular meshwork and the anterior surface of the iris, about one-third of the distance from its periphery. An estimation of the angle width is achieved by observing the amount of separation between the inner surface of the trabeculum and the anterior iris surface. In practice, the examiner grades the angle according to the visibility of the various angle structures.
  • the Shaffer grading system provides a method of comparing the widths of different angles. The system assigns a numerical grade to each angle with associated anatomical description, the angle width in degrees and implied clinical interpretation. Table 1 summarize the Shaffer grading system.
  • RetCam is the trademark of an imaging system available from Clarity Medical Systems, Inc., Pleasanton, CA.
  • Retcam is a fundus camera which has traditionally been used to capture retinal images. With some technical modifications, the camera can be used to capture an image of the iris, and cornea (I. I. K. Ahmed AND L. D. Mackeen, "A new approach for imaging the angle", Glaucoma Today, pp. 27-30, JULY/AUGUST 2007).
  • the RetCam and similar devices have wide field of view, and a fibre optic light system which focuses light to an angle which minimally affects the pupil (in contrast to gonioscopy).
  • Other fundus camaras can be modified to do similar work, but the RetCam device appears to have better resolution under similar lighting conditions.
  • RetCam and similar devices may be referred to by the term "goniophotography” or "angle imaging".
  • the camera captures a view of the eye in a direction which faces the planes of the iris and cornea, and not from a viewpoint which looks between the iris and the cornea.
  • goniophotography systems there is direct viewing of the eye (i.e. without a solid intermediate tool; for example, the lens of Retcam touches the surface of eye with gel between them), whereas gonioscopy uses an indirect view (the camera is away from eye with an enlarge-mirror in between).
  • goniophotography has a wider field of view than gonioscopy.
  • Goniophotography has some advantages compared with gonioscopy.
  • RetCam imaging subjects are in a supine position on a bed or reclining chair.
  • An advantage of this is that the patients cannot pull away from the camera lens' tip, which is a common problem when using goniolenses.
  • the tip of the RetCam's lens is smaller than the standard goniolens and does not sit on the patient's lower eyelid.
  • patients are less likely to squeeze their eyelids shut and to struggle against the stimulus of a foreign object in their eyes.
  • the illumination from the camera sweeps across their pupils, which means that patients are also less likely to experience discomfort from the light's intensity, so angle-related artifacts are minimal.
  • the RetCam makes goniography fast, easy, and effective.
  • the new retcam imaging modality enhances the use of a camera as a clinical, educational, and research tool. It makes goniography easier and thus more images can be captured.
  • Manual grading is currently adopted clinically. As we can see that manual grading of medical images is usually tedious and time consuming. This holds true for retcam images as well. Moreover, the angle usually lies in a small portion in the retcam image which makes the manual grading more difficult. Manual grading is often subjective by the grader and thus the reproducibility is a concern.
  • the present invention aims to provide methods and apparatus to analyse automatically an image of at least part of the anterior chamber of an eye. This image is collected without a gonioscope, by a system such as the RetCam camera.
  • the invention proposes that the image is processed to identify from this eye image the structure of the junction of the corea and iris.
  • the method may identify whether the region of the eye image including the junction of the iris or cornea contains proximate arcuate lines, since it is experimentally observed that this is characteristic of a higher value of the iridocorneal angle. That is, for patients for whom the cornea and iris are not touching, there are two substantially parallel (that is, co-centric) arcuate lines having very similar radii of curvature.
  • the image is formed by a goniophotography (angle imaging). It does not include portions which show the mutually-facing surfaces of the iris and cornea (i.e. the inwardly facing walls of the anterior chamber of the eye).
  • the method includes an edge detection step, a step of arc identification, and then a method of determining whether arcs have been detected which comprise proximate lines.
  • the lines may be required to meet pre-known criteria, which are obtained from knowledge about the structure of the eye. In this way, it is possible to reduce the influence of other spurious edges in the eye image.
  • the invention may make possible a precise, efficient and preferably automatic system to grade the angle, which in turn would make it possible to provide large- scale clinical use. That is, a images from a large number of subjects could be processed at low cost, for example as part of a screening process for glaucoma.
  • automated is meant that the technique is performed by a computer without human interaction, except that a human may initiate the process.
  • the term “semiautomatic” is used to describe a process including automatic steps but with certain steps performed by with the aid of a human operator.
  • Preferred steps of the process involve further processing of the image utilizing medical image processing, pattern recognition, statistical learning, and/or other technologies. These steps are important to improve the accuracy of the result.
  • Fig. 1 is a known cross-sectional image of a human eye
  • Fig. 2 is an expanded portion of the Fig. 1 ;
  • Fig. 3 is a first flow diagram of steps of an embodiment of the invention.
  • Fig. 4 is a more detailed form of the flow diagram of Fig. 3;
  • Fig. 5 is composed of Figs 5(a)-(d), which are four sample Retcam images;
  • Fig. 6 is composed of Fig. 6(a)-(d), and shows images obtained from the images of Fig. 5 by a first step of the embodiment of Fig. 3;
  • Fig. 7 is composed of Figs 7(a) to (d), and shows image obtained from a second step of the embodiment of Fig. 3;
  • Fig. 8 is composed of Figs. 8(a) and 8(b), and illustrates the known Hough transformation
  • Fig. 9 illustrates a step of a fourth step of the method of Fig. 3.
  • the input to the method is a RetCam image.
  • the method contains the following steps: edge detection (step 1 ), arc detection (step 2), closed/open angle glaucoma classification (step 3), and identifying angle structures and grading (step 4).
  • the edge detection step is a pre-processing step for the arc detection step.
  • the arc detection aims to find the boundary between different angle structures.
  • the third step determines whether an angle is open or closed, and thus is an initial grading of the eye.
  • the fourth step is optional, but, when carried out, obtains the degree of the closure and the potential to become closed. The width and the area of the angle are measured.
  • Fig. 4 shows the flow of Fig. 3 in more detail, as will be explained below.
  • Steps 1 to 4 will now be explained in more detail with reference to four RetCam images which are shown (in black-and-white versions which are derived from coloured original RetCam images) in Fig. 5. These are images viewing the anterior chamber of the eye from the top, bottom, left and right. In these images, the inner black area is the pupil. The grey area surrounding the pupil is the iris (this is yellow in the original RetCam images). The outer whitish area is the cornea. The angle which is of interest is located at the area between the iris and the cornea. The portion of the image, which contains the junction of the cornea and iris is viewed in a direction which is substantially perpendicular to those structures, and the mutually- facing surfaces of the iris and cornea are not directly viewed.
  • the edge detection step is to find the candidate region of interest. As shown in Fig. 4, it has two sub-steps.
  • a first sub-step shown as “segment effective image area” in Fig. 4, is to identify background areas of the image. This is done by “pixel thresholding (which means that a pixel is regarded as background (and excluded from further analysis) unless its intensity is above a predefined value, in which case it is regarded as "foreground”) followed by a morphological erosion processing.
  • a second sub-step (shown as "contrast image”) in Fig. 4 is to form a binary image which shows edges in the foreground. This is done by the well-known technique of "Canny edge detection" in the embodiment. Since the background has been excluded, this two sub-step process excludes some false edges, i.e., the edges in the background area.
  • Fig. 6(a)-(d) are the respective outputs of the edge detection step when the four sample images of Fig. 5(a)-(d) are used as the input. Figs. 6(a)-(d) still contain noise, false edges, broken edges, and etc. Thus, further processing is performed.
  • step 2 applies arc detection algorithms to detect true arcs and remove false arcs.
  • the task here is to detect a known shape: an arc.
  • a Hough transform is suitable for solving such a problem. Since the edges are approximately part of a circle, a circular Hough transform is applied. In contrast to a Hough transform for straight line detection, a circular Hough transform has three unknowns instead of two: the locations of centre in the 2D coordinates and the radius of the circle.
  • the output of the Hough transform is an accumulator, where the local maxima correspond to the arcs.
  • the algorithm for Circular Hough Transformation can be summarized into following three steps. It employs an "accumulator" which is a value for each point on the 2-D plane.
  • the input to the method is the set of edge points found in step 1.
  • Fig. 8 The edge points are shown by the solid line in Fig. 8(a). These points all lie on a single arc (in fact, a circle). For three of the edge points, the corresponding circles obtained in step (1 ) of the Hough transform are shown by the dashed lines in Fig. 8(b). As seen in Fig. 8(b), the point which lies on the intersection of the three dashed circles has a higher value for the accumulator than any other point in the plane. The existence of this maximum confirms that the edge points lie on an arc with radius r. If there were multiple arcs, this would lead to multiple points having a local maximum of the accumulator value.
  • the embodiment may include restrictions on the three unknowns to reduce the computation load.
  • the parameter r is selected based on prior knowledge. For example, r can be optimised (in the sense of maximising the value of the highest accumulators) within a pre-known range. The experimental results given below optimised r within the range 450-650 pixels (if another resolution is used, of course, a different value of r would be appropriate). This value was selected based on test images which Were analysed manually. A wider range may optionally be used, but this leads to a higher computational cost.
  • Fig. 7(a)-(d) shows the show results of Hough transform for each of the respective images 6(a)-(d).
  • Each ' * ' denotes a local maxima, which corresponds to an arc.
  • Figs. 7(a) and 7(b) have two such maxima.
  • Figs. 7(c) and 7(d) have one such maximum.
  • the local maximum for a given arc is at a position which is called the arc centre.
  • step 4 it will be useful (in step 4) to have "quadrant information", which indicates whether the position of the centre of the pupil relative to the centre of the image, and specifically to determine if the direction from the pupil to the centre of image is upward (as perceived by the subject), downwards, in the lateral direction towards the nose, or in the lateral direction away from the nose.
  • quadrant information indicates whether the position of the centre of the pupil relative to the centre of the image, and specifically to determine if the direction from the pupil to the centre of image is upward (as perceived by the subject), downwards, in the lateral direction towards the nose, or in the lateral direction away from the nose.
  • Fig. 6(d) is an example with a false arc (the arc which is to the right of Fig. 6(d)) which is due to other structures instead of the angle.
  • step 6(d) means the right-side arc. Note that although we use the strongest arc to determine the quadrant, that strongest arc may later be excluded. This is because some eye images contain one or more arcuate edges which are centred on the pupil but which are radially outward from the ⁇ dges associated with the iridocorneal angle, and sometimes such radially-outward edges are stronger edges than those associated with the iridocorneal angle. In summary, we use the strongest arc to determine the quadrant, and then, if the distance between the arcs is greater than the predetermined value, we exclude all but the one closer to the pupil. Another sub-step of step 3, which gives a result which is useful in step 4 is segmenting a region of interest.
  • the strongest arc (corresponding to the maximum among the all local maximums) is adopted. This process is shown in Fig. 9, where the strongest arc is marked as line 11. Denoting its arc radius as r, and its arc center as (x c ,y c ) , we draw another two arcs 12, 13 with same arc centre
  • a basic task of grading is to tell whether an angle is open or closed. When an angle is closed or nearly closed, the two sides of the angle (i.e. the mutually-facing surfaces of the iris and cornea) overlap. In terms of the arc, we would be able to detect one arc only.
  • the classification step 3 we classify a patient known to be suffering from glaucoma as having open angle glaucoma if we can detect two or more arcs. Otherwise, the patient is classified as having close angle glaucoma.
  • An automatic algorithm is used to determine if there is one maximum, or two (or more) . We find the local maximum in a predefined region by scanning over the images.
  • Step 4 is a more advanced analysis which includes identifying angle structures and grading the angle.
  • the grading of angle width is an important part of the ocular examination. It is desirable to determine:
  • the amount of trabecular pigmentation is the amount of trabecular pigmentation.
  • the Shaffer grading system may be applied to determine the score of the grading results, which would be verified with the clinical grading results.
  • the arc detected earlier would be mapped to possible angle structures. The area and the width of the angle would be examined and measured for the automatic grading.
  • step 4 is performed using the segmented "region of interest" obtained in step 4, to identify angle structures.
  • a first sub-step shown in Fig. 4 we extract features from the region of interest. Suppose (just as an example) that the features we use are the edges obtained in step 1. If so, step 4 uses those of the edges which are within the region of interest. For image classified in step 2 as superior or inferior, divide the region of interest into parallel columns. In each column, we find the top edge and the bottom edges, and we assume that these edges are part of the structure of the iridocorneal angle structure. In a next sub-step shown in Fig.
  • angle width for this column as the distance between the top edge in the column and the bottom edge in the column (for columns having just one edge, this distance is zero. If a different structure than edge had been used, then this sub-step would obtain the average value of the dimensions of that structure along the region of interest.
  • Step 3 we compute the average of the angle width over the columns of the region of interest, to give the mean angle width. If the mean angle width is less than a predetermined threshold, the angle is determined to be "closed angle glaucoma". Otherwise, it is determined that it is open angle glaucoma. Note that this process provides an alternative to the process of step 3 above, and we have found that it is more accurate. (Step 3 may therefore be omitted.)
  • step 4 includes advanced grading by comparing the mean angle width with predetermined ranges, each corresponding to a respective grade (e.g. Shaffer grade). In other words, if the mean angle width is found to be in a certain one of the predefined ranges, the image is graded as a having the corresponding grade.
  • predetermined ranges e.g. Shaffer grade
  • the computation is similar to that explained in the preceding three paragraphs, except that we process the region of interest row-by- row, rather than column by column, and in each row we find the leftmost and rightmost edges, and work out the angle width based on their mutual distance. The average angle width is then found as an average angle width over the rows.
  • An alternative approach is to use the angle width in each column or row to form a profile of angle width. Then the angle width profile is used as a feature and combined with machine learning to train classifiers to predict grades.
  • Steps 1-3 of the embodiment of Fig. 3 were performed on RetCam image sets of 99 patients from the Singapore Eye Research Institute. Of these, 54 are patients with open angled glaucoma, and 45 are patients with closed angle glaucoma. One image from each patient was randomly selected for testing. The embodiment classified 45 of the 54 open angled glaucoma cases correctly as open angle, and 39 of the 45 closed angle glaucoma correctly as closed angle. Thus, the accuracies of the embodiment in correctly identifying open/closed angle glaucoma were 83.3% and 86.7% respectively.
  • step 4 the embodiment classified the 50 of 54 open angle glaucoma cases correctly as open angle and 44 of 45 closed angle glaucoma cases correctly as closed angle.
  • the accuracies are 92.6% and 97.8%.
  • the embodiment provided promising and encouraging open/closed angle classification.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Geometry (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Eye Examination Apparatus (AREA)

Abstract

La présente invention a trait à une image de l'œil incluant une image de la zone contenant l'intersection de la cornée et de l'iris qui est traitée de manière à identifier si elle contient deux lignes rapprochées, dans la mesure où il est observé de façon expérimentale qu'il s'agit d'une caractéristique d'une valeur non nulle de l'angle iridocornéen. Le procédé inclut une étape de détection de bord, puis un procédé permettant de déterminer si les bords ont été détectés qui comprend ces deux lignes rapprochées. Il peut être nécessaire que les lignes répondent à des critères connus au préalable, qui sont obtenus grâce à la connaissance de la structure de l'œil. De cette manière, il est possible de réduire l'influence des autres bords parasites dans l'image de l'œil. L'invention permet d'obtenir un système précis, efficace et automatique permettant de régler l'angle, qui à son tour permet de fournir une utilisation clinique à grande échelle.
PCT/SG2011/000088 2010-03-04 2011-03-04 Analyse automatique d'images de la chambre antérieure d'un œil WO2011108995A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US31063310P 2010-03-04 2010-03-04
US61/310,633 2010-03-04

Publications (1)

Publication Number Publication Date
WO2011108995A1 true WO2011108995A1 (fr) 2011-09-09

Family

ID=44542451

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/SG2011/000088 WO2011108995A1 (fr) 2010-03-04 2011-03-04 Analyse automatique d'images de la chambre antérieure d'un œil

Country Status (1)

Country Link
WO (1) WO2011108995A1 (fr)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110310254A (zh) * 2019-05-17 2019-10-08 广东技术师范大学 一种基于深度学习的房角图像自动分级方法
IT201800004938A1 (it) * 2018-04-27 2019-10-27 Metodo per determinare una curva di trabecolato basato sulla applicazione di soglie ai canali di una immagine a colori
IT201800004934A1 (it) * 2018-04-27 2019-10-27 Metodo per determinare una curva di trabecolato basato su minimi locali negativi multipli in una immagine a toni-di-grigio
WO2019207481A1 (fr) * 2018-04-27 2019-10-31 Nidek Technologies Srl Procédés de détermination d'une courbe de réseau trabéculaire

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6942343B2 (en) * 2003-04-07 2005-09-13 Arkadiy Farberov Optical device for intraocular observation
US20090157062A1 (en) * 2007-12-13 2009-06-18 Christoph Hauger Systems and methods for treating glaucoma and systems and methods for imaging a portion of an eye

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6942343B2 (en) * 2003-04-07 2005-09-13 Arkadiy Farberov Optical device for intraocular observation
US20090157062A1 (en) * 2007-12-13 2009-06-18 Christoph Hauger Systems and methods for treating glaucoma and systems and methods for imaging a portion of an eye

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
CORON, A. ET AL.: "Automatic Segmentation of the Anterior Chamber in in vivo High- frequency Ultrasound Images of the Eye", IEEE ULTRASONICS SYMPOSIUM, 2007, pages 1266 - 1269 *
HE, M. ET AL.: "Heritability of the Iridotrabecular Angle Width Measured by Optical Coherence Tomography in Chinese Children: The Guangzhou Twin Eye Study", INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE, vol. 49, no. 4, 2008, pages 1356 - 1361 *
RAMANATHAN, S. ET AL.: "Automatic Detection of Accretion of Glaucoma in Eye", SYSTEMS, SIGNALS AND IMAGE PROCESSINGAND 6TH EURASIP CONFERENCE FOCUSED ON SPEECH AND IMAGE PROCESSING, 14TH INTERNATIONAL WORKSHOP ON MULTIMEDIA COMMUNICATIONS AND SERVICES, 2007, pages 441 - 445 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
IT201800004938A1 (it) * 2018-04-27 2019-10-27 Metodo per determinare una curva di trabecolato basato sulla applicazione di soglie ai canali di una immagine a colori
IT201800004934A1 (it) * 2018-04-27 2019-10-27 Metodo per determinare una curva di trabecolato basato su minimi locali negativi multipli in una immagine a toni-di-grigio
WO2019207481A1 (fr) * 2018-04-27 2019-10-31 Nidek Technologies Srl Procédés de détermination d'une courbe de réseau trabéculaire
CN110310254A (zh) * 2019-05-17 2019-10-08 广东技术师范大学 一种基于深度学习的房角图像自动分级方法

Similar Documents

Publication Publication Date Title
Haleem et al. Automatic extraction of retinal features from colour retinal images for glaucoma diagnosis: a review
Dutta et al. Glaucoma detection by segmenting the super pixels from fundus colour retinal images
Tavakoli et al. A complementary method for automated detection of microaneurysms in fluorescein angiography fundus images to assess diabetic retinopathy
JP5582772B2 (ja) 画像処理装置及び画像処理方法
SujithKumar et al. Automatic detection of diabetic retinopathy in non-dilated RGB retinal fundus images
Khalil et al. Detection of glaucoma using cup to disc ratio from spectral domain optical coherence tomography images
De La Fuente-Arriaga et al. Application of vascular bundle displacement in the optic disc for glaucoma detection using fundus images
WO2018116321A2 (fr) Procédé de traitement d'image de fond rétinien
Hassan et al. Automated retinal edema detection from fundus and optical coherence tomography scans
US20110243410A1 (en) Method and system for determining the position of an optic cup boundary
Hunter et al. Automated diagnosis of referable maculopathy in diabetic retinopathy screening
Cheng et al. Closed angle glaucoma detection in RetCam images
WO2011108995A1 (fr) Analyse automatique d'images de la chambre antérieure d'un œil
Muramatsu et al. Determination of cup-to-disc ratio of optical nerve head for diagnosis of glaucoma on stereo retinal fundus image pairs
Tan et al. Automatic notch detection in retinal images
Giraddi et al. Optic disc detection using geometric properties and GVF snake
Singh et al. Assessment of disc damage likelihood scale (DDLS) for automated glaucoma diagnosis
Joshi et al. Fundus image analysis for detection of fovea: A review
Ruggeri et al. Analysis of corneal images for the recognition of nerve structures
Khatter et al. Retinal vessel segmentation using Robinson compass mask and fuzzy c-means
Tamilarasi et al. Template matching algorithm for exudates detection from retinal fundus images
Syga et al. A fully automated 3D in-vivo delineation and shape parameterization of the human lamina cribrosa in optical coherence tomography
Lotankar et al. Glaucoma Screening using Digital Fundus Image through Optic Disc and Cup Segmentation
Pandey et al. Automatic detection of red lesions in diabetic retinopathy using shape based extraction technique in fundus image
Pathan et al. A methodological review on computer aided diagnosis of glaucoma in fundus images

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 11751002

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 11751002

Country of ref document: EP

Kind code of ref document: A1