Title: Ocular fundus auto imager
Abstract: An ocular fundus imager (8) automatically aligns fundus illuminating rays to enter the pupil (P) and to prevent corneal reflections from obscuring the fundus image produced. Focusing the produced fundus image is automatically performed and is based upon the fundus image iself. A head restraint for the patient undergoing examination is in the form of a pair of spectacles which is not only easy to use accurately but significantly reduce the gross alignment between the optical system (8) and the patent's pupil (P).
Patent Number: 7,025,459 Issued on 04/11/2006 to Cornsweet, et al.
| Inventors:
|
Cornsweet; Tom N. (Prescott, AZ);
Buck; Gary F. (Prescott, AZ)
|
| Assignee:
|
Visual Pathways, Inc. (Prescott, AZ)
|
| Appl. No.:
|
311492 |
| Filed:
|
July 6, 2001 |
| PCT Filed:
|
July 6, 2001
|
| PCT NO:
|
PCT/US01/21410
|
| 371 Date:
|
December 16, 2002
|
| 102(e) Date:
|
December 16, 2002
|
| PCT PUB.NO.:
|
WO02/05705 |
| PCT PUB. Date:
|
January 24, 2002 |
| Current U.S. Class: |
351/208; 351/200; 351/205; 351/206; 351/221 |
| Current Intern'l Class: |
A61B 3/14 (20060101); A61B 3/10 (20060101) |
| Field of Search: |
351/200,205,206,208,209,210,211,214,221,246
|
References Cited [Referenced By]
U.S. Patent Documents
| 3915564 | Oct., 1975 | Urban.
| |
| 4019813 | Apr., 1977 | Cornsweet et al.
| |
| 4187014 | Feb., 1980 | Kato et al.
| |
| 4281926 | Aug., 1981 | Cornsweet.
| |
| 4283124 | Aug., 1981 | Matsumura.
| |
| 4329049 | May., 1982 | Rigg et al.
| |
| 4405215 | Sep., 1983 | Sano et al.
| |
| 4436388 | Mar., 1984 | Takahashi et al.
| |
| 4469416 | Sep., 1984 | Isono.
| |
| 4526450 | Jul., 1985 | Suzuki et al.
| |
| 4544248 | Oct., 1985 | Nunokawa.
| |
| 4579430 | Apr., 1986 | Bille.
| |
| 4580885 | Apr., 1986 | Takahashi.
| |
| 4591249 | May., 1986 | Takahashi et al.
| |
| 4666268 | May., 1987 | Ito.
| |
| 4673264 | Jun., 1987 | Takahashi.
| |
| 4690525 | Sep., 1987 | Kobayashi et al.
| |
| 4715703 | Dec., 1987 | Cornsweet et al.
| |
| 4732466 | Mar., 1988 | Humphrey.
| |
| 4773749 | Sep., 1988 | Ohtomo et al.
| |
| 4834526 | May., 1989 | Nunokawa.
| |
| 4902121 | Feb., 1990 | Shinn.
| |
| 4989023 | Jan., 1991 | Sakurai et al.
| |
| 4993827 | Feb., 1991 | Benedek et al.
| |
| 5064286 | Nov., 1991 | Ai et al.
| |
| 5072731 | Dec., 1991 | Taratuta et al.
| |
| 5090799 | Feb., 1992 | Makino et al.
| |
| 5114222 | May., 1992 | Cornsweet.
| |
| 5125730 | Jun., 1992 | Taylor et al.
| |
| 5129400 | Jul., 1992 | Makino et al.
| |
| 5196872 | Mar., 1993 | Beesmer et al.
| |
| 5202708 | Apr., 1993 | Sasaki et al.
| |
| 5210554 | May., 1993 | Cornsweet et al.
| |
| 5233517 | Aug., 1993 | Jindra.
| |
| 5349419 | Sep., 1994 | Taguchi et al.
| |
| 5371557 | Dec., 1994 | Nanjho et al.
| |
| 5382988 | Jan., 1995 | Nanjo.
| |
| 5410376 | Apr., 1995 | Cornsweet et al.
| |
| 5422690 | Jun., 1995 | Rothberg et al.
| |
| 5504543 | Apr., 1996 | Ueno.
| |
| 5508760 | Apr., 1996 | Kobayashi et al.
| |
| 5539487 | Jul., 1996 | Taguchi et al.
| |
| 5540226 | Jul., 1996 | Thurston et al.
| |
| 5542422 | Aug., 1996 | Hayden.
| |
| 5572266 | Nov., 1996 | Ohtsuka.
| |
| 5662706 | Sep., 1997 | Legerton et al.
| |
| 5697006 | Dec., 1997 | Taguchi et al.
| |
| 5710630 | Jan., 1998 | Essenpreis et al.
| |
| 5713047 | Jan., 1998 | Kohayakawa.
| |
| 5764341 | Jun., 1998 | Fujieda et al.
| |
| 5844658 | Dec., 1998 | Kishida et al.
| |
| 5896198 | Apr., 1999 | Chou et al.
| |
| 5905561 | May., 1999 | Lee et al.
| |
| 5908394 | Jun., 1999 | Kandel et al.
| |
| 5912720 | Jun., 1999 | Berger et al.
| |
| 5914771 | Jun., 1999 | Biber.
| |
| 5943116 | Aug., 1999 | Zeimer.
| |
| 5993001 | Nov., 1999 | Bursell et al.
| |
| 6086205 | Jul., 2000 | Svetliza.
| |
| 6112114 | Aug., 2000 | Dreher.
| |
| 6179421 | Jan., 2001 | Pang.
| |
| 6257722 | Jul., 2001 | Toh.
| |
| 6276799 | Aug., 2001 | Van Saarloos.
| |
| 6304723 | Oct., 2001 | Kohayakawa.
| |
| 6309068 | Oct., 2001 | Kohayakawa.
| |
| 6361167 | Mar., 2002 | Su et al.
| |
| Foreign Patent Documents |
| WO96/17545 | Dec., 1995 | WO.
| |
Primary Examiner: Winakur; Eric F.
Assistant Examiner: Sanders; John R
Attorney, Agent or Firm: Cahill, von Hellens & Glazer P.L.C.
Parent Case Text
CROSS REFERENCE TO RELATED APPLICATIONS
The present application claims priority to the subject matter disclosed in a
provisional application entitled "FUNDUS AUTO IMAGER", filed Jul. 17, 2000 and
assigned Ser. No. 60/218,757 directed to an invention made by the present inventor.
This application is a continuation-in-part of application Ser. No. 09/649,462,
filed Aug. 25, 2000, now U.S. Pat. No. 6,296,358, which claims priority from provisional
Application No. 60/218,757, filed Jul. 17, 2000.
Claims
The invention claimed is:
1. Apparatus for imaging the ocular fundus, said apparatus comprising in combination:
a) a source of illumination for illuminating at least a section of the ocular
fundus of a patient;
b) means for directing the illumination along an optical axis to enter the pupil
of the patient;
c) a video sensor responsive to an image of the pupil reflected from the eye
of the patient for depicting the alignment of the optical axis with the pupil of
the patient;
d) a beam splitter and a lens for directing the reflected image of the ocular
fundus toward said video sensor and for creating a reflected image, respectively;
e) positioning means for aligning the optical axis in response to the depicted alignment;
f) a focusing assembly for focusing the image of the ocular fundus as a function
of the image itself;
g) a further beam splitter for reflecting the image to said focusing assembly;
h) a further video sensor for recording the focused image; and
i) the aerial focal plane of said lens being between said further beam splitter
and said focusing assembly and said focusing assembly including a) a further lens
located at its focal distance from the aerial focal plane, b) an aperture located
at the back focal plane of said further lens, c) a yet further lens for providing
a further image, and d) said further video sensor being located at the back focal
plane of said yet further lens for receiving the image of the ocular fundus section
under examination.
2. The apparatus as set forth in claim 1 including a mounting for supporting
a lens of the patient's glasses intermediate said lens and said aperture.
3. The apparatus as set forth in claim 1 including means for translating said
aperture orthogonally to the optical path of the image between two positions to
obtain sequentially two images forming, in combination, a stereo image of the ocular fundus.
4. Apparatus for imaging the ocular fundus, said apparatus comprising in combination:
a) a source (S) of illumination for illuminating at least a section of the ocular
fundus (
14) of a patient;
b) a first lens (L
1) for directing the illumination through a first aperture (A
1);
c) a filter (F) for controlling the bandwidth of the illumination emanating from
said first aperture (A
1);
d) a mirror (M
1)for finely aligning the optical axis of the illumination,
said mirror including means (
10) for repositioning said mirror in two axes;
e) a lens (L
2) for collecting the illumination reflected from said mirror;
f) a lens (L
3) for receiving the collected illumination and forming an
image of said aperture (A
1) in the pupil of the patient to illuminate a
section (
12) of the ocular fundus (
14) of interest;
g) a first beam splitter (BS
2) for receiving the image of the section
(
12) of the ocular fundus (
14) of interest and reflecting the image
to a second collecting lens (L
4);
h) a lens (L
5) for focusing the image received from said first beam splitter
(BS
1) upon a first video sensor (C
1);
i) a second beam splitter (BS
3) for reflecting a part of the collected
light to a lens (L
6);
j) a second aperture (A
2) for passing the image from said lens (L
6)
to a lens (L
7);
k) a second video sensor (C
2) for receiving the collimated image from
said lens(L
7); and
l) means (
20) for relocating said apparatus in three axis to grossly align
the optical axis of the ocular fundus (
14) illumination with the pupil of
the patient.
5. The apparatus as set forth in claim 4 including a source of light (FIX) and
a third beam splitter (BS
1) disposed in the path of the collimated illumination
from said first collimating lens (L
2) to permit the patient to fixate upon
said source of light resulting in exposure of the section (
12) of the ocular
fundus (
14) of interest to the illumination.
6. The apparatus as set forth in claim 5 including a plurality of said sources
of light (FIX) at spaced apart locations to correspond with different sections
of the fundus being exposed to the illumination.
7. The apparatus as set forth in claim 4 including a plurality of photodetectors
(PD) for providing a signal responsive to iris reflection.
8. The apparatus as set forth in claim 4 including a further source of light
(
16) for illuminating the eye of the patient.
9. The apparatus as set forth in claim 4 including a focusing assembly (FA) translatable
as a unit for focusing the image upon said second video sensor (C
2), said
focusing assembly comprising said lens (L
6), said second aperture (A
2),
said lens (L
7) and said second video sensor (C
2) and means (
18)
for rectilinearly translating said focusing assembly.
10. The apparatus as set forth in claim 4 including a mounting (
29) for
supporting a lens of the patient's pair of glasses adjacent said second aperture (A
2).
11. The apparatus as set forth in claim 4 including means (
24) for translating
said second aperture (A
2) orthogonally to the axis of the image passing
therethrough between a first and a second position to obtain a pair of stereo images
having a stereo base equivalent to the distance between said first and second positions.
12. The apparatus as set forth in claim 4 wherein said source (S) of illumination
provides illumination having a wavelength in the infrared region.
13. The apparatus as set forth in claim 4 wherein said source (S) of illumination
provides illumination having a wavelength in the near visible infrared region.
14. The apparatus as set forth in claim 4 wherein said source (S) of illumination
provides illumination having a wavelength in the visible light region.
Description
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to the field of ocular imaging, and, more particularly,
to devices for imaging the ocular fundus.
2. Description of Related Art
The term ocular fundus refers to the inside back surface of the eye containing
the retina, blood vessels, nerve fibers, and other structures. The appearance of
the fundus is affected by a wide variety of pathologies, both ocular and systemic,
such as glaucoma, macular degeneration, diabetes, and many others. For these reasons,
most routine physical examinations and virtually all ophthalmic examinations include
careful examination of the ocular fundus.
Routine examination of the ocular fundus (hereinafter referred to as fundus)
is performed using an ophthalmoscope, which is a small, hand-held device that shines
light through the patient's pupil to illuminate the fundus. The light reflected
from the patient's fundus enters the examiner's eye, properly focused, so that
the examiner can see the fundus structures.
If a hard copy of the fundus view is desired, a device called a fundus camera
can be used. However, to use existing fundus cameras successfully is a very difficult
undertaking. The operator must (1) position the fundus camera at the correct distance
from the eye, (2) position it precisely in the vertical and horizontal directions
in such a way that the light properly enters the pupil of the patient's eye, (3)
refine the horizontal and vertical adjustments so that the light reflected from
the front surface of the eye, the cornea, does not enter the camera, (4) position
a visual target for the patient to look at so that the desired region of the fundus
will be imaged, and (5) focus the fundus image. All these operations must be performed
on an eye that is often moving. Therefore, the use of existing fundus cameras requires
a significant amount of training and skill; even the most skilled operators often
collect a large number of images of a single eye in order to select one that is
of good quality.
In existing fundus cameras, alignment and focusing are performed under visual
control by the operator. This usually requires that the patient's eye be brightly
illuminated. Such illumination would normally cause the pupils to constrict to
a size too small to obtain good images. Therefore, most existing fundus cameras
require that the patient's pupil be dilated by drugs.
U.S. Pat. No. 4,715,703 describes an invention made by one of the present inventors
and discloses apparatus for analyzing the ocular fundus. The disclosure in this
patent is incorporated herein by reference.
SUMMARY OF THE INVENTION
The present invention is in the nature of a fundus camera which automatically
and quickly performs all the aligning and focusing functions. As a result, any
unskilled person can learn to obtain high quality images after only a few minutes
of training and the entire imaging procedure requires far less time than existing
fundus cameras. Moreover, all of the automatic aligning and focusing procedures
are performed using barely visible infrared illumination. With such illumination,
the patient's pupils do not constrict and for all but patients with unusually small
natural pupils, no artificial dilation is required. The fundus images can be obtained
under infrared illumination and are acceptable for many purposes so that the patient
need not be subjected to the extremely bright flashes required for existing fundus
cameras. To obtain standard color images using the present invention, it is sometimes
necessary to illuminate the eye with flashes of visible light. However, such images
can be obtained in a time appreciably shorter than the reaction time of the pupil,
so that the pupil constriction that results from the visible flash does not interfere
with image collection. Unlike existing fundus cameras, the present invention provides
for automatic selection of arbitrary wavelengths of the illuminating light. This
facility has two significant advantages. First, it is possible to select illuminating
wavelengths that enhance the visibility of certain fundus features. For example,
certain near-infrared wavelengths render the early stages of macular degeneration
more visible than under white illumination. Second, by careful selection of two
or more wavelengths in the near infrared, it is possible to obtain a set of images
which, when properly processed, generate a full color fundus image that appears
very similar to a color image obtained with white light. Thus, it is possible to
obtain acceptable color fundus images without subjecting the patient to bright flashes.
It is therefore a primary object of the present invention to provide a fundus
imager which automatically positions fundus illuminating radiation to enter the
pupil while preventing reflection from the cornea from obscuring the fundus image,
irrespective of movement of the eye or the patient's head within the head restraint.
Another object of the present invention is to provide automatic focusing
of the fundus image based upon the image itself.
Yet another object of the present invention is to provide automatic positioning
of one or a sequence of fixation targets to select the sections(s) of the fundus
to be imaged.
Still another object of the present invention is to provide a fundus imager
for collecting a set of images that can be arranged in a montage to provide a very
wide angle fundus image facilitated by the capability of the fundus imager to automatically
align and focus the images.
A further object of the present invention is to provide automatic setting of
video
levels in a fundus imager to use the full range of levels available.
Yet another object of the present invention is to permit aligning and focusing
a fundus imager under infrared illumination to permit imaging without drug induced
dilation of the pupil.
A yet further object of the present invention is to provide for automatic selection
of illumination wavelength.
A yet further object of the present invention is to provide a colored image from
a fundus imager by sequential imaging and registration of images.
A yet further object of the present invention is to provide an apparently normally
colored image generated by two infrared wavelengths.
A yet further object of the present invention is to provide for automatic acquisition
by a fundus imager of a stereo image pair having a known stereo base.
A yet further object of the present invention is to provide a head positioning
spectacle frame for use with a fundus imager.
A yet further object of the present invention is to accommodate for astigmatism
and/or extreme near and far sightedness by placing a lens of the patient's glasses
in the path of illumination of the fundus imager.
A yet further object of the present invention is to provide a method for automatically
positioning the illuminating radiation of a fundus imager to prevent corneal reflections
from obscuring the fundus image obtained.
A yet further object of the present invention is to provide a method for automatic
focusing in a fundus imager.
These and other objects of the present invention will become apparent to those
skilled in the art as the description thereof proceeds.
BRIEF DESCRIPTION OF THE DRAWINGS
The present invention will be described with greater specificity and clarity
with reference to the following drawings, in which:
FIG. 1 is a schematic diagram illustrating the functional elements of the present invention;
FIG. 2 illustrates the location of the photo detectors relative to lens L3;
FIG. 3 is a block diagram illustrating a representative computer system for
operating the present invention;
FIG. 4 illustrates the effect of corneal reflections to be avoided;
FIG. 5 is a schematic illustrating an alignment of the optical axis to avoid
corneal reflections;
FIG. 6 is a graph illustrating determination of an acceptable video level;
FIG. 7 illustrates determination of edge points;
FIGS. 8A, 8B and 8C depict the light rays from a point to an
image plane without an interposed aperture, and with an interposed aperture at
two locations displaced from one another;
FIGS. 9A and 9B illustrate the shift of an image upon an image plane located
beyond the focal plane in response to displacement of an interposed aperture from
one location to another;
FIGS. 10A and 10B illustrate the shift of an image upon an image plane located
short of the focal plane in response to displacement of an interposed aperture
from one location to another;
FIG. 11 illustrates a head restraint in the form of a pair of spectacles;
FIG. 12 illustrates a first variant for realigning the optical axis;
FIG. 13 illustrates a second variant for realigning the optical axis; and
FIG. 14 illustrates a third variant for realigning the optical axis.
DESCRIPTION OF THE PREFERRED EMBODIMENT
Referring to FIG. 1, there is illustrated a preferred embodiment of optical
system
8 of the present invention. Lens L
1 focuses light from a light
source S onto a small aperture A
1. The light source may be a source of visible
light, infrared radiation or of a wavelength in the near visible infrared region.
Light passing through aperture A
1 passes through a filter F and is reflected
by mirror M
1 toward lens L
2. Mirror M
1 is pivotally mounted
to permit rotation about two orthogonal axes, which pivotal mounting is represented
by device
10 attached to the mirror. Lens L
2 collimates (makes parallel)
light from aperture A
1. A beam splitter BS
1 transmits about ninety
percent (90%) of the incident light from lens L
2 to lens L
3. Half
of the light passing through lens L
3 is reflected by beam splitter BS
2
and is absorbed by light trap LT. The other half of the light passing through lens
L
3 forms an image of aperture A
1 in the focal plane of lens L
3,
which focal plane lies in the plane of a patient's pupil P. The light passing through
the pupil illuminates a section
12 of ocular fundus
14 (hereinafter
only the term fundus will be used).
Light diffusely reflected from fundus
14 emerges from pupil P and half
of it is reflected by beam splitter BS
2 toward collimating lens L
4,
which lens is at its focal distance from the pupil. An infrared light emitting
diode (LED), representatively shown and identified by reference numeral
16,
diffusely illuminates the region of the front of the eye. About ten percent (10%)
of the light is transmitted through beam splitter BS
3, which light passes
through lens L
5. Lens L
5 forms an image of the pupil and the front
of the eye in the plane of a video sensor C
1. The video output from video
sensor C
1 is displayed on an operator's monitor (on computer screen shown
in FIG. 3) to provide a view of the eye and of the pupil.
If the patient's eye is focused at infinity, the light reflected from each point
on fundus
14 will be collimated as it is incident on lens L
4. Therefore,
90% of the light reflected from beam splitter BS
3 will form an aerial image
of the fundus in the focal plane of lens L
4, which focal plane is represented
by a dashed line identified as FI (Fundus Image). The light passes through lens
L
6, which lens is at its focal distance from fundus image FI. Thus, lens
L
6 will collimate light from each point on the fundus. Further, because
the light considered as originating in the plane of pupil P is collimated by lens
L
4, lens L
6 will form an image of the pupil in its back focal plane,
which is coincident with the location of second aperture A
2. Light passing
through second aperture A
2 is incident on lens L
7, which lens will
then form an image of the fundus in its back focal plane which is coincident with
second video sensor C
2. The video image produced by video sensor C
2
represents an image of the fundus.
If the eye is not focused at infinity, the aerial fundus image FI will be moved
away from the back focal plane of lens L
4. For example, if the eye is nearsighted,
the aerial fundus image will move toward lens L
4. Such movement would cause
the fundus image to be defocused on video sensor C
2. Focusing the image
under these conditions is accomplished as follows. Lens L
6, aperture A
2,
lens L
7, and video sensor C
2 are mechanically connected to one another
by a focusing assembly labeled FA; that is, these elements are fixedly positioned
relative to one another and move as a unit upon movement of the focusing assembly.
A unit identified by reference numeral
18 provides rectilinear movement
of the focusing assembly on demand.
A set of photodetectors PD, of which three are shown in FIG. 1, lie in a plane
between the eye and beam splitter BS
2. As further shown in FIG. 2 from the
viewpoint of the eye, orthogonal pairs of photodetectors are located in the vertical
and horizontal axes relative to lens L
3. The purpose of these photodetectors
is that of sensing any light diffusely reflected from the iris.
The entire optical system (
8) discussed above and illustrated in FIG.
1 is supported upon an assembly identified by reference numeral
20. The
assembly includes motive elements, such as rectilinear actuators and related servomechanisms
responsive to commands for translating the entire optical system horizontally (laterally),
vertically and toward and away from the eye, as representatively depicted by set
of arrows
22.
To operate optical system
8, a computer control system
30 is required,
which is representatively illustrated in FIG. 3. The computer control system includes
a central processing unit (CPU)
32, such as a microprocessor, and a number
of units interconnected via a system bus
34. A random access memory (RAM)
36, a read only memory (ROM)
38 are incorporated. An input/output
adapter
40 interconnects peripheral devices, such as a disk storage unit
42. A user interface adapter
44 connects the keyboard
46,
a mouse (or trackball)
48, a speaker
50, a microphone
52,
and/or other user interface devices, such as a touch screen (not shown) with system
bus
34. A communication adapter
54 interconnects the above described
optical system
8 through a communication network
56. A display adapter
58 interconnects a display unit
60, which maybe a video screen, monitor,
or the like. The computer operating system employed maybe any one of presently
commercially available operating systems.
In operation, an operator enters patient information data into the computer control
system using the keyboard and also enters the location or set of locations on the
fundus that is/are to be imaged. It may be noted that the field of view of the
optical system is preferably 30° in diameter while the ocular fundus is about
200° in diameter. To image various regions of the 200° fundus, the eye
can be rotated with respect to the optical system; such rotation is achieved by
having the patient look from one reference point to another. After entry of the
raw data, the patient's head is juxtaposed with a head positioning apparatus to
locate the eye in approximate alignment with respect to the optical axis. An image
of the front of the eye produced by video sensor C
1 appears on computer
screen
60. The operator may use a trackball or mouse
48 or similar
control to move the image horizontally and vertically until the pupil is approximately
centered on a set of cross-hairs displayed on the computer screen. A further control
is used to focus the image of a pupil. Such horizontal and vertical movements,
along with focusing of the image of the pupil, are achieved by moving the entire
optical system
8 through energization of assembly
20 (see FIG. 1).
That is, the horizontal and vertical movements of the image are achieved by moving
the entire optical system horizontally and vertically and the focusing of the pupil
image is accomplished by moving the entire optical system toward or away from the
eye. When the operator is satisfied that the pupil image is in sharp focus and
that the pupil is approximately centered, the operator de-energizes LED
16
(which illuminated the front of the eye) and then initiates the automatic alignment
and image collection procedure.
To achieve proper alignment of the optical system with the eye requires that
the
light from light source S enter the pupil. Initially, the angular position of mirror
M
1 is set so that the image of aperture A
1 lies on the optical axis
of the system. It is noted that the image of aperture A
1 contains the light
used to illuminate the fundus. Since video sensor C
1 also lies on the optical
axis, if the operator has initially centered the pupil image even crudely, light
from light source S will enter the pupil. About three percent (3%) of the light
incident on the eye will be reflected from the corneal surface and if this light
reaches video sensor C
2, it would seriously obscure the image of the fundus.
Therefore, the optical system includes the following elements for preventing corneal
reflection from reaching video sensor C
2.
If the light rays forming the image of aperture A
1 were aligned so that
the central ray were perpendicular to the corneal surface, then many of the rays
in the corneal reflection would pass backward along the incident light paths. As
shown in FIG. 4, the central ray would pass back on itself; the ray labeled Ray-
1
would pass back along the path of the incident ray labeled Ray-
2, etc. (The
angle at which a ray is reflected from a shiny surface can be determined as follows.
First, find the line that is perpendicular to the surface at the point that the
ray hits. Then find the angle between the incident ray and the perpendicular ray;
this is called the "angle of incidence". Finally, the ray will be reflected at
an angle equal to the angle of incidence but on the other side of the perpendicular
line. This is called the angle of reflection.) It is therefore evident from the
schematic shown in FIG. 4 that many rays reflected from the corneal surface and
impinging upon beam splitter BS
2 would enter lens L
4 and impinge
upon video sensor C
2.
However, the corneal surface is steeply curved and if the central ray of
the incident light is moved far enough away from the perpendicular to the cornea,
as shown in FIG. 5, the reflected light will be deflected far enough to miss beam
splitter BS
2 and therefore miss passing through lens L
4 and therefore
not impinge upon video sensor C
2. The method for achieving this deflection
will be described below.
The image of aperture A
1 is appreciably smaller than the smallest pupil
for which optical system
8 will operate correctly. In the preferred environment,
the smallest useful pupil is four millimeters (4 mm) in diameter and the image
aperture A
1 is one millimeter (1 mm) in diameter. Initially, the image of
aperture of A
1 lies on the optical axis and is thus approximately centered
on the pupil. Mirror M
1 is actuated by signals generated by the computer
system to rotate about a vertical axis to cause the image of aperture A
1,
and thus the light that illuminates the fundus, to move horizontally, laterally
in small increments (e.g. 0.1 millimeters), to the left across the pupil. When
the image of aperture A
1 just begins to fall beyond the pupil, that is to
fall upon the iris, the light scattered by the iris will fall on all four photo
detectors PD (see FIG. 2). The photodetectors become enabled to generate a signal
supplied to the computer system indicative of such event. Thereafter, the image
of aperture of A
1 is moved backward one step and a video image of the fundus
is saved. Mirror M
1 is then moved back to the center and stepped in increments
toward the top of the pupil until the photo detectors indicate light reflected
from the iris. Thereafter, the image of aperture A
1 is moved backward one
step and a video image of the fundus is saved. This step is repeated for each of
the right and bottom edges of the pupil.
The computer now contains four fundus images taken with light at four locations
at the edge of the pupil. If the corneal reflection has reached video sensor C
2
in one of those images, the amount of light forming that image will be greater
then the light forming the other images. The computer system examines each of the
four images and selects the one for which the average video level is lowest. This
image is presumed not to contain light from any corneal reflection. It may be noted
that the geometry of the cornea and pupil are such that for a pupil four millimeters
(4 mm) or larger, the corneal (but not iris) reflection will always be absent from
at least one of the four images.
When the image of aperture A
1 falls on the iris, the diffuse reflection
will illuminate all four detectors. Most of the time that the image of aperture
A
1 falls on the pupil and not the iris, the corneal reflection will illuminate
one or two photo detectors. However, the corneal reflection will never fall on
all four detectors. Therefore, to achieve the goal of placing the image of aperture
A
1 into the pupil, it is necessary to determine location of the edge of
the pupil by moving the image of aperture A
1 until all four of the four
detectors simultaneously generate a signal indictive that they are illuminated.
The four edges of the pupil are located as a function of the signals generated
by the photodetectors, as described above. From the location of these edges, the
center of the pupil can be determined by the computer system with respect to the
optical axis of the instrument. If the center of the pupil does not lie approximately
on the optical axis, the computer system commands the horizontal and vertical motors
(assembly
20) to move the entire optical system
8 until the pupil
is centered. The servomechanisms actuating the horizontal and vertical motors are
slow compared to the motions of mirror M
1. These servomechanisms are intended
to permit limiting the motions of mirror M
1 within a restricted range to
reduce the sizes of the entrance and exit pupils of the optical system and to simplify
the optical design of the lenses. In this way, light is continuously and automatically
introduced through the pupil to illuminate the fundus and images contaminated by
light reflected from the cornea surface can be automatically discarded.
An alternative method for tracking the pupil and positioning the image of aperture
A
1 on the pupil of the eye will be described hereafter. In the above described
procedure, an image of the patient's pupil is formed on video sensor C
1.
The image was used by the operator to perform rough alignment of the optical system
with the eye. However, image appearing on video sensor C
1 can also be used
for automatic tracking of the eye and the positioning of the image of aperture
A
1. This is done by using the computer system for extracting the edges of
the pupil from the video signal and computing the coordinates of its center and
of its edges.
When the image of aperture A
1 falls within the pupil, the light it contains
passes through the pupil, falls on the fundus and is scattered by the fundus. Some
of that scattered light exits the pupil. Thus, when the image of aperture A
1
falls within the pupil, the pupil is backlighted by light reflected from the fundus,
and an image of the pupil on video sensor C
1 consists of a bright disk on
a dark background. The goal is to determine the location of the center and of the
edges of this image so that aperture A
1 can be automatically placed where
the fundus will be illuminated and the image of the fundus on video sensor C
2
will then not be spoiled by light reflected from the cornea. If the pupil is correctly
centered on the optical axis of the optical system, the pupil image will be centered.
If not, the direction and distance between the center of the pupil and the center
of the field of view of the camera can be used to drive servomechanisms (assembly
20 in FIG. 1) to correct the error by moving the entire optical system.
Further, if the edges of the pupil are located, those locations can be used to
position the image of aperture A
1 just inside the edge of the pupil. There
are a number of ways of finding the center and the edges of the pupil image from
the video signal produced.
A method for finding the center and the edges of the pupil image will now be
described.
It involves finding the edges of the pupil image on each video line that intersects
the edges and then computing the most likely position of the center and of the
edges of the actual pupil. The image from video sensor C
1 is read out, as
is the standard video practice, by reading the values of the various points along
a horizontal line and then the values along the next horizontal line, etc. (neglecting
the detail of interlacing). If a given video horizontal line intercepts the image
of the pupil, the video level will abruptly rise from the dark background level
to the brighter level of the pupil. To locate this transition and find the position
of each edge, it is necessary to define the values of the background and of the
pupil. To do this, a histogram of pixel values is formed during the first few video
frames. It will contain a large peak with values near zero, representing dark background
pixels, and additional peaks at higher values that represent the pupil and various
reflections to be discussed below. A typical histogram is illustrated in FIG. 6.
Each point along the horizontal axis represents a different video signal level
and each point on the vertical axis indicates the area of the image that displays
the corresponding video level.
The "background level" is defined as the level just below the first minimum.
Specifically, the histogram is first smoothed using a running block filter. That
is, for a position on the horizontal axis the vertical value on the curve is replaced
by the average of the vertical value and its adjoining values. This computation
is performed in steps along the horizontal axis (video level) until there are ten
consecutive values for which the vertical axis increases. The "background value"
is then defined as the lowest of these ten values. An "edge point" on each horizontal
line is defined as the horizontal location for which the video level changes from
equal to or below the "background value" to above that value or changes from above
that value to equal or below that value. As the video scan proceeds, the location
of each point is saved. Thus, at the end of each video frame, a set of point locations
is stored in the computer memory (see FIG. 3).
If the pupil image consists solely of a bright disk on a dark background, the
above described procedure would essentially always be successful in finding a close
approximation to the actual pupil edges. However, for real pupil images the procedure
is confounded by two sources of reflections. First, light reflected from the cornea;
if this light reaches video sensor C
1, it will form a bright spot superimposed
on the pupil image. If that spot were entirely within the margins of the pupil,
it would not interfere with the process described above. However, if it falls on
the edge of the pupil image, as it may when a patient is looking at an angle to
the optical axis of the optical system, then it will appear as a bulge on the edge
of the pupil, as illustrated in FIG. 7. Therefore, some of the "edge points" located
by the above computations will actually be edges of the corneal reflection instead
of the edge of the pupil. Second, a similar problem arises if the image of aperture
A
1 falls on the edge of the pupil, as it might during an eye movement too
fast to be accurately tracked and compensated. In that event, finding the center
and the edges of the pupil requires special procedures.
One such special procedure will described below. The edge points are collected
as described above. There will typically be several hundred such points. An ellipse
is then found (determined) that best fits the set of edge points. The pupil of
the human eye is usually circular, but if it is viewed from an angle, as it will
be if the patient is looking at a point other than on the optical axis, then the
image of the pupil will approximate an ellipse. So long as the reflections from
the cornea and iris do not overlap a major part of the pupil edge (and so long
as the pupil is not of grossly abnormal shape), such a procedure yields a good
estimate of the locations of the actual pupil center and the edge.
One method for finding the best fitting ellipse will be described. Assuming that
200 hundred points have been labeled edge points by the above procedure, each of
such points has a horizontal (x) and a vertical (y) location. Assume that these
200 hundred points, that is pairs of values (x,y), are in a consecutive list. Five
points are selected at random from the list, requiring only that each selected
point be separated from the next selected point by ten or so positions on the list.
This process will then yield the locations of five putative edge points that are
some distances apart on the pupil. These five pairs of values are substituted into
the equation for an ellipse and solved for the five ellipse parameters. One form
of equation for an ellipse is:
c1*x^2+
c2*xy+c3*y^2+
c4*x+c5*y=1
Substitute the five putative edge points as the pairs (x,y) of values
in that equation. Invert the matrix to find the values for c
1 through c
5.
Then the angle that the ellipse makes with the xy axis is:
θ=½*arc cot((
c1-c3)/
c2)
Then if u=x*cos θ+y*sin θ and v=-x*sin θ+y*cos θ, then d
1*u^2+d
3*v^2+d
4*u+d
5*v=1
Where d
1=c
1*cos ^2θ+c
2*cos θ*sin θ+c
3*sin ^2θ
d
3=c
1*sin ^2θ-c
2*cos θ*sin
θ+c
3*cos ^2θ
d
4=c
4*cos θ+c
5*sin θ
d
5=-c
4*sin θ+c
5*cos θ
The center of the ellipse has u coordinate u=-d
4/(s*d
1) and v coordinate
V=-d
5/(2*d
3) so the center of the ellipse has the x coordinate
x=u*cos θ-
v*sin θ
and the y coordinate
y=u*sin θ+
v*cos θ
If R=1+d
4^2/2d
1+d
5^2/2d
3 then the semiaxes of the
ellipse have lengths
Square root (R/d
1) and square root (R/d
3)
This entire procedure is repeated, say, 100 times for 100 different sets of
putative points yielding 100 different estimates of the x,y location of the center.
The best fitting ellipse is the one for which the center is closest to the median
x and y values of the set of 100.
The resulting deviations between the horizontal and the vertical locations of
the center of the chosen ellipse and the optical axis of the optical system can
be used directly as error signals to drive the positioning servos associated with
assembly
20 and the image of aperture A
1 can be directly and finely
positioned such as by moving mirror M
1 so that the image lies just inside
the pupil.
The corneal reflections can be prevented from spoiling the image of the fundus
by the following procedure. The method involves directing the patient's line of
sight to certain selected positions. If the selected position is straight ahead,
that is, the line of sight is directed along the optical axis, then positioning
the image of aperture A
1 in any direction at the margin of the pupil will
cause the corneal reflection to be sufficiently deflected (assuming a pupil of
4 mm diameter or larger). If the selected position is in any other direction, then
positioning the image of aperture A
1 on t
