WO2019158916A1

WO2019158916A1 - Imaging device and method of imaging a subject's eye

Info

Publication number: WO2019158916A1
Application number: PCT/GB2019/050383
Authority: WO
Inventors: Louise Allen; Peter Easey
Original assignee: Neocam Limited
Priority date: 2018-02-13
Filing date: 2019-02-13
Publication date: 2019-08-22
Also published as: GB201802322D0; GB2570939A; GB2570939B

Abstract

An imaging device (1) comprising a first light source (6) and a second light source (7), in which the first light source (6) is arranged to provide infrared illumination and the second light source (7) is arranged to provide green or short wavelength visible light illumination, and an image capture device (3) arranged to capture images of a scene, typically a subject's eye, using the infrared illumination and the green or short wavelength visible light illumination. The imaging device can be used to screen for cataracts, retinoblastomas and other media defects of the eye and its ocular media.

Description

IMAGING DEVICE AND METHOD OF IMAGING A SUBJECT’S EYE

This invention relates to an imaging device and a method of imaging a subject’s eye.

Cataracts are an opacity within the lens of the eye and are present at birth in 1 in 2000 babies. It can affect one or both eyes. Reliable screening and early identification of cataract in the initial six to eight weeks of life is crucial, since early curative surgery can prevent life-long blindness.

During the first two to three months of life, critical neural connections are made between the eye and the brain. This process relies on sight. If cataracts are not treated within this time period then, irrespective of whether the condition is subsequently solved, the baby will most likely remain visually impaired as the brain will not connect with the eye.

Retinoblastoma is a potentially fatal rare cancer of the retina affecting 1 in 20,000 children under three years of age. If detected and treated early this type of cancer is curable. In the United Kingdom and most developed countries, babies are screened by non-specialist staff for cataract and retinoblastoma twice: prior to discharge from the maternity ward and subsequently at the six to eight week Neonatal Infant Physical Examination check. A direct ophthalmoscope is the current device used for this screening test. Studies show that approximately 50% of congenital cataracts are missed using this device.

The eye screening staff are not ophthalmic experts and the current technique is not easy, especially since many staff will have never examined a baby with congenital cataract. Quite rightly, they will err on the side of caution which leads to high rates of false positives being referred to ophthalmic specialists. In the United Kingdom, this costs around £180 per general practitioner (GP) referral and blocks hospital appointments needlessly. The reverse also applies: cataracts are missed, leading to late diagnosis and inability to resolve lifelong blindness. 13% of childhood blindness is caused by cataracts being undetected at birth. The economic loss over 10 years due to preventable visual impairment from childhood cataract is in the order of US$5,000 million worldwide and approximately £2 l4m for the United Kingdom National Health Service (NHS) specifically. Ensuring an accurate screening with an easy to use and easy to understand device has significant potential to reduce the volume of lifelong blindness caused by baby cataracts, free up ophthalmic consultations (reduce waiting times), reduce GP referrals (improve budget performance) and give the NHS significant financial and performance gains.

The fundus (back of the eye) consists of many layers. The innermost layer is the light sensing, film like, transparent retina. Under the retina is a pigment layer which varies in darkness dependent on the skin colour of the individual. Under the pigment layer is the choroid - a layer full of blood vessels which supplies nutrients to the eye tissues. White light entering the eye passes through the cornea, lens and retina reflecting back off the red choroidal and deeper pigmented layer behind the retina at the back of the eyeball. It is this effect that gives rise to red eye or red reflex in photos and on ophthalmic examination.

Opacities in the ocular media e.g. cataract, cause a dark shadow on retro-illumination from the red reflex, whereas retinoblastoma causes a white reflex when using visible light.

Due to differences in the melanin content of the pigmented layer, those of Caucasian ethnicity typically have a bright red reflex, whereas individuals of Asian or Afro- Caribbean ancestry have a muted red / brown or occasionally yellowish choroidal reflex, which can be harder to differentiate using white light. Additionally, use of visible light causes pupillary constriction which makes assessment of the red reflex more difficult.

Current screening utilises a direct ophthalmoscope, a system where the examiner studies the reflection of light (“red reflex”) from the back of the eye to a coaxial white light source. Most non- specialists find this technique difficult particularly in babies, who tend to have smaller pupils than adults, and will wriggle and shut their eyes when confronted with a bright light. Pupil-dilating drops are not used for screening examinations. Assessment of the red reflex is particularly difficult in babies of Asian or Afro-Caribbean babies since the increased ocular pigmentation reduces the brightness and colour properties of the reflected red light. According to a first aspect of the invention, there is provided an imaging device comprising a first light source and a second light source, in which the first light source is arranged to provide infrared illumination and the second light source is arranged to provide green visible light illumination, and an image capture device arranged to capture images of a scene using the infrared illumination and the green visible light illumination.

According to a second aspect of the invention, there is provided an imaging device comprising a first light source and a second light source, in which the first light source is arranged to provide infrared illumination and the second light source is arranged to provide visible light illumination, in which no more than a fraction of the power of the visible light illumination has a wavelength longer than 570nm, the fraction being at most 5%, and an image capture device arranged to capture images of a scene using the infrared illumination and the visible light illumination.

With the above two aspects, we have appreciated that it is possible to provide two different illuminations in order to detect abnormalities in the eye; the choroid has maximal reflectance to red and infra-red (IR) light and minimal reflectance to green or short wavelength light. Using the first light source (infrared illumination) will create a bright choroidal reflection at the pupil when suitably imaged, with any opacities within the eye showing as darker patches on retro-illumination. Using the second light source (green/visible light of short wavelength), a normal ocular fundus will reflect only a small proportion of light from this portion of the visible spectrum and so a dark pupil image will result; any (typically white) media opacity - for example a retinoblastoma - will be imaged as a bright reflection on a dark background. In both cases, the changes are visible regardless of ethnicity. Furthermore, there is no or little pupil reaction to infrared illumination, and only moderate constriction to low intensity green or short wavelength light.

As such, the scene will typically comprise a subj ect’s eye, typically the ocular media of an eye of a subject. The scene may comprise opacities in the optical media.

The imaging device may comprise a control circuit which is arranged to control the capture of images. The control circuit may be arranged so as to sequentially cause the capture of images with different illumination, typically within 500ms of each other. In particular, the control circuit may be arranged to cause the capture of at least one first image with the first light source (and typically not the second light source) illuminating the scene and at least one second image with the second light source (and typically not the first light source) illuminating the scene. Thus, the differing effects of the illumination can be separately recorded and compared.

Furthermore, the control circuit may be arranged to capture at least one third image without either of the first or second light source illuminating the scene. Each third image can be used to determine whether the first and/or second light sources are illuminating the scene, as there should be an increase in image intensity in the first and second images as compared to the third images.

The order in which each first, second and third image is taken may be irrelevant, so that the second images could be taken before or after the first or the third images, and the third images could be taken before or after the first or second images. Where multiple first, second and/or third images are taken, first, second and/or third images could be interleaved (for example, first image, second image, first image, third image, second image in that order).

The infrared illumination may comprise infrared radiation having a power spectrum in which 50%, 75%, 90%, 95% or even all of the power transmitted has a longer wavelength than 700 nanometres (nm) (that is, outside the visible spectrum). Typically at least at least 50%, 75%, 90%, 95% or even all of the power transmitted will have a shorter wavelength than l400nm (700nm to l400nm being near infrared). In a particularly preferred embodiment, at least 50%, 75%, 90%, 95% or even all of the power transmitted will have a wavelength of 850nm, plus or minus 20nm, lOnm or 5 nm .

The green/visible light illumination may comprise visible light having a power spectrum in which 45% to 100% of the power transmitted has a wavelength between 490 and 570nm. As such, the fraction may be 5%, 4%, 3%, 2%, or 1%.

The image capture device may be a charge coupled device (CCD) or other image sensor. Typically, the image capture device may be sensitive to the illumination of the first and second light sources. The image capture device may be monochromatic, in that it records images composed of image intensity and not hue. As such, the image capture device may be sensitive to both visible and infrared light,

The image capture device may be arranged to collect light along an optical axis; the imaging device may comprise an optical system arranged to collect light along the optical axis and to pass it to the image capture device. The first and second light sources may each provide illumination as a beam non-parallel to the optical axis, but typically within +/-one degree of parallel thereto, in order to provide the best illumination of the interior of the eye and avoid Purkinje images (images of the imaging device in the captured images).

Typically, the imaging device (and any optical system) may be arranged so that the image capture device is arranged to capture images of a subject’s eye when within a range of distances from the subject’s eye; typically the range of distances would comprise 10 to 500mm, or between 100 and 300mm, or between 150 and 250mm. The range will typically comprise a point or region where the beams of each of the first and second light sources and the optical axis overlap.

Each of the first and second light sources may be arranged so that, when they are providing illumination, they each provide simultaneous illumination of at least 25% of the ocular fundus.

The imaging device may comprise an electronic circuit to power the first light source and/or the second light source.

The imaging device may comprise an electronic circuit to manage the first light source and/or the second light source.

The imaging device may comprise an electronic circuit to power and manage the first light source and/or the second light source, preferably such that light of a hazardous intensity is prevented from impinging on a patient’s eye. The powering and management of the first and/or second light sources is in one embodiment in compliance with BS EN ISO l5004_2. A single electronic circuit may power and manage the first light source and/or the second light source.

The imaging device may comprise a housing in which the first and second light sources and the image capture device are supported. The imaging device may be portable in that it may have a mass of less than lOkg (preferably less than lkg). The imaging device may be provided with a power source, such as a battery, typically within the housing.

The housing may be provided with a high contrast pattern for attracting the attention of young subjects such as babies. The imaging device may also comprise a noise- emitting device for attracting the attention of young subjects such as babies.

The imaging device may comprise a memory arranged to store images captured by the image capture device. Additionally or alternatively, it may be provided with an output for the images, such as a transmission port (e.g. a USB port), network connection (wireless or wired).

The control circuit may be arranged so as to process the captured images so as to determine whether there may be an abnormality in the eye of a subject. As such, the control circuit may be arranged so as to determine a presence of a potential abnormality of a first type if a dark area appears against a lighter background within a subject’s eye in a first image of the subject’s eye. The first type may comprise potential cataracts.

When determining a potential abnormality of the first type, the control circuit may be arranged to ignore any areas that are lighter than a threshold; such areas may represent Purkinje images of the imaging device. The threshold may be set based upon a mean pixel intensity within the subject’s pupil; for example, the threshold may be at least one or two standard deviations above mean pupil pixel intensity.

The control circuit may also be arranged to determine a presence of a potential abnormality of a second type if a light area appears against a darker background within a subject’s eye in a first and/or second image of the subject’s eye. The second type may comprise potential retinoblastoma or white coloured cataract. When determining a potential abnormality of the second type, the control circuit may be arranged to disregard any circumscribed circular areas. Typically, these would represent reflections of the incident light from the anterior and posterior surface of the cornea and the anterior surface of the lens (Purkinje reflexes).

The control circuit may also be arranged to compare each first and second images on the one hand with each third image on the other hand. If there is an increase in pixel intensity (typically averaged over the captured image) between each third image on the one hand and the first and second images on the other hand, then the illumination of the respective first or second image was functioning.

According to a third aspect of the invention, there is provided a method of imaging a subject’s eye, comprising:

sequentially illuminating the eye with infrared illumination and green visible light illumination; and

capturing images of the eye as illuminated with the infrared illumination and with the green visible light illumination. According to a fourth aspect of the invention, there is provided a method of imaging a subject’s eye, comprising:

sequentially illuminating the eye with infrared illumination and visible light illumination, in which no more than a fraction of the power of the visible light illumination has a wavelength longer than 570nm, the fraction being at most 5%; and capturing images of the eye as illuminated with the infrared illumination and with the visible light illumination.

With the above two aspects, we have appreciated that it is possible to provide two different illuminations in order to detect abnormalities in the eye. Using the infrared illumination, then a normal choroidal reflection will be light, with any opacities in the eye will show as darker patches. Using the green or (short wavelength) visible light, there will normally be little reflection in the eye and so a dark image will result; any media opacity - for example a retinoblastoma - will show up as a light patch. In both cases, the changes are visible regardless of ethnicity. Furthermore, there is no or little pupil reaction to infrared illumination, and only moderate constriction to low intensity green or short wavelength light.

The method may comprise imaging the ocular media of the subject’s eye, typically comprising opacities in the optical media.

The method may also comprise capturing at least one image without either of the illuminations. The infrared illumination may comprise infrared radiation having a power spectrum in which 50%, 75%, 90%, 95% or even all of the power transmitted has a longer wavelength than 700 nanometres (nm) (that is, outside the visible spectrum). Typically at least at least 50%, 75%, 90%, 95% or even all of the power transmitted will have a shorter wavelength than l400nm (700nm to l400nm being near infrared). In a particularly preferred embodiment, at least 50%, 75%, 90%, 95% or even all of the power transmitted will have a wavelength of 850nm, plus or minus 20nm, lOnm or 5nm.

The green/visible light illumination may comprise visible light having a power spectrum in which 50%, 75%, 90%, 95% or even all of the power transmitted has a wavelength between 490 and 570nm. As such, the fraction may be 5%, 4%, 3%, 2%, or 1%.

The images may be captured so as to represent the intensity of radiation within the spectra transmitted by the illuminations. The images captured may be monochromatic, in that they record image intensity and not hue.

The method may comprise processing the captured images so as to determine whether there may be an abnormality in the eye of the subject. As such, the method may comprise determining a presence of a potential abnormality of a first type if a dark area appears against a lighter background (retro-illuminated) within a subject’s eye in a first image of the subject’s eye. The first type may comprise potential cataracts. When determining a potential abnormality of the first type, the method may comprise ignoring any areas that are lighter than a threshold; such areas may represent Purkinje images of the imaging device. The threshold may be set based upon a mean pixel intensity within the subject’s pupil; for example, the threshold may be at least one or two standard deviations above mean pupil pixel intensity.

The method may comprise determining a presence of a potential abnormality of a second type if a light area appears against a darker background within a subject’s eye in a first and/or second image of the subject’s eye. The second type may comprise potential retinoblastoma or white coloured cataract.

When determining a potential abnormality of the second type, the method may comprise disregarding any circumscribed circular areas. Typically, these would represent reflections of the incident light from the anterior and posterior surface of the cornea and the anterior surface of the lens (Purkinje reflexes).

The method may also comprise comparing each first and second images on the one hand with each third image on the other hand. If there is an increase in pixel intensity (typically averaged over the captured image) between the first and second images on the one hand and each third image on the other hand, then the illumination of the respective first or second image was functioning.

There now follows, by way of example only, description of an embodiment of the invention, described with reference to the accompanying drawings, in which:

Figure 1 shows a schematic view of an imaging device in accordance with an embodiment of the invention;

Figure 2 shows a photograph of a subject’s eye taken with infrared illumination, where the subject has a normal eye;

Figure 3 shows a photograph of a subject’s eye taken with infrared illumination, where the subject has a cataract; Figure 4 shows a further photograph of a subject’s eye taken with infrared illumination, where the subject has a normal eye;

Figure 5 shows a photograph of a subject’s eye taken with green illumination, where the subject has a normal eye;

Figure 6 shows a further photograph of a subject’s eye taken with infrared illumination, where the subject has a cataract;

Figure 7 shows a photograph of a subject’s eye taken with infrared illumination where the subject has a retinoblastoma; and

Figure 8 shows a photograph of a subject’s eye taken with green illumination, where the subject has a retinoblastoma.

Figure 1 of the accompanying drawings shows an imaging device 1 in accordance with an embodiment of the invention.

The imaging device 1 comprises a handheld housing 2 capable of being operated single-handed, in which there is provided a high mega-pixel NOIR camera 3 having a window 4 which passes light captured along an optical axis 5 outside of the housing 2 to the camera 3 inside the housing 2. The camera 3 can capture images, typically still but potentially also video, in both visible monochrome and infrared.

Two light sources 6, 7 are provided within the housing. A first light source 6 provides infrared illumination, whereas a second light source 7 provides visible green light illumination, although blue or violet light could also be used in place of green light. Both light sources are formed as light emitting diodes (LEDs) . Typically, the first light source 6 would provide its output around 850nm, whereas the second light source 7 would provide its output over a 490 and 570nm wavelength range. The light sources 6, 7 are slightly off-centre with respect to the optical axis 5 (to reduce catch light or Purkinje effects). A control circuit 8 of the form of a microprocessor and storage is provided in the housing 2, along with a power source comprising at least one (typically rechargeable) battery. The imaging device 1 can therefore be used to image a subject’s eye 100. Typically, the subject will be a young human, typically an infant under one year of age. The imaging device uses the reflectance properties of the fundus to infrared and green light to improve and optimise the contrast between light reflecting from the choroid 101 and the reflectance of opacities within the ocular media. Using infrared illumination, the normal choroidal reflection looks white and opacities cause a grey / black shadow. Using green (or blue or violet) light, there is very little reflection, so the pupil looks dark and any media opacity which reflects green (etc) light, e.g. a white retinoblastoma, will look white when the image is viewed in monochrome. The effects can be summarised in the following table:

As such, the control circuit 8 is arranged to capture at least three images using the camera 3 : · A first image or images, taken in an infrared spectrum using infrared illumination from the first light source 6.

• A second image or images, taken using monochrome visible light with visible green light illumination from the second light source 7.

• A third image or images, taken without any illumination.

As such, the third image can be used to check that the illumination is functioning; in both infrared and visible spectrums with the appropriate illumination functioning, the pixel intensity should be higher with the illumination on (first or second images) than with the illuminations both off (the third image(s)) .

To prove that the light sources are functioning, the difference between images 1 and 3 and images 2 and 3 shall then be taken as a basis for deciding whether the lights were on. The pixels of each image subtraction shall be added to generate a number that corresponds to how much illumination the light sources 6, 7 have put into the images. A difference above a threshold shall indicate that the illumination was indeed functioning. The device shall auto power off after xlO minutes of inactivity. The device has a touchscreen 10 for patient data entry, image alignment and review. There will typically be facilities to upload the captured image(s) via a removable memory card, a wired communication link such as USB and/or a wireless link such as Bluetooth or Wifi.

Thus, a screener using the device can quickly check for the defects set out in the third and fourth columns of the table above, which are much more clearly apparent than would be the case with the prior art (set out in the second column of the table).

The housing front has a distinctive high contrast pattern 11 to attract the attention of the subject - typically a baby. A buzzer 12 that produces a noise to attract the subject’s attention during illumination.

The camera 3 is of fixed focal length with a deep depth of field; this is to facilitate a rapid image capture. The first and second images are captured and displayed in monochrome to enhance contrast ratios. The power levels of the first and second light sources 6, 7 are within medical limits.

Additionally or alternatively, the control circuit 8 may be arranged and/or programmed to detect potential defects in a subject’s eye by detecting the features set out in the third and fourth columns of the table above.

In such a case, Purkinje images are more of a nuisance for the first images being used for detection of a small central cataract - a small central shadow from a congenital cataract might be occluded by the Purkinje image if the Purkinje image is central, hence the need to slightly decentre the beam. Since these corneal reflections are much brighter than the diffuse reflection from the back of the eye, the control circuit 8 will be arranged to ignore a light patch brighter than two standard deviations above the mean pupil pixel intensity for each first image.

Retinoblastomas are never seen as a circumscribed bright round circle - they cause a diffuse white reflectance from the back of the eye. The control circuit 8 can therefore be arranged to ignore any white reflex of the calculated size which is circumscribed (surrounded by either the red reflection from the back of the eye or the darkness of the iris if decentred off the pupil) since this would be a reflection from the cornea. Additional features may be added to enable further functions. For ocular use, this may include a white LED to enable imaging of the anterior segment of the eye; lenses can be added to enable imaging of the retina. A blue visible light LED can be added to enable detection and imaging of fluorescein dye staining in corneal pathology. An otoscope attachment can be provided to enable imaging of the eardrum.

Claims

1. An imaging device comprising a first light source and a second light source, in which the first light source is arranged to provide infrared illumination and the second light source is arranged to provide visible light illumination in which no more than a fraction of the power of the visible light illumination has a wavelength longer than 570nm, the fraction being at most 5%, and an image capture device arranged to capture images of a scene using the infrared illumination and the visible light illumination, in which the scene comprises ocular media of an eye of a subject.

2. The imaging device in which the scene comprises opacities in the optical media.

3. The imaging device of claim 1, comprising a control circuit which is arranged to control the capture of images, and in which the control circuit is arranged so as to sequentially cause the capture of images with different illumination within 500ms of each other.

4. The imaging device of claim 3, in which the control circuit is arranged to cause the capture of at least one first image with the first light source and not the second light source illuminating the scene and at least one second image with the second light source and not the first light source illuminating the scene.

5. The imaging device of claim 3 or claim 4, in which the control circuit is arranged to capture at least one third image without either of the first or second light source illuminating the scene.

6. The imaging device of any preceding claim, in which the image capture device is arranged to collect light along an optical axis, and in which the first and second light sources each provide illumination as a beam non-parallel to the optical axis.

7. The imaging device of claim 6, in which each beam is within +/- one degree of parallel to the optical axis.

8. The imaging device of claim 6 or claim 7, in which the imaging device is arranged so that the image capture device is arranged to capture images of a subject’s eye when within a range of distances from the subject’s eye; the range will comprising a point or region where the beams of each of the first and second light sources and the optical axis overlap.

9. The imaging device of any preceding claim, in which each of the first and second light sources are arranged so that, when they are providing illumination, they each provide simultaneous illumination of at least 25% of the ocular fundus of the subject’s eye.

10. The imaging device of any preceding claim, provided with at least one of a high contrast pattern for attracting the attention of young subjects such as babies and a noise-emitting device for attracting the attention of young subjects such as babies.

11. The imaging device of any preceding claim, in which the control circuit is arranged so as to process the captured images so as to determine whether there may be an abnormality in the eye of a subject.

12. The imaging device of claim 11 as it depends from claim 4, in which the control circuit is arranged so as to determine a presence of a potential abnormality of a first type if a dark area appears against a lighter background within a subject’s eye in a first image of the subject’s eye.

13. The imaging device of claim 12, in which, the control circuit is arranged so that, when determining a potential abnormality of the first type, it ignores any areas that are lighter than a threshold.

14. The imaging device of claim 13, in which the threshold is set based upon a mean pixel intensity within the subject’s pupil.

15. The imaging device of claim 11 as it depends from claim 4 or any of claims 12 to 14, in which the control circuit is arranged to determine a presence of a potential abnormality of a second type if a light area appears against a darker background within a subject’s eye in a first and/or second image of the subject’s eye.

16. The imaging device of claim 15, in which the control circuit is arranged so that, when it is determining a potential abnormality of the second type, it disregards any circumscribed circular areas.

17. The imaging device of claim 5 or any of claims 6 to 16 as dependent thereon, in which the control circuit is arranged to compare each first and second image on the one hand with each third image on the other hand, and to determine an increase in pixel intensity between each third image on the one hand and the first and second images on the other hand.

18. The imaging device of any preceding claim, in which the visible light illumination is green light illumination.

19. A method of imaging the ocular media of a subject’s eye, comprising:

sequentially illuminating the eye with infrared illumination and visible light illumination in which no more than a fraction of the power of the visible light illumination has a wavelength longer than 570nm, the fraction being at most 5%; and capturing images of the optical media of the eye as illuminated with the infrared illumination and with the visible light illumination.

20. The method of claim 19, comprising capturing at least one image without either of the illuminations.

21. The method of claim 19 or claim 20, in which the images are of opacities in the ocular media of the eye.

22. The method of any of claims 19 to 21, in which the visible light illumination is green light illumination.