Patent Application
20060176448
The invention relates generally to instruments for examining and treating the eye and specifically to a scanning laser ophthalmoscope based on a relaxed confocal catadioptric design, in combination with various external light sources, and optimized for advanced non-invasive diagnostic procedures.
The ophthalmoscope is well known as an important device for examining the eye, and in particular the retina. As a result of great interest in preserving eyesight, ophthalmoscopes of various constructions have been built. The latest version of the ophthalmoscope, a scanning laser ophthalmoscope, is particularly appealing because of its unique capability of combining a high resolution and high contrast infra-red or angiographic 60 Hz interlaced video imaging of the retina with psychophysical procedures such as microperimetry. A precise correlation between retinal anatomy and retinal functioning can be established with the scanning laser ophthalmoscope. This retinal function mapping that includes information about fixation characteristics, visual acuity and retinal sensitivity, is now known to be very helpful to the surgeon when applying various forms of therapeutic laser and for the purpose of low vision rehabilitation. Van de Velde has disclosed before in U.S. Pat. Nos. 5,892,569, 5,923,399, 5,943,117, 6,186,628 and 6,789,900, which are all herein incorporated by reference, different embodiments of the relaxed confocal scanning laser ophthalmoscope that use various external light sources for therapeutic and diagnostic applications. Such functional extensions of the basic instrument rely to a great extent on its relaxed confocal and catadioptric design, incorporating one or two synchronized avalanche photodetector pathways.
Additional extended functionality has been disclosed in the U.S. Pat. Nos. 5,543,866, 5,568,208 and 6,000,799, herein incorporated by reference. Such embodiments have been constructed and the results have been reported in the following publications, that are herein incorporated by reference:
It is the principle object of this invention to disclose a number of additional embodiments and optical constructions of the forementioned relaxed confocal instrument that further improve the functionality of microperimetry, wavefront analysis, Stiles-Crawford measurments, therapeutic laser, and especially the integration of the existing instrument with the capabilities of spectral domain optical coherence tomography (SD-OCT).
In summary, the improvements to the generic relaxed confocal catadioptric instrument, optionally combined with external light sources include:
Further objects and advantages of the proposed embodiments will become apparent from a consideration of the drawings and ensuing description of a preferred embodiment.
The reference numerals will be adjusted per preliminary amendment to conform to the specification and drawings
A typical embodiment of the relaxed confocal scanning laser ophthalmoscope adaptable for several functional extensions is illustrated in the figures. The principles of scanning laser ophthalmoscopy are described in detail in the prior art (Pomerantzeff, Webb, Hughes, Saban-Cohen et Al., Plesch and Klingbeil). Only the features of the relaxed confocal scanning laser ophthalmoscope that are relevant to the enhancements are further discussed.
I. Enhanced Acousto-Optic Modulation
The use of an acousto-optic modulator to draw a variety of objects in the visible laser raster dates back to the proposals of Hugh and Webb. Before that, a slide or screen was used at a confocal retinal plane to generate psychophysical stimuli. In its simplest embodiment, a typical collimated Gaussian laser beam of about 1 mm diameter traverses the AOM at the well-known Bragg angle. The 1st order diffracted beam is used with about 85% efficiency. Modulation can occur over a dynamic range of several magnitudes. It was quickly noted that if a two laser system was used for separate imaging and psychophysics, the acousto-optical delay time would have to be taken into account, software or hardware wise. Typically this delay is about 50 pixels of each 90 ns duration, i.e. about 0.5 microseconds. Fortunately, this delay turns out to be a stable value. It was also suggested at that time, but never reduced to practice, that a prism could be directly attached to both ends of this AOM in order to combine two or even more wavelengths, presumbly at their optimal Bragg angle.
Two problems have been observed by the author with regard to the above. First, detailed psychophysics requires a much higher bandwith of modulation in order to correctly and evenly represent the correct light intensity of even single pixel objects. This is particularly clear in the case of the smallest acuity targets (about 1.5 minarc on the retina) that are possible within the laser raster. The second problems relates to the prism. From the Bragg formula it can be seen that shorter wavelengths require a smaller Bragg angle and vice-versa upon entering the AOM. A prism however will cause the shorter wavelength to have a larger Bragg angle and could therefore not be used as previously illustrated. Both problems can actually be solved at the same time with an appropriate optical construction.
In order to decrease the rise-time for modulating the traversing laser beam, it is necessary to focus and collimate the orginal 1 mm beam to a size that is approximately 0.2 mm. In order to accomplish this, the AOM has to be driven at a higher frequency, 80 Mhz at a minimum. To reduce this beam diameter a Gallilean collimator can be used. If this collimating telescopic lens system is used after the prism, then the divergence of wavelengths is switched and corresponds to the correct configuration to enter the AOM. Actual prism values to be used and lens distances to be observed depend on the refractive index of the prism glass, desired Bragg angles and power of the telescope, but are readily calculated from tables provided by Melles-Griot. Also, the prismatic effect itself can be obtained by entering the collimator off-axis on one side and appropriately angulating the second element of the collimator.
At the other end of the AOM, a similar prism and beam expander will realign the different wavelength laserbeams and provide the required beam diameter. In view of what will follow, it is not unusual to obtain a fairly large beam diameter, larger than 1 mm, suitable for truncation at some further point in the illuminating pathway.
It is also understood that the prisms are not necessary if only one wavelength is used for which a short rise time is desired. Rise-times in the order of 30 ns, which are definitely subpixel, are not difficult to obtain and compare favorably with the current rise-times of about 200 ns (beyond two pixels)
If both IR and visible wavelengths are equipped with appropriate AOMs, then a large number of possibilities exist with regard to multiplexing time or area wise areas of interest for various wavelengths. For example, the neighborhood of a therapeutic laser spot could be illuminated with blue or green to better visualize or measure the change in backscatter through the retinal pigment epithelium during (micro)photocoagulation. Many other similar applications can be envisaged. The capability of spatially or temporally modulation over a large dynamic range of several wavelengths and relating this under one CPU—multilayer graphics control card is one of the biggest assets of scanning laser ophthalmoscopy. Currently it allows multiplexing to avoid the need of synchronized detector pathways.
When passing different wavelengths through one AOM is should be noted that the RF efficiency is related to wavelength, even at the correct Bragg angle for each wavelength. The manuals and specification sheets with formulas of Intraaction Inc. AOM models E801D680 Mhz driver and ATM-80A1 are herein incorporated by reference. For refractive index values of different Schott type glasses such as BK-7, and the specifications of dispersing prisms we incorporate the tabular data of Melle-Griot catalogue, copyright 2002.
II. Enhanced Aperture for Separting Light Pathways Inside the SLO
Every ophthalmoscope since its conception needs to separate at one point the illuminating pathway from the pathway that is followed by the returned light. Obviously because the illuminating source and the observer's eye (or detector) cannot be in the same place, but perhaps more pertinent to understand, the need to avoid strong reflexes from mostly the anterior surface of the cornea. This can be accomplished in two ways. A central stop or aperture is used to separate both pathways. Conveniently the aperture's size and location are such that all of the spurious reflexes are transmitted or stopped by the same stop or aperture (essentially returning to the light source itself). This mechanism is found traditionally in ophthalmoscopes and flndus cameras of all sorts. Confocal (scanning) microscopes use the fact that if the detector is confocal with the object, reflexes coming from other layers would arrive at this detector much diluted and would therefore not contribute signifcantly to an undesirable contrast reduction. Although this principle, thus solely using a beamsplitter, alone could be used in a confocal SLO to accomplish this, the relaxed confocal design of the SLO—which means that the confocal aperture is at least four times laser in diameter than the illuminating laser spot on the retina—would render this mechanism inadequate. Only a tighly confocal instrument could therefore use a beamsplitter to this effect.
As a result, the separating aperture in the relaxed confocal SLO is wholly transparent. Under the current invention we propose to provide a series of apertures, e.g. placed on a larger rotating wheel to be able to select one at will, with the following characteristics:
As a minor remark, we would like to draw the attention to the fact that the confocal detecting apertures also come in different diameters, with a central stop, and coatings. It is even possible to create assymetric apertures to emulate indirect lighting from one side. After years of experience, we do believe that a variety of direct confocal apertures are sufficient for diagnostic purposes. In the above paragraphs we proposed to do something similar to the other critical aperture in the SLO optics.
With regard to the illustrations we incorporte by reference the Melles-Griot catalogue optics information on truncation of Gaussian beams, copyright 2002.
III. Enhanced Electro-Optical Control of the Galvanometer-Polygon Scanning Elements
Very early co-pupillary SLO designs by Pomerantzeff and Webb used two mirror-galvanometers in tandem to obtain the desired scanning laser raster. Remarkable at that time was the flexibility of the laser raster dimensions by simply varying the amplitude and frequencies of these galvanometers. Later confocal SLO models adhered closely to the universal NTSC or PAL standards for generating the raster. Also, the speedier polygon was introduced for the fast horizontal scanning. Typically, the polygon would run at about 40,000 RPM and would have either 24 or 25 facets to create the horizontal raster lines. The vertical scanning is typically 50 or 60 Hz interlaced. Interlacing has a long history dating back to the origin of TV. It exists to conserve bandwith and to cope with psychophysical flicker that would otherwise be visible to the observer. Zooming in such an SLO is typically accomplished with the help of a Campini type of telescope-invertor. It conveniently changes the horizontal angle of deflection between two values, and also at the same time adjust the beam diameter of the illuminating lasers. Such constellation is seen in the accompanying
At present, there is a tendency to digitally create and control both video in and out for the SLO, and one result is that video standards have become more flexible with regard to refresh rate and pixel dimensions. The SLO can take advantage of this by electronically adjusting the polygon and galvanometer characteristics, rather than by inserting additional optical elements. This would render the instrument even more compact and economical. As will be explained, the previously mentioned capability of beam truncation is important in this respect. It is quite important to understand that the laser raster characteristics or requirements are different for pure imaging or psychophysical purposes. This will be explained below.
One known formula is basic to our discussion: it describes the relation between the number of image lines (raster lines), including any number of “virtual” lines drawn during the vertical retrace, the number of facets on the polygon, the polygon speed and the refresh rate of the raster that is determined by the galvanometer. The formula is thus: Refresh Frequency X # of lines=# of facets X RPM of polygon. This is applicable to both interlaced or non-interlaced schemes. Note that the angular amplitude of the galvanometer and the actual beam diameter of the flying spot on the retina do not appear in this formula. Those last two parameters are to be independently controlled. By varying the above mentioned parameters the following results can be obtained:
Typically for imaging purposes, the spot diameter on the retina will not change. Although this may seem rather strange, it should be observed that the light being collected is more determined from the size of the confocal aperture than from the initial impact zone on the retina. This means that basically the same image area on the retina will be visualized, for example a 40 degree diagonal FOV, regardless of zoom, but the magnification factor changes, i.e. the number of horizontal and vertical pixels. The price that is paid for a lower resolution is “subsampling” of the same area on the retina, the bonus we get is increased refresh rate. This increase in refresh rate (or decrease) has some implications in the presence of eye movements and has a very important impact on psychophysical applications because of the human eye's sensitivity to flicker under certain conditions. Three types of fixational eye movements exist that merrit our attention: a very fast 60 Hz, small amplitude oscillation that is generally below the resolution of a regular SLO (not adaptive optics SLO). A second movement is a slow drifting of typically 1 degree per second. In general this will only cause a slight distortion of the consecutive images even at low refresh rates. For the purpose of fiducial landmark tracking, these movements will not affect close TV lines but might lines that are spaced at the edges of the screen. A third type of eye movement are the “corrective” micro saccades” or flicks. They are a rapid restoration of the line of sight onto the object of fixation. Such flicks will either case a clearly visible abrupt shift in the image, visible as a break or discontinuation of the vessel pattern upon freezing the image or they will cause the same fiducial landmarks to be appreciably separated from each other in consecutive images. During active tracking using e.g. two-dimensional normalized gray-scale correlation these flicks can be distinguished from the slower drifting based on aforementioned characteristics. Because of this and because of the typical distortion of the laser raster, fiducial landmarks for tracking should be chosen as close as possible to the area of interest to be tracked.
Adjustment of the angular vertical amplitude of the laser raster with the galvanometer is useful in other situations: e.g. to calibrate square pixels and if the lines are made to overlap, a higher resolution can be obtained when confocally detecting the position of the laser spots projected by the previously mentioned external wavefront measurement module.
IV. Interfacing with External Light Sources for Various Purposes, Requiring Spatial Light Modulation
Some embodiments have been described in the referenced patents. For example, the use of a wavefront measurement module and selective PDT or TTT laser treatment to the retina. Both applications are different in some respects: PDT or TTT requires a fairly large amount of power to be delivered to the retina, hence the use of a higher powered multimode laser and a simple mode scrambling device. The spatial pattern that is desired is most often stationary and requires only some adjustment in positioning if the retina were to move. The laser light is delivered in Maxwellian view mode.
The wavefront module essentially requires parallel pencils of light to cover systematically the pupillary area. Because those (thick) rays are parallel they are supposed to focus on the same spot on the retina in the absence of refraction anomalies. If a low power laser source is used to generate the pencils, the laser could first be focused to a very small spot onto a thin diffuser. On the other side of the diffuser a pinhole is attached from which the light will fan out, its coherence being broken. Divergence can be controlled if necessary with a negative lens. The N.A. can be fixed with a subsequent large aperture (about 10 mm) and collimation will prepare the bundle of light to hit the LC-SLM. Several adjacent pixels would let an approximately 1 mm beam travel towards the pupil. It is possible to use phase retardation LC-SLM for nulling procedures. In this case the orientation of the ray would be slightly adjustment to create a spot on the retina that is coincident with a reference spot.
Using the Mazwellian view projection of the TTT method and the LC-SLM described above, it is straightforward to project microperitric stimuli onto to the retina. Additional modules comprising two lenses, can be inserted in the optical pathway to take care of magnification or focus issues (these extra modules have been omitted from the drawings). For further reference with include by reference a paper presented at the International Conference on Lasers '96, Society of Optical and Quantum Electronics, 2-6 December 1996, entitled “Applications of programmable spatial light modulators”, and selected references, by author Michael K. Giles.
V. Interfacing with Fast Scanning External Light Sources for Diagnostic or Therapeutic Purposes
The electro-optical characteristics of the relaxed confocal scanning laser ophthalmoscope can be used to great benefit when combining this instrument with external laser sources, in particular scanning laser sources. Several examples of increasing complexity have been outined in the incorporated patents by reference. We would like to give an overview and describe an additional possibility below.
For therapeutic purposes there has been an increasing interest in targetting selectively the retinal pigment epithelium (Although we could argue that preferably the photoreceptor outer segments should be disabled in a certain pattern or holes created within an obstructive and thickened Bruch's membrane). The technique has been referred to as microphotocoagulation. In brief, scanning is performed with the help of a two dimensional or one-dimensional acousto-optic deflector. The details have been outlined in U.S. Pat. No. 6,789,900.
There is also another important diagnostic application of scanning external light sources based on the integration of spectral domain optical coherence tomograhy (SD-OCT) into the SLO based on some general principles outlined before and taking into account some particular observations with regard to the use of SD-OCT itself.
In brief, SD-OCT was conceived by Fercher in 1995 as an alternative to the classic time domain OCT. This time domain OCT requires a lay-out that is similar to a Michelson interferometer to record reflections within the retina (Thus exploring the retina in depth or z-direction; the SLO typical explores the x-y plane and sort of integrates the information from the z direction). The time domain OCT requires a reference arm that rapidly scans back and forth. This is fairly time consuming. Using broadband light sources, this interference method can obtain a resolution in the z-direction of as little as 5 micron. This compares very favorably with the limited z-resolution of a confocal SLO which is at best about 150 micron. Thus both diagnostic techniques are very complimentary and their combination would give rise to an instrument in which the ophthalmologist has an immediate overview of the retina and where he would be able to obtain a section of histologic quality (much like a slitlamp would do) between any two points of interest.
Up to now such combinations exist, combining a variation of a classic findus camera with time domain OCT. However several disadvantages exist: the time domain is slow, about 400 samples can be taken per secon, several thousand are needed to collect a reasonable amount of information. In this time interval, eye movements occur and also, the fundus visualization is either non-confocal IR which does not permit viewing details on the retina, or it is a color reference picture on which it is difficult to outline precisely where the OCT samples have been taken.
Some of these problems have been resolved by construction an SLO-OCT combination (Podoleanu) that uses the x-y scanning lasers to obtain the interference data as it scans along a raster pattern on the retina (C-plane scanning). This however does not resolve some of the issues regarding speed, S/N ration and immunity against eye movements, since the technique is still essentially within the time domain.
Our SLO-OCT combination is based on the spectral domain method and exploits the unique capabilities to combine scanning laser sources as described in referenced patents. Only recently has SD-OCT become mainstream (in combination with a classic slitlamp or fundus camera) because of the intense calculations (Fourier transforms) that have to be performed. Curent computational devices are upt to this task and it is not unusual to be able to derive 40,000 samples per second. It is precisely this high speed acquisition of information that makes integration within the SLO possible. Besides the general issues that have been addressed in previous patents and for which the solutions are here applicable by comparision, two special considerations have to be made:
The two systems (SD-OCT and relaxed confocal SLO) can be combined optically using similar constructions as described in previous patents. Ideally, the scanning of the OCT component happens to originate from a pivot point and through the similar pivot point of the SLO scanning lasers. The module can be completely external, or it can still make use of the main reflecting mirror inside the SLO. In the later case, the pivoting mechanism is smaller and consist of a mirror that is allowed to move in x-y directions on top of a ball bearing. Focussing however requires at least one more lens element.
At this point it is useful to remind that fast IR modulation (TTL or AOM based) can have an additional application in video blanking for microperimetry purposes: when measuring absolute thresholds on the retina; even “invisible” IR light of 800 nm still retains the capability of stimulation the retina: the IR laser could then be turned of in the neighborhood of the stimulus. Also multiplexing in the case of combined ICG and fluorescein angiography, and multiplexing in general when only one detector pathway instead of two are available, are important applications of fast IR modulation.
As a last remark we point to the development of polarization sensitive OCT techniques: these would have the capability of analysing the state of polarization, hence birefringence of the nerve fiber layer in the retina based on an absolute high resolution space interval (typically 3 micron). This cannot be realized with the current polarization sensitive SLOs that suffer from the interference of corneal birefringence, measure the total birefringence in toto of the nerve fiber layer. Such technique cannot readily distinguish between true nerve fiber layer thinning and molecular changes (microtubules) within the nerve fiber axons that eg. limit axonal transport in the early stages of neuro-degenerative diseases such as glaucoma and Alzheimer's.
For further information on SD-OCT we include by reference the following U.S. Pat. No. 6,3777,349 by Adolf Fercher, and the following publications: (1) Measurement of intraocular distances by backscattering spectral interferometry, optics communications, by A. Fercher et Al, 1995; (2) Optical Coherence Tomography Journal of biomedical optics, vol 1(2) pp 157-173 1995 by A. Fercher; (3) Optical coherence with spectral radar by M. Bail et Al. Photon Propagation in Tissues II Proc. SPIE vol 2925 page 298-303, 1996.
Additional embodiments of the relaxed confocal scanning laser ophthalmoscope have been disclosed that relate to:
Although the descriptions above contain many specifications, these should not be construed as limiting the scope of the invention but as merely providing an illustration of the presently preferred embodiments. Other embodiments including additions, subtractions, deletions, or modifications will be obvious to those skilled in the art and are within the scope of the following claims. The scope of the disclosure should be determined by the appended claims and their legal equivalents, rather than by the examples given.
This application is acontinuation of U.S. Pat. Nos. 6,186,628 and 6,789,900. These patents are incorporated in this application by reference.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 09431680 | Nov 1999 | US |
| Child | 11182538 | Jul 2005 | US |
| Parent | 10291958 | Nov 2002 | US |
| Child | 11182538 | Jul 2005 | US |
