TITLE
FIELD SYNTHESIS AND OPTICAL SUBSECTIONING FOR STANDING WAVE MICROSCOPY
Background of the Invention
The invention relates to luminescence optical microscopy and particularly to an apparatus and method for selectively optically illuminating particular zones in a specimen such as a biological cell or tissue.
Field of Invention
Optical microscopy using fluorescence tagging for the determination of three-dimensional structure of cells and tissues is an important diagnostic and research procedure. There are a great number of dyes which can be attached to various structures within the cell. When excited by a particular wave length of light these dyes will fluoreεce or phosphoresce. For example, several common dyes will emit a red glow when excited by green light. Hence, one can see structures to which a fluorescent dye is attached. The presence and location of the tagged structures can provide important diagnostic and structural information for basic research and clinical diagnostics.
Fluorescent imaging, in particular, is of vast utility in cell biology because of the high specificity of fluorescently labeled protein analogs, antibodies,
hybridization probes, enzyme substrates, lipid analogs, and peptideε, as well as stains. Fluorescence micrographs of extremely complicated objects such as intact cells typically show clearly the distribution of the tracer molecules, all other components being "invisible". The important optical characteristics of many biological specimens that allow for this simple interpretation of the image field is that cells are generally weakly refractive and weakly absorptive objects .
Conventional microscope images contain information about the 3-D structure of the object when the depth of field of the lens system is smaller than the axial dimension of the specimen. This means that in a single image, the axial location of a particular feature is encoded by its degree of defocus. A 3-D image data set, which is a "zero-order" estimate of the true structure of the object, is obtained by recording a series of images as the object is stepped through the focal plane of the microscope, a procedure known as optical sectioning microscopy (OSM) . Each image is a spatially filtered axial projection of the object, and each generally contains in-focus and out-of-focus features. One of the central problems in 3-D microscopy is the removal of out-of-focus features from the 3-D image by optical and image processing methods thereby deriving a refined estimate of the true object.
In fluorescence microscopy there is a linear relationship between the emitter distribution in the object and the intensity distribution in the image field. This is caused by the mutual incoherence of fluorescence emission. Dye molecules in the specimen radiate independently so that the individual intensity fields are simply superimposed in the image plane. It is possible to deconvolve the 3-D image to eliminate the out-of-focus portions. However, for the various methods that have been proposed there is a trade-off of recovery of high frequency structure for accuracy or stability.
The alternative to computational refinement of optical sectioning microscopy (OSM) is confocal scanning fluorescence microscopy (CSF ) in which direct optical spatial filtering is used to remove out-of-focus light waves from the detector field. In one version of this type of instrument fluorescence is excited in the specimen by a highly focused beam. In the image plane of the microscope a pin hole is placed at the point optically conjugate to the focal point of the beam and a high gain, low noise detector is placed behind the pin hole. The microscope acts as a spatial filter that detects efficiently only fluorescence photons that originate near the beam focus . 3-D image data is obtained by a raster scanning of the beam relative to the specimen, either optically or mechanically, and stepping
the specimen axially through the focal plane to get stacked images. Confocal methods have several shortcomings. For example, such images often have a low signal to noise ratio. Hence, the resolution of the image is often severely compromised. Also, scanning usually is relatively slow, with scan times up to 64 seconds per frame for high signal-to-noiεe images. Indeed, there are many circumstances in which this technique cannot be utilized.
For fundamental reasons, a fluorescence microscope is more severely limited in axial (depth, or inter-image plane) resolution, as opposed to transverse (in the image plane) resolution. Consider a microscope with a lens having a high numerical aperture (NA) and a specimen of refractive index n illuminated by a light beam having a wavelength λ. The well-known Rayleigh resolution formula, 0.61λ/(NA), sets transverse resolution at about 0.2 μm via direct imaging. This can be halved, in principle, by confocal scanning. In comparison, the axial equivalent of the Rayleigh formula,
2 2nλ/(NA) , is in the range 0.7-0.9 μm, typical for high- quality fluorescence OSM image sets. Computational image
processing or confocal scanning can reduce this to
0.4-0.5 μm . A more restrictive analysis, the Rayleigh quarter-wave criterion, λ/8n sin 2 (1/2 si•n-1 NA/n), gives a theoretical axial resolution in the range 0.13-0.17 μm for the best microscope lenses. This has been demonstrated in transmitted light microscopy, but not in fluorescence, due to the lack of mutual coherence in fluorescence imaging, and the generally lower signal-to- noise level. Therefore when the specimen contains fine stratified structural features, or simply when it is thinner than the depth-of-field, fluorescence OSM or even CSFM is unable to yield significant 3D information.
United States Patent No. 4,621,911 discloses a method and apparatus called standing wave luminescence or fluorescence microscopy (SWFM) in which a specimen is illuminated in a fluorescence microscope by means of a standing wave field at the excitation wave length. This field is preferably produced by crossing two equal amplitude coherent beams from a laser. The direction of the beams is such that the nodal and anti-nodal planes in this field are parallel to the object plane of the microscope. Under this condition fluorescence is excited in laminar zones in the specimen, maximally at the location of each anti-nodal plane. One of these planes
can be made coincident with the in-focus plane. In this way in-focus features of the specimen are made brightly fluorescent. Immediately adjacent out-of-focus features above and below the in focus plane are in nodal zones and are, therefore, only weakly fluorescent.
United States Patent No. 4,621,911 teaches a theory and embodiments for creation and manipulation of a periodic standing wave field superimposed with the specimen in a fluorescence microscope, and that sets of images obtained by standing wave excitation contain Fourier coefficient information on the axial (depth) structure of the object down to an axial resolution limit of λ/4n, as small as 0.068 μm. The embodiments of the patent include several methods for producing a standing wave field by crossing two equal-amplitude collimated s- polarized coherent beams at complementary angles relative to the axis of the microscope. These embodiments include the use of total internal reflection (TIR), a mirror or prism, or a wavelength-selective high reflector to fold a laser beam in the specimen region of the microscope, independent coherent beams entering the specimen from opposite sides, or a re-entrant beam that first emerges from the objective lens into the specimen. The patent also includes embodiments where the nodal planes are not parallel to the object focal plane although the parallel
condition is of principal interest here. Nevertheless, the method and apparatus of the '911 patent do not overcome the problems associated with the presence of out of focus luminescent portions in the image.
Summary of the Invention
We provide improved methods and apparatus which overcome the out of focus problems and clearly display the transverse and axial position of luminescent structures in the specimen. As in the apparatus of Patent 4,621,911, a fluorescence microscope is equipped with an optical system for standing wave excitation of the specimen. In a first present preferred method we manipulate a single standing wave field to show the axial structure of a specimen, even when the specimen is so thin that it is entirely within the depth-of-field of a high resolution microscope. We call this extension optical εubsectioning, and have found a practical axial resolution limit of λ/8n (approximately 0.05 μm ) in thin specimens. In a second present preferred embodiment a nonperiodic excitation field is generated in the specimen such that the excitation intensity is peaked only at the object focal plane. In this case, intersecting beams entering the specimen from opposite sides are swept in angle to generate a continuous series of standing wave fields that differ in node spacing, but all with an
antinode at the object focal plane. A single image is recorded with the swept excitation, with the net result of the time-multiplexing being the preferential excitation of structures in the specimen that lie in the in-focus plane. This differs from the method of the '911 patent in that the net excitation field is peaked only at the in-focus plane, and not at evenly-spaced antinodal planes. We call this method excitation field synthesis (EFS) or field synthesis fluorescence microscopy (FSFM). It represents an extension of standing wave microscopy theory and practice to the high-aperture limit, where the depth-of-field of the optics is less than the axial dimension of the specimen.
In our method and apparatus we direct two beams, which can be a single beam reflected back through a specimen, to create distinct nodes and anti-nodes such as is taught in U.S. Patent No. 4,621,911. We prefer to direct the beams through the lens of the microscope. In one embodiment the beam reflects directly back from a mirror positioned underneath the specimen. Alternatively, we use an active phase conjugator to generate the return or re-entrant beam in a standing wave illuminator. The phase conjugator generates the time- reversed version of the excitation beam that first passes through the specimen, so that a standing wave field of
high spatial modulation is created, even in specimens where the incoming planar wavefronts have been aberrated due to refractive index inhomogenetieε . In another embodiment we use a beam splitter to create two beams from a single light source. The two beams preferably are directed to enter the specimen from opposite sides.
In yet another embodiment the specimen is illuminated in a manner previously discussed and an image of the specimen is recorded in a camera and stored. Then the specimen is moved a selected axial distance. Again the specimen is illuminated and a second image is created and stored. This process can be continued until a desired number of images of the specimen are created. These images are then combined by image procceεsing to produce one or more combined images of the specimen.
Other objects and advantages of the invention will become apparent from a description of the preferred embodiments shown in the figures.
Brief Description of the Figures
Figure 1 is a schematic representation of the standing wave illumination of the prior art wherein a specimen is in a standing wave field of ε-polarized light and on the optical axis of a microscope.
Figure 2 is a schematic representation illustrating the formation of a standing wave field by
total internal reflection of an incident beam from a cover glass.
Figure 3 is a side view of a specimen placed in the object focal plane of a microscope.
Figure 4 is a diagram showing a first present preferred embodiment of our improved standing wave microscope.
Figure 5 is an enlarged view of the specimen being illuminated in the microscope of Figure 4.
Figure 6 is a block diagram of a image processor used with the microscope of Figure 4.
Figure 7 is a schematic diagram of the second preferred embodiment of our microscope.
Figure 8 is a graph of overlapping light beams having different nodal spacings.
Figure 9 is a graph illustrating movement of the nodal plane for optical subsectioning.
Principles of Operation
In a standing-wave microscope, two plane-wave fields from a laser are crossed at complementary angles in the specimen volume, where they interfere (Fig. 1). When the two fields are s-polarized and of equal amplitude, the resulting interference pattern has an electric field intensity that varies only axially, as
Ieχ(ez)/ = I0 fi i cos (vKz + Φ) 1j where K= ( 47rn/λ )cos£, λ is the wavelength and n is the specimen refractive index. Fluorescence is excited in the specimen in proportion to I (z). The nodes or antinodeε, which are planes parallel to the focal plane, have a spacing Δs = λ/2n cos# . By controlling the angle θ , the node spacing can be varied down to a minimum value of λ/2n, when the two beams are counterpropagating along the axis of the microscope. By shifting the phase of one of the beams, the relative position of the antinodeε within the specimen can be changed, at constant node spacing.
It is straightforward to estimate the axial resolution in SWFM. Two particles which are separated axially by half the node spacing can be illuminated alternately by shifting the phase of the standing wave pattern. With blue light excitation and a specimen refractive index of 1.33 (water) or higher, λ/4n is 0.09 μm or less. In practice, it has been possible to resolve
particles separated axially by a quarter fringe (0.045 μm ) in specimens where there is little overlapping structure. Because these distances are considerably less than the depth-of-field normally obtained in fluorescence microscopy, "optical εubεectioning" is possible. That is, in thin specimens that fall entirely within the depth-of-field of a high numerical aperture (NA) objective lens, axial structure can be observed purely by calibrated movement of an antinodal or nodal plane through the specimen, with no mechanical refocusing required. Even when the specimen is thick, discrete outlying structure may be discriminated by degree of defocus from in-focus features, so that optical subsectioning will still be useful. In mathematical terms, standing wave excitation is equivalent to axial modulation of the point spread function (PSF) which in turn is equivalent to shifting the optical transfer function (OTF) axially in reciprocal space by a distance equal to the spatial frequency of the standing wave field. This permits recovery of spatial frequency information that is absent or very heavily attenuated under incoherent excitation.
Optical subsectioning is a εubtractive process in which a single standing wave field is manipulated to resolve the relative axial position of two or more stratified structures in a thin fluorescently labeled
specimen. A thin specimen is one which lies entirely within the depth-of-field of the microscope and, strictly, within one node period of the standing wave field, 0.17 μm under typical conditions. In practice, the phase of the field is adjusted so that a nodal plane is coincident with one stratum. The structural features of that stratum then fluoresce only weakly, and an image is recorded that shows primarily the structures in other strata, i.e., a subtractive image. The nodal plane is then moved by a known distance to null the fluorescence from a neighboring stratum, and a complementary image is then recorded. In the simplest case, two structures that overlap in a conventional fluorescence image will each appear distinctly in two standing wave images, where a nodal plane is located first at one structure, and then the other.
A standing-wave microscope is, in fact, a type of inter erometer in which the dye molecules in the specimen act as the primary detectors of the excitation field. In our original design (Fig. 2), total internal reflection was utilized to fold a collimated laser beam at the specimen cover glass to form the periodic field, which made the use of a high numerical aperture (NA) immersion lens difficult. This problem iε solved by use of low-divergence gauεsian beams propagating within the aperture of a high-NA lens, giving maximum resolution and
light collection efficiency. In the simplest configuration (Figs 4 and 5), the beam 21 emerges from the lens 8, passes through the specimen 2, and is back- reflected by a cloεely-apposed mirror 16. One could use a phase conjugator in place of a mirror 16. The mirror iε moved axially by a piezoelectric drive 14, which causes an equal axial shift of the excitation field planes through the specimen. The optics are adjusted so that the gaussian beam exiting the objective 8 contracts slightly to a large-diameter waist (150 μm) at the mirror, 0.1-0.2 mm beyond the specimen. The standing wave field iε then due to the superpoεition of the gaussian field and its reflection. In this condition, the nodal surfaces of the unperturbed standing wave field, although curved in principle, are flat to better than 1 part 20,000 over the field of view. Therefore, these surfaces are called nodal planes. Nodal planes are always parallel to the mirror in this system.
A more versatile optical system having a laser 64 for a light source was also designed, with objective lenses 8 and 55 positioned on opposite sides of the specimen (Fig. 7). A prism splitter 56 iε used to amplitude-divide the expanded beam, so that a low divergence gaussian enters the εpeci en 2 independently from each εide. Since the coherence length of a 1 m ion laser is only about 30 mm, the two beam paths in the
microscope are typically matched to within 5 mm. A piezoelectric drive 7*1 on a mirror 72, 73, 74 in one beam path serves to adjust the phase, in this case mirror movement being a full wavelength per cycle of the standing wave field. One advantage of the two beam system iε that unaberrated wavefrontε enter the specimen on both sides, compared to the mirror system where phase errors accumulate on both passes when the specimen refractive index is heterogeneous. The second advantage of this configuration iε that by sweeping the beam angle ( θ ) in the specimen, standing wave fields of different spatial period can be time-multiplexed in the εpecimen during the acquisition of a single image. The advantage of this is described below.
The two-beam εyεtem provideε the means for excitation field synthesis (EFS). The microscope operates like an OSM εyεtem, in that fluoreεcence images of the εpecimen are recorded in an electronic camera as the specimen iε stepped through the focal plane. It differs from OSM in that for each image, the εpecimen will be excited by a continuoυε sequence of standing wave fields that differ in axial node spacing, but all with an antinode located at the focal plane of the microscope. The sequence of fields iε generated by εweeping the beam croεεing angle during excitation of the specimen. For every field in the sequence, fluoreεcence will be excited
maximally at the focal plane. Away from this plane, the phases of each field differ, so that the excitation intensity averaged over all fields in the sequence will be less. Optionally, a second fluorescence image can then be recorded with the same εequence- of standing wave fields, except that each is adjusted to have a node at the focal plane. In this case, fluorescence will be ' excited minimally at the focal plane, but at similar averaged levels away from it. Digital pixel-by-pixel subtraction of the nodal image from the antinodal image gives a result for which the effective excitation field iε peaked at the geometric focal plane, but decays to zero above and below this plane.
In practice, the εequence of standing wave fields can be generated by sweeping the beam-crossing angle θ through the full range accessible within the aperture of the objective lens and condenser. For a water-immersion system having a numerical aperture of 1.2, for example, the range is +64°. The synthesized field for the antinodal image will then be
z cos dθ
IEFS<
Z>
θ
m ( l + Jo(K
0z)) + έ (-l)
π_1 Sin 2n °m 2
0n„(K
0z n n=l
= 1.12 (1 + J„(K<>Z)) t 0.78 J (K0Z) + 0.49 J4(K0z) + 0.15 J6(KCZ) - 0.10 J8(K0Z) - 0.19 J10(KoZ) - 0.13 J12(K0z) - 0.01 J14(K0z) + ... where Kc = 47rn/λ. For the nodal image, only the constant term doeε not change sign, so that the difference image iε weighted by the sum of the Bessel terms. This εum iε peaked at the focal plane, and decays to zero in an oscillatory manner above and below it (Fig. 8). For an idealized optical εyεtem in which the numerical aperture iε equal to the εpecimen refractive index (NA = n, θm =
90° ) , the weighting of the difference image iε simply
Jo (KoZ) .
The effect of field synthesis on the OTF can be illustrated directly. Fourier transformation of the above equation gives
< Ko '
which is a piecewise discontinuous function of axial spatial frequency. J
vτ, (k ) is convolved with the OSM
OTF to give the EFS OTF. Therefore, it can be seen that for an EFS system operating within the aperture of existing high-NA objective lenses, the OSM OTF gets expanded piecewise into an extended axial εpatial frequency band. The EFS OTF haε the same transverse limit as in OSM, but iε εignificantly extended axially. Purely axial wave vectors with frequencies in the range of K
0cos0 to K
0 are recovered directly, and the bandpaεε limit extends beyond this for near-axial wave vectorε.
Description of the Preferred Embodiments
The present invention utilizes intersecting beams to create a standing wave pattern. Several methods for creating such a pattern are described in United States Patent No. 4,621,911. Figure 1, which was taken from that patent, shows the intersection at a fluorescent or phosphorescent specimen of two coherent, collimated, monochromatic beams of light, A and B, of a wavelength suitable for excitation of fluorescence or phosphoreεcence in the εpecimen. Rays A represent the propagation of a collimated beam (plane waves) that makes an angle θ with the optical axis of the microεcope. Rayε B repreεent the propagation of a εecond collimated beam that makes an angle 180° - θ with the optical axis, and iε coplanar with A and the optical axiε.
It iε alεo poεεible to create a standing wave pattern using reflection so that the reflected beam intersects the incident beam. This technique is illustrated in Figure 2. A sufficiently coherent light source (not shown) directs a collimated beam C which passes through the εpecimen 2. That beam εtrikeε the cover glass 4 and is reflected as collimated beam D. Since beams C and D are s-polarized and intersect, a standing wave pattern 6, indicated by dotted lines, is formed. The microscope lens 8 is positioned so that itε focal point 9 in focal plane 10 is within the standing wave pattern in the specimen.
When both beams A and B or C and D make the same angle θ with respect to the optical axis of the microscope, as εhown in Figures 1 and 2, the antinodal and nodal planes of the standing waves are parallel to the focal plane. Therefore, fluorescence will be excited in the εpecimen in laminar zoneε that εhow the axial εtructure of the object. This can be most clearly seen in Figure 3.
A εide view of the εpecimen 2 mounted on a glaεε εlide 6 iε εhown in Figure 3. The specimen 2 is under a cover glass 4. The microscope lens 8 iε positioned so that the object focal plane 10 is within the εpecimen 2. If the εpecimen iε illuminated in the manner εhown in Figureε 1 or 2 a εeries of laminar zoneε 12 will be
created within the specimen. The node spacing (ΔS) of the excitation field varies with changes in the wavelength (λ) of the beam and its angle ( θ ) relative to the optical axis 3. That iε, (ΔS) = λ/2n cos θ which is a minimum of λ/2n at θ = 0° . The relative position of the nodes and the εpecimen can be varied at constant node spacing by shifting the phase of one of the beams. In practice, total internal reflection was utilized to fold one beam at the cover glass 4 so as to set up a standing wave as εhown in Figure 2. In the method of Figure 5, θ is limited by J the lens sy Jstem to a maximum value of θ m = sin" NA/n' . In the εyεtem of Figure 2, θ iε limited by a minimum value egual to the critical angle at the cover glaεε 4.
Many cells can be tagged with a dye that is excited by green light to fluoresce red; so we can use a green laser for our light source. For other dyes we may also use blue, red, yellow and even ultraviolet lasers. As shown in Figure 4 it is also possible to use an incoherent light source 20 εuch aε a high preεsure mercury lamp with a beam collimator 19, polarizer 22 and bandpasε excitation filter 23. For the microscope 18 εhown in Figure 4 the imaging system is comprised of
lenεeε 8 and 28, dichroic reflector 26, emiεεion filter 27, beam εplitter 25, eyepiece 24, camera 30, image processor 32 and display 34. Light source 20 emits a light beam 21 which iε expanded s-polarized and collimated. Then the beam iε reflected by dichroic reflector 26. Green light passeε through lenεeε 28 and 8, cover glass 4 and specimen 2 until it iε reflected by mirror or phase conjugator 16. This causes a standing wave pattern shown in Figure 5 to be created in the specimen. The incident and reflected green light beams also cause luminescent tags within the specimen to emit red light 29. Emitted red light paεseε from the specimen through lenses 8 and 28 and through dichroic filter 26 and emission filter 27 and is directed by beam splitter or reflector 25 to eyepiece 24 or camera 30 or both. An electronic camera 30 with image procesεor 32 and display 34 records and displays an image. The display could be a cathode ray tube or film.
Referring to Figure 5, the incident beam 21 of green light passeε through the cover glaεε and specimen and is reflected by reflective surface 17 of mirror 16 as reflected beam 31. The interεecting incident beam 21 and reflected beam 31 create a εtanding wave pattern within the εpecimen. Furthermore, thiε light cauεeε luminescent tag 36 within the specimen 2 to emit red light indicated by beam 26. The red light 26 iε then directed to the
camera image processing unit and display as shown in Figure 4. The optics are adjusted so that a gaussian beam 21 exiting the objective 8 contracts slightly to a large-diameter waist (150 m) at the reflective surface 17 of the mirror 16 which is preferably 0.1-0.2 mm beyond the specimen 2. The standing wave field is then created by the εuperpoεition of the gaussian field and its reflection. In thiε condition, the nodal surfaces of he unperturbed standing wave field, although curved in principle, are flat to better than 1 part in 20,000 over the field of view.
The image which iε created from the emiεsion of light 29 by the luminescent tag 36 can be enhanced using known image proceεεing technology. That image can be further improved uεing the methods described herein. Therefore, the image processor 32 should include a central processing unit 37, read only memory 38, and a random accesε memory 39 aε indicated by Figure 6. Normally, the image will be digitized uεing an A/D converter 40 in the image proceεsor 32 or in the camera 30. The digital image is enhanced by the central processing unit 37 according to a program in memory 38. Both the original and enhanced images can be stored in memory 39. The enhanced image is converted to analog form by D/A converter 41 for display.
A more versatile optical system with objective lenses positioned on opposite sides of the specimen is diagrammed in Figure 7. A laser emits a light beam 65 which paεεes through beam expander 66 and lens 67 to pivotable mirror 68 and a scan system 69 for field εyntheεiε. The εcan εystem 69 contains a movable scan mirror 68 and telescope 58. The beam 65 is split by beam splitter 56. A portion of the light is directed by mirrors 71, 72 and 73 through phase control 52, tube lens 53 and objective 55 through the εpecimen 2. A second portion of the beam is directed by prism 54, lens 22, dichroic reflector 26 and objective lens 8 through the εpecimen 2. Light 29 emitted from the specimen 2 paεεes to camera 30 or oculars 42 through lens 8, dichroic reflector 26, barrier filter 46 and tube lens 44. The beam splitter and prism are used to amplitude-divide the expanded gauεsian beam, so that nearly flat wave fronts enter the specimen independently from each side. Since the coherence length of a 1 m laser is only about 30 mm, the two beam paths in the microscope are matched to within 5 mm. A piezoelectric drive 74 can be provided on a mirror 71, 72, 73 or dichroic reflector 26 in one beam path to adjust the phase. In this case mirror movement iε a full wavelength per cycle of the standing wave field. One advantage of the two beam system s that unaberrated wavefrontε enter the specimen on both sides,
compared to the mirror εyεtem of Figure 5 where phase errors accumulate on both passes when the specimen refractive index is heterogeneous. The second advantage of the configuration of Figure 7 is that by sweeping the beam angle ( θ ) in the εpecimen, εuch aε by rotation of mirror 68, standing wave fields of different spatial period can be time-multiplexed in the specimen during acquisition of a single image. If this iε done while keeping an antinode fixed at the object focal plane, εweeping provideε a means for synthesizing a nonperiodic excitation field peaked at the in-focus plane of the specimen.
The devices shown in Figures 4 thru 7 are particularly useful for two methods of specimen imaging. In one method, the specimen 2 is excited by a time- multiplexed sequence of standing wave fields that differ in axial node spacing. This generates a set of wave patterns which if superimposed would look like the upper graph of Figure 8. There three waves 75, 76, 77 indicated by solid, dotted and chain lines are shown. By superimpoεing a set of standing wave fieldε having different node spacing, but all having an antinode at the focal plane in the specimen an effective field that iε peaked at the focal plane can be εyntheεized.
Therefore, for every field in the εequence, fluorescence will be excited maximally at the focal
plane. Away from thiε plane, the phases of each field differ, so that the excitation intensity averaged over all fields in the εequence will be less. Optionally, a second fluorescence image can then be recorded with the same sequence of standing wave fields, except that each is adjusted to have a node at the focal plane. In this case, fluorescence will be excited minimally at the focal plane, but at similar averaged levels away from it. Digital pixel-by-pixel subtraction of the nodal image from the antinodal image gives a result for which the effective excitation field iε peaked at the geometric focal planeε, but decays at zero above and below the plane. The sum of a series of curves, such as the curves 75, 76 and 77, iε shown in the bottom graph of Figure 8. In practice, the sequence of standing wave fields can be generated by sweeping the beam-crossing angle θ through the full range acceεεible within the aperture of the objective lens and condenser by use of scan mirror 68.
Figure 9 illustrateε optical subsectioning. A specimen 2 has tagged objects a and b. In the first case 90, a nodal plane is made coincident with object "a", so that the image will show the fluorescence of "b". In the second case 92, the node has been shifted to the axial location of "b", so that a second image will then show "a". The process can be easily extended to three or more
close stratified objects, with the result that a series of images are obtained which contain linear combinations of contributions from each stratum. Linear digital processing can then be used to extract images corresponding to each stratum. ..
Two problems with standing wave fluoreεcence microscopy were immediately evident in our early work. First, the use of total reflection in the illuminator was convenient, but precluded the use of high-aperture immersion lenses. Second, as described, the standing wave data set consisted of a large number of images for each specimen focal plane position, since both θ and could be varied. Thiε made the method impractically εlow. The preεent methodε and apparatus of excitation field syntheεis reduce the data set to one or two images per focal plane, by multiplexing different standing-wave fieldε.
With the beamε counter propagating on axis ( θ = 0°), and an excitation wavelength of 514.5 nm, the node spacing in the specimen was determined to be equal to λ/2n in both the mirror and crosεed beam systems. In the first case, mirror movement of 0.17 μm corresponds to a shift of one fringe. In the two beam system, external mirror movement of 0.514 μm had the same effect.
The standing-wave microscope was used to view cytoεkeletal actin fibers in ^ixed 3T3 fibroblaεt cells. Cells were grown at a low density on 40mm cover glasses, fixed permeabilized, stained for F-actin with rhodamine- phalloidin, and mounted in a thin film of medium with a second cover glass on top. In the mirror εyεtem the specimens were oil-immerεed to both the objective and the mirror. In the two-beam εyεtem, oil-immersion lenses were used on both sides of the specimen. With the standing-wave field adjusted for maximum fringe flatness, the laminar pattern of fluorescence excited in the specimen could be easily seen as contour-like rings resulting from the intersection of the planar antinodeε with the mound-like volume occupied by the cytoεkeleton. Thiε was visible in even highly-flattened cells in which the cytoskeleton was entirely in sharp focus. By shifting the fringe position, different εetε of fiberε could alternatively be made to fluoreεce. In certain cases, this type of contrast reversal was caused by reflector movement corresponding to a fringe diεplacement of 0.04-0.05 μm . From the direction of fringe displacement, and known orientation of the εpecimen, the axial order of distinguishable features could be inferred.
A comparison of standing wave to uniform excitation was made by blocking one path in the
microscope immediately after the beam splitter. In thiε case, cytυεketetal fibers in all parts of the specimen fluoresce, and refocusing affected nearly all distinguishable features equally, imparting little 3d information. With the interference restored, regions of the cell coincident with antinodeε become approximately 4X brighter and nodal regionε nonemiεsive. An image of the same cell by confocal scanning fluorescence microscopy showed less discrimination between sets of actin fibers that show up in different standing wave images .
Although we have described and illustrated certain present preferred embodiments of our method and apparatuε for field εyntheεiε and optical εubsectioning for standing wave microscopy, it should be understood that our invention is not limited thereto, but may be variously embodied within the scope of the following claims .