WO2013160619A1 - Method for analysing a photograph of the retina or a bank of photographs of the retina - Google Patents

Method for analysing a photograph of the retina or a bank of photographs of the retina Download PDF

Info

Publication number
WO2013160619A1
WO2013160619A1 PCT/FR2013/050918 FR2013050918W WO2013160619A1 WO 2013160619 A1 WO2013160619 A1 WO 2013160619A1 FR 2013050918 W FR2013050918 W FR 2013050918W WO 2013160619 A1 WO2013160619 A1 WO 2013160619A1
Authority
WO
Grant status
Application
Patent type
Prior art keywords
view
field
size
method according
photograph
Prior art date
Application number
PCT/FR2013/050918
Other languages
French (fr)
Inventor
Xiwei ZHANG
Etienne Decenciere
Guillaume THIBAULT
Original Assignee
Association Pour La Recherche Et Le Developpement De Methodes Et Processus Industriels "Armines"
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

Links

Classifications

    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B3/00Apparatus for testing the eyes; Instruments for examining the eyes
    • A61B3/10Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions
    • A61B3/12Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions for looking at the eye fundus, e.g. ophthalmoscopes
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B3/00Apparatus for testing the eyes; Instruments for examining the eyes
    • A61B3/10Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions
    • A61B3/14Arrangements specially adapted for eye photography
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/00597Acquiring or recognising eyes, e.g. iris verification
    • G06K9/00604Acquisition

Abstract

The invention relates to a method for analysing a photograph of the retina, comprising a step of measuring a dimension representing a field of vision (10) of a pre-determined shape visible on the photograph, followed by a step of adapting at least one size parameter to be used for the detection of retinal lesions, taking into account said dimension representing a field of vision (10).

Description

A method of analyzing a retinal photograph or a bank of

Retinal photographs Technical field and prior art

The invention is in the field of automatic analysis of medical images for purposes of preventing, diagnosing or monitoring the evolution over time of a pathology. It applies specifically to digital images of the retina, possibly present in an image bank.

Image analysis of the retina becomes an essential element in the detection, diagnosis assistance and monitoring of retinal diseases. image processing methods are being developed in many laboratories, and begin to be marketed to meet this need.

For various reasons, image scale, in other words the size in microns of the surface of the retina corresponding to a pixel on the image is not known. Moreover, acquisition systems continue to evolve, bringing new resolutions and therefore new scales. But most of image processing methods are set according to the scale of the image. It is therefore essential to determine the scale from the images.

Determine the absolute scale of a given image is difficult in practice, because the dimensions of the patient's eye and the exact optical characteristics of the different environments which are traversed by the light beams are not known. Moreover, should be adapted a procedure for determining the absolute scale to each acquisition system, which would be a time consuming task and impractical to implement, since, firstly, the number of available systems and, secondly, the steady development of new equipment with different features.

An indirect determination of the scale is therefore an essential step in any method of analyzing an image of the retina. In the case of stock images from different sources, the images are several different scales. In all cases reported in the literature, the authors use the definition images (ie the dimensions of the image in pixels) to perform this normalization, but the methods based on this value give sufficiently accurate results.

For example, Figure 1 shows two images of the same definition, vested with the same angle. With a method based on the image size, they would be considered the same scale. But they were acquired with the same angle of view, and therefore the field of view visible in both images should, as a first approximation, be of the same actual size if they were actually on the same scale. Remember that the field of view of an imaging device, here the fundus, is the space in the picture where an image was actually photographed. In Figure 1, the fields of view are circular. although they are not the same size is seen (in pixels), and mentioned method is not satisfactory. Another type of possible method is based on the size of the visible anatomical structures. In theory, the size of anatomical structures could indeed be a reference. In practice, either these anatomical structures vary too greatly from one subject to another (the radius of the optical disk, for example, can vary from simple to double) or they are difficult to extract automatically. Thus, the distance between the center of the optical disk and the center of the fovea is a reliable reference, but unfortunately not only its automatic extraction is difficult and time consuming calculation for the optical disc and the fovea are not easy to spot automatically, but in addition one or the other of the structures may be missing from the image. Summary of the Invention

The invention in this context on the spatial calibration images of the retina, acquired with an image acquisition device of the fundus such as a fundus camera, since a field of view appears on the image. Specifically, there is provided an analysis of a retinal photography method, comprising a step of measuring a representative size of a field of view of predetermined shape visible in the photograph, then a step of adapting at least one size parameter to be used for detecting lesions or retinal anatomical structures taking into account said representative dimension of a field of view.

If the size parameter is a linear parameter, the adaptation may be a multiplication by a factor proportional to said representative dimension of a field of view. If the size parameter is a surface parameter, the adaptation may be a multiplication by a factor proportional to the square of said representative dimension of a field of view.

The invention thus proposes using a relative scale: it is then taken as the reference one or more images acquired with a given system, under the same conditions, and the scale of the other images is given in relation to the scale of reference images, by an adaptation taking account of said representative dimension of a field of view using, for example, a simple multiplicative factor proportional to said representative dimension of a field of view or to its square. Suppose that all the images were acquired with the same angle. So we can assume that the representative dimension of the field of view, usually the width is the same on all images. Considering therefore this width as a constant, we get a common reference, which allows us to calculate the difference in scale between any two images. Suppose that all the images have not been acquired with the same angle. So, provided you have the angle of view for both reference images than other images, one can also determine the multiplication factor corresponding to the relative scale, using the representative dimension of the field of view and knowledge of these angles.

measuring the reference is specified that can be done long before the analysis of photography to analyze, including a third party who merely provide the laboratory conducting the analysis a multiplication factor corresponding to the relative scale .

In one implementation, the method uses a reference retinal photograph which was also measured said representative dimension of a field of view of the predetermined shape and wherein said size parameter has previously been determined.

It is also suggested - and this is the preferred application of the invention - a method of analyzing a library of retinal photographs, the photographs of the bank having fields of view, each photograph of the bank being analyzed according to the mentioned method, and a multiplication of at least one size setting for the detection of retinal lesions for each photograph being powered by a factor proportional to the representative dimension of the corresponding field of view. This method can comprise in particular a step of detecting the field of view based on its high contrast relative to the dark background of the photograph, and a step of detection of retinal damage using an image processing method and said size setting. However, when the process is implemented by software, the detection steps may be implemented by separate software or separate software.

In this method, the image or images to be analyzed may be in black and white or in color. The invention also relates to a computer program comprising instructions, which when executed by a microprocessor, cause the implementation of a method as mentioned above.

BRIEF DESCRIPTION OF FIGURES

The invention will now be described in conjunction with the following accompanying figures:

Figure 1 shows two retinal photographs.

Figure 2 shows an implementation algorithm of the invention.

Detailed disclosure of the invention

The field of view, which is the generally circular area that appears on these images (see Fig. 1 for two examples) is easy to automatically extract through the image processing, because of its high contrast relative to the dark background of the 'picture. width or diameter in pixels We can then measure, which is referenced 10 in Figure 1. The field of view may not be circular and be truncated example circular or elliptical. then chosen to measure the apparant large diameter, which is a dimension.

In order to apply any for analysis of these images algorithm, size parameters of the applicant, simply choose the settings for a reference image, or a set of reference images acquired under identical conditions, then to generalize these parameters, any other image acquired with the same angle.

This method was validated by comparing the scale factors obtained with the scale factors obtained through manual measurements of the distance between the center of the fovea and the center of optical disk. The correlation is almost perfect. In the case of an image that would not have been acquired with the same angle as the image or reference images, provided you have the angle of view, both for the image or images reference for the image in question, you can perform trigonometric calculation to proceed with the adaptation of processing parameters. The industrial feasibility of this invention was demonstrated in the TeleOphta project, funded by the French National Research Agency (France) through the TecSan program. One objective of this project is to develop tools for image classification of the back of the eye to help in the detection of retinal diseases, especially diabetic retinopathy. Diabetic retinopathy, a disease of the retina caused by diabetes is the leading cause of blindness in 20-65 years. Health organizations recommend that diabetic patients to spend at least an examination of the fundus year. screening networks, such as OPHDIAT were deployed to assist in screening for diabetic retinopathy. Given the growing number of diabetics, and the workload ophthalmologists, these networks are already saturated.

One solution is to automatically detect clear sound images to alleviate the workload of ophthalmologists. These automatic methods involve image analysis, and therefore require scale calibration methods. The feasibility of the invention was demonstrated by data from the OPHDIAT network. The invention has been implemented as follows:

For a given image, detecting the field of view using image processing methods;

· Measure the width of the field of view;

• Detection of retinal lesions, e.g. microaneurysms, using image processing methods after having adjusted the size parameters (e.g. lower bound and upper bound of microaneurysms size) through the width previously calculated by the rule of three mentioned. In this case, the size of terminals of the linear parameters, and they are multiplied by a parameter proportional to the width of the field of view. If the size parameter was an area-setting, it would have been multiplied by a factor proportional to the square of the width of the field of view.

As shown in Figure 2, the steps performed thus the detection field of view for a reference image and measuring its width (step 100), for determining the size parameters of the reference image to detect retinal damage (step 110), then to an image analyzing data derived from the image database, the detection field of view and the measurement of its width (step 120), the application for the image to be analyzed to a rule of three to the reference image size parameters using the ratio of the field of view widths (step 130), and the search for retinal damage on the image to be analyzed (step 140). Then repeats the steps 120 to 140 for another image to be analyzed removed from the image bank.

It is specified that the retinal camera can be a black and white retinal or retinal colors.

The method may be implemented by software, which performs at least the measurement of the width of the field of view for the image to be analyzed and the step of multiplying the useful size parameter in the search for retinal damage. The software can also perform the other steps, but they can be implemented by separate software, for example pre-existing. The invention is not limited to the disclosed embodiment, but extends to all variants within the scope of the claims.

Claims

A method of analyzing a retinal photography, comprising a step of measuring (120) a size of a field of view (10) of predetermined shape visible in the photograph, then an adaptation step (130) at least one size setting for the detection of retinal lesions taking into account said size of a field of view (10) and the angle of view of acquiring the photography.
The analysis method according to claim 1, wherein the size parameter being a linear parameter adaptation is multiplication by a factor proportional to said representative dimension of a field of view.
The analysis method according to claim 1, wherein the size parameter is a surface parameter adaptation is multiplication by a factor proportional to the square of said dimension of a field of view.
The analysis method according to one of Claims 1 to 3, using a reference retinal photograph which was also measured (100) said size of a field of view of the predetermined shape and wherein said size parameter been previously determined (110).
A method of analyzing a library of retinal photographs, bank photographs showing the fields of view (10) of the same shape, each photograph of the bank is analyzed according to a method according to one of claims 1 to 4, and adapting (130) at least one size setting for the detection of retinal damage being done for each photograph using the dimension of a corresponding field of view.
The analysis method according to one of Claims 1 to 5, comprising a detection step (120) of the field of view based on its high contrast relative to the dark background of the photograph.
The analysis method according to one of Claims 1 to 6, comprising a detection step (140) of retinal damage using an image processing method, and said size setting.
The analysis method according to one of Claims 1 to 7, the image or the images are in black and white or in color.
9. Computer program comprising instructions, which when executed by a microprocessor, cause the implementation of a method according to one of claims 1 to 8.
PCT/FR2013/050918 2012-04-27 2013-04-25 Method for analysing a photograph of the retina or a bank of photographs of the retina WO2013160619A1 (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
FR1253929 2012-04-27
FR1253929A FR2989875A1 (en) 2012-04-27 2012-04-27 Method of analysis of a retinal photograph or a bank of photographs of retina

Publications (1)

Publication Number Publication Date
WO2013160619A1 true true WO2013160619A1 (en) 2013-10-31

Family

ID=48468635

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/FR2013/050918 WO2013160619A1 (en) 2012-04-27 2013-04-25 Method for analysing a photograph of the retina or a bank of photographs of the retina

Country Status (2)

Country Link
FR (1) FR2989875A1 (en)
WO (1) WO2013160619A1 (en)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110051089A1 (en) * 2009-08-27 2011-03-03 Canon Kabushiki Kaisha Fundus imaging apparatus and method therefor

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110051089A1 (en) * 2009-08-27 2011-03-03 Canon Kabushiki Kaisha Fundus imaging apparatus and method therefor

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Giancardo L., Meriaudeau F., Karnowski T.P., Chaum E. and Tobin K.: "Quality Assesment of Retinal Fundus Images using Elliptical Local Vessel Density" In: 2010, Intech, XP002692748, ISBN: 978-953-7619-57-2 pages 201-224, DOI: 10.5772/7618, section Vessel Segmentation; page 211 - page 212 *
None

Also Published As

Publication number Publication date Type
FR2989875A1 (en) 2013-11-01 application

Similar Documents

Publication Publication Date Title
Gang et al. Detection and measurement of retinal vessels in fundus images using amplitude modified second-order Gaussian filter
Monaghan Role of real time 3D echocardiography in evaluating the left ventricle
Chapman et al. Computer algorithms for the automated measurement of retinal arteriolar diameters
Wilson et al. Computerized analysis of retinal vessel width and tortuosity in premature infants
Niemeijer et al. Automated measurement of the arteriolar-to-venular width ratio in digital color fundus photographs
US5233517A (en) Early glaucoma detection by Fourier transform analysis of digitized eye fundus images
Grewal et al. Diagnosis of glaucoma and detection of glaucoma progression using spectral domain optical coherence tomography
Phillips et al. Quantification of diabetic maculopathy by digital imaging of the fundus
Spencer et al. Automated detection and quantification of microaneurysms in fluorescein angiograms
US20100202677A1 (en) Image processing apparatus and image processing method for a tomogram of an eye region
Corona et al. Digital stereo image analyzer for generating automated 3-D measures of optic disc deformation in glaucoma
US20110249911A1 (en) Spectral unmixing for visualization of samples
JP2004321793A (en) Method and system for computational analysis of skin image
Mozley et al. Measurement of tumor volumes improves RECIST-based response assessments in advanced lung cancer
Sigal et al. Eye-specific IOP-induced displacements and deformations of human lamina cribrosa
Arnold et al. Point-of-care assessment of microvascular blood flow in critically ill patients
CN104964973A (en) Test paper reading and analyzing method and system based on mobile terminal camera
Dias et al. Retinal image quality assessment using generic image quality indicators
US20130058553A1 (en) Image processing apparatus and image processing method
Jelinek et al. An automated microaneurysm detector as a tool for identification of diabetic retinopathy in rural optometric practice
JP2007068928A (en) Eyeball surface analyzing system
Zubal et al. Optimized, automated striatal uptake analysis applied to SPECT brain scans of Parkinson's disease patients
JP2007114843A (en) Quality deciding device
Zinchuk et al. Quantifying spatial correlations of fluorescent markers using enhanced background reduction with protein proximity index and correlation coefficient estimations
US20130226008A1 (en) Calculating Conjunctival Redness

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: 13723868

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase in:

Ref country code: DE

122 Ep: pct app. not ent. europ. phase

Ref document number: 13723868

Country of ref document: EP

Kind code of ref document: A1