WO2009087392A1 - Optical beam deflection apparatus and methods - Google Patents

Abstract

Apparatus and methods for deflecting a beam, preferably a laser beam are disclosed. The invention resides in the use of two deflectors with the deflection provided by the second deflector (30) at least partially counteracting the deflection provided by the first deflector (75). This allows a laser beam to be pointed to a selected spot for a predetermined dwell time and allows for rapid movement of the laser beam from point to point. Preferably, an inertia-limited mechanical deflector, such as a galvanometer mirror (75) or piston mirror is used to deflect the laser beam and a diffractive deflector, such as an acousto-optic deflector (30), is used to at least partially counteract the deflection of the mechanical deflector. It can be arranged that the laser beam points to a stationary position in space for a predetermined dwell time or for the laser beam to scan over a predetermined path.

Description

OPTICAL BEAM DEFLECTION APPARATUS AND METHODS

The present invention relates to apparatus and methods involving the manipulation of a beam of electromagnetic radiation, such as a laser beam. More particularly, the invention relates to apparatus and methods which use a laser beam to image a target space, such as by selectively focussing the laser beam in the target space, which may be a 2-D plane or a 3-D volume.

The ability to steer and focus electromagnetic radiation, such as a laser beam, rapidly in three-dimensions is very attractive for several applications in biology, microfabrication and data storage.

Laser scanning confocal imaging is an important and widely used tool in biology because it allows high contrast visualization of subcellular structures and monitoring of physiological processes with fluorescence indicators within living tissue by excluding contaminating of out-of focus light. Conventional confocal methods work best at relatively shallow depths where light penetration is good and scattering is minimal. Unfortunately, conventional confocal imaging cannot be used to image biological activity deep (>100 μm) within the living tissue. However, more recently, a new type of laser scanning confocal microscopy has been developed that relies on non-linear multiphoton excitation to selectively activate fluorophores where the light intensity exceeds the 2-photon threshold at the centre of the focal volume. Fluorescent light is emitted in all directions by these fluorophores and is picked up by a high numerical aperture lens system and photomultipliers. As the focal spot is scanned through the tissue the light intensity emitted by the fluorophores varies according to the intensity of staining by the fluorescence indicators in that part of the tissue. Combining the photomultiplier signal with the known position of the 2-photon focal volume enables a 2D or 3D image of the fluorescence intensity within the tissue to be reconstructed. This technique, known as two-photon microscopy, allows imaging at much greater depth because of the longer excitation wavelengths used for multiphoton excitation (wavelengths of 600-1150 nm), which scatter less than those used in conventional confocal imaging, and because confocality arises intrinsically from the excitation volume allowing all emitted photons to be used to construct the image. These properties together with the low levels of photodamage achievable have made 2-photon imaging an extremely powerful method for examining physiological processes at the cellular and subcellular levels both in vitro and in vivo. US 5,034,613 discloses a two-photon laser microscopy apparatus. Galvanometer mirrors are used to deflect the beam over the target volume of interest. We previously described in PCT/GB07/003455 filed on 12 September 2007 a system utilising Acousto-optic deflectors (AODs) to scan a target volume that overcomes a number of problems. The present invention relates to a further development of that system and the present invention can be used in combination with any of the solutions previously described in PCT/GB07/003455.

Two-photon imaging has been particularly popular in neurόscience, as it has allowed the dynamic properties of neuronal network activity to be imaged in intact brain tissue using calcium indicators. The spatial resolution of 2-photon microscopy is well suited to this task even allowing the small synaptic connections between neurons to be resolved. Multiphoton excitation has also begun to be used to photolyse "caged compounds" that release signalling molecules and dyes, allowing signalling within and between cells to be mimicked. This technique is potentially very important for understanding cell signalling and synaptic integration thus determining how individual neurons carry out low-level computations.

Conventional laser scanning microscopes, for example as disclosed in US 5,034,613, have traditionally used mechanical deflectors, such as galvanometer mirrors to scan a laser beam. Such galvanometer mirrors are configured to scan in the X-Y plane only. Focussing in the Z direction is achieved by moving the apparatus relative to the sample (for example by moving the objective lens closer to or further away from the sample). The use of mechanical deflectors such as galvanometer mirrors has an inherent disadvantage in that the deflectors necessarily have a mass and the speed at which the deflector can be moved from one position to another is limited by inertia. In practical terms, this means that it takes of the order of 200-1000 μs to move a typical galvanometer mirror from one selected position to another selected position. In turn, this limits the number of spots upon which a laser beam can be focussed during a given time frame. Gδbal et al, "Imaging Cellular Network Dynamics in 3-Dimensions Using Fast 3-D Laser Scanning", Nature Methods, Vol. 4, No. 1, January 2007 disclose a galvanometer based system in which the objective lens is vibrated to alter the Z focal position. This allows a 3-D volume to be scanned but is still not fast enough for neuroscience applications. The mass of the objective lens means that it can be vibrated only at 10 Hz and the use of continuously moving galvanometer mirrors mean that it is not possible to focus the laser point at a selected position for a useful dwell time (e.g. 20 μs) without slowing down the speed of movement of the galvanometer mirrors and thereby increasing the total time to scan the whole volume.

The temporal resolution of the present state of the art galvanometer-based two- photon imaging systems is one or two orders of magnitude too slow to accurately image signalling in a network of neurons. In such neurons, the elementary signal event (action potentials) occurs on the millisecond time scale. Moreover, the signals are spatially distributed in three-dimensions as they flow through the neural networks and building a 3-D stack of images using galvanometer-based technology takes minutes. Furthermore, galvanometers are too slow for studying synaptic integration from large numbers of widely distributed inputs in individual neurons using photolysis because the excitation beam needs to be moved to many (for example 30) sites within a millisecond in order to stimulate synapses distributed over the dendritic tree. For example, assuming that it takes 300 μs to move from one spot to another using a galvanometer mirror and assuming a dwell time at each spot of 5 μs, it would take 9.15 ms to image 30 sites. This is approximately 10 times too slow for current needs.

One approach suggested in the prior art to overcome some of the disadvantages is to use rapid non-mechanical deflectors, such as acousto-optic deflectors (AODs) instead of galvanometers to steer the two-photon laser beam. The advantage of using AODs is that they allow the laser beam to be moved much more rapidly from point-to- point than in a galvanometer-based system (compare a movement time of 5 - 25 μs with AODs to 200 - 400 μs with galvanometers). This has several potential advantages. Firstly, images can be scanned rapidly. Secondly, multiple point measurements can be made with long dwell times at very high temporal resolution (e.g. using an AOD system with 15 μs movement time, 33 points can be simultaneously sampled at 1 KHz sample rate with a 15 μs dwell time or in other words 33 points can be monitored 1000 times per second). The use of AODs therefore allows more of the time to be devoted to collecting photons from the regions of interest rather than being taken up in moving the laser beam between sites.

As well as deflecting the laser beam in the X-Y plane, the use of two AODs per axis can, in principle, also be used to focus the laser beam in the Z dimension. For example, Kaplan et al describe in "Acousto-Optic Lens with Very Fast Focus Scanning" Optics Letters, Vol. 26, No. 14, July 15 2001, pp 1078-1080, the use of two or four AODs to focus a laser beam in the X and Z plane or anywhere in an X, Y and Z volume. To achieve focussing in a 3D volume, two AODs for focussing in the X-Z plane are followed by two AODs for focussing in the Y-Z plane.

Although AOD-based systems give great advantages in terms of the speed at which points can be visited, there is an inherent problem in that AOD devices are usually designed to have a good transmission efficiency (e.g. approximately 80%) only for a narrow range of input acceptance angles. As the input acceptance angle increases, efficiency typically reduces. The use of two or more AODs in series only exacerbates this problem as the input acceptance angle of the second AOD defines the maximum deflection possible with the first AOD and this in turn defines the maximum deflection possible with the AOD system in total. Typically, deflection angles of 10-20 mrad are possible. Accordingly, AOD based systems can usually only be used for quite small target volumes, for example 250 x 250 x 250 μm.

Galvanometer mirror based systems do not suffer from this problem and can deflect the laser beam typically by angles of up to 40 mrad (or more) with no reduction in efficiency. Accordingly, the present state of the art provides two broad choices for practitioners. The first possibility is to use mechanical deflectors such as galvanometer mirrors so as to scan a larger target volume. The inherent inertia of the mechanical deflectors means that fast beam deflection is not possible and even the fastest, lightest, galvanometer mirrors struggle to reach speeds of 1 KHz when moving point-to-point. In non-pointing mode, when moving at their fastest speeds, the galvanometer mirrors create a constantly and smoothly moving laser spot that does not dwell at the point of interest for enough time to obtain a useful signal. This problem can be overcome by slowing the laser spot down so as to achieve a predetermined dwell time but this only means it takes longer for the laser beam to travel in between points. Accordingly, mechanical deflector based systems suffer the problem that only a very low fraction of the total scan time can be used for signal collection.

The second possibility is to use a non-mechanical deflector system, such as an AOD-based system. Here the beam can be deflected very quickly and speeds of 30-60 KHz are possible. However, only small target volumes can be scanned. For important biological applications like imaging neural networks over larger volumes, it would be desirable to have available a system which combined the large target volume achievable with galvanometer mirror based apparatus with the fast beam deflection obtainable with AOD systems. Furthermore, it is desirable if such a system has a good spacial resolution while maximising the light intensity for multi-photon excitation and the dwell time for which the laser remains at each point of interest, in particular sparsely distributed pre-chosen points of interest selected from the full scan volume.

These and other problems are addressed by embodiments of the present invention.

Apparatus and methods for deflecting a beam, preferably a laser beam are disclosed. The invention resides in the use of two deflectors with the deflection provided by the second deflector at least partially counteracting the deflection provided by the first deflector. This allows a laser beam to be pointed to a selected spot for a predetermined dwell time and allows for rapid movement of the laser beam from point to point. Preferably, an inertia-limited mechanical deflector, such as a galvanometer mirror or piston mirror is used to deflect the laser beam and a non-mechanical deflector, such as a diffractive deflector, or more particularly an acousto-optic deflector, is used to at least partially counteract the deflection of the mechanical deflector. It can be arranged that the laser beam points to a stationary position in space for a predetermined dwell time or for the laser beam to scan over a predetermined path.

The present invention accordingly provides a method of deflecting a beam using at least two beam deflectors. The'method preferably comprises using a first deflector to introduce a first deflection to said beam so as to move said beam in a first direction while simultaneously using a second deflector to introduce a second deflection to said beam.

The use of first and second deflectors allows high resolution to be achieved at fast speeds over a larger target volume.

Preferably, the deflections introduced by each deflector are continuously changing.

Advantageously, the movement of the beam that would be provided by said second deflector comprises a component in a direction opposite to said first direction so as to at least partially counteract the movement of the beam provided by said first deflector. This allows the spot to be held stationary so as to obtain a predetermined dwell time on the point of interest. Alternatively, it allows the beam to be scanned in a pattern that is different to the scanning pattern provided by the first deflector.

Preferably, the deflection introduced by the second deflector causes the beam to be focussed at a point of interest that does not lie along the path of the beam that would result if only the^;first deflector was present. Once the beam is focussed at the point of interest, a third, continuously changing, deflection may be applied to the beam such that the beam remains substantially stationary at the point of interest. Such continuously changing third deflection is preferably substantially equal and opposite to the first deflection. The first deflection is preferably a mechanical (i.e. inertia-limited) deflector such as a galvanometer mirror or a piston mirror.

The second deflector is preferably a diffractive deflector, such as an acousto- optic deflector.

The present invention also provides an apparatus for deflecting a beam. The apparatus preferably comprises a first deflector and a second deflector.

Preferably, the movement of the beam that would be provided by said second deflector comprises a component in a direction opposite to the direction of movement of the beam that is provided by the first deflector.

The method, apparatus and system of the present invention is particularly useful for implementing non-linear optical processes, such as multi-photon processes or two- photon processes.

The present invention is applicable to X-Y area scanning in two dimensions, X- Y-Z scanning in three dimensions, or may be applied only to a single axis, such as the X-axis alone, the Y-axis alone or the Z-axis alone. Any of the acousto-optic deflectors of the present invention are preferably made from a higher frequency anisotropic acousto-optic crystal of which TeO₂ is one example.

In the embodiments, the first and second deflectors are arranged in series along the path of the laser beam. It is preferred, but not essential, that the second deflector be arranged before the first deflector such that the laser beam encounters the second deflector first.

The present invention will now be further described, by way of non-limitative example only, with reference to the accompanying schematic drawings, in which:- Figure 1 shows an acousto-optic deflector (AOD) and the principle of diffraction of a laser beam using an ultrasonic acoustic wave;

Figure 2 shows an AOD focussing a laser beam; Figure 3 shows the moving focal spot obtainable with a single AOD (note that for a single AOD this spot represents a line focus perpendicular to the plane of the paper);

Figure 4a shows a graph of the frequency of the acoustic wave as it varies with time;

Figure 4b shows a graph of the frequency of the acoustic wave as it varies with distance across the AOD;

Figure 5 shows a configuration comprising two AODs which allow a laser beam to be focussed to a fixed spot in the X-Z plane; Figure 6 is a similar view to Figure 5 but additionally shows the undiffracted zeroth order component of diffraction;

Figures 7a - 7c show how a lens 70 can be used to focus the AOD output to a real position in a target;

Figure 8 shows a configuration of two parallel AODs in accordance with the present invention;

Figure 9 shows two examples of 25 x 25 point target areas and two paths of a focussed laser point provided by a galvanometer mirror;

Figure 10 shows an overview of the components of a two-photon system according to the present invention; Figure 11 shows two orthogonal views of an arrangement of four AODs in accordance with the present invention;

Figure 12 shows a 25 x 25 point target area and a point of interest lying outside of the scan path of a galvanometer mirror;

Figure 13 shows a close up of the vectors required to divert the beam from the galvanometer scan path to the point of interest;

Figure 14 shows the path sequence of the focussed light spot;

Figure 15 shows a close-up of part of Figure 14;

Figure 16 shows how the invention may be arranged to allow a plurality of spatially distributed spots to be visited in sequence; Figure 17 is a graph comparing the duty cycle possible with combined AOD and galvanometer scanning as compared to galvanometer mirror scanning alone;

Figure 18 illustrates how a larger area may be covered by four smaller scans using the present invention; Figures 19a and 19b show how the AODs may be used to move the focal point of the laser beam in the Z-direction and the spherical aberration that is caused; and Figure 20 shows a preferred embodiment of the invention in which the laser beam is deflected by AODs, galvanometer mirrors and a piston mirror.

Technical Background

In order to fully understand the invention, it is useful to explain the technical effects relevant to the invention. Figure 1 illustrates the principle of Bragg diffraction in an acousto-optic deflector. The acousto-optic deflector comprises a crystal 10 and a crystal transducer 12.

The crystal is preferably a high-efficiency anisotropic acousto-optic crystal such as a TeO₂ crystal. The crystal transducer 12 is attached to one side of the crystal and is arranged to propagate an ultrasonic acoustic wave 14 through the crystal, preferably using the slow shear mode of propagation. An incoming laser beam 16 entering the crystal at an angle Φj will be diffracted by the acoustic wave and the first order component of diffraction will have an angle Φ_∑ as shown in Figure 1. The first order component of diffraction is labelled 18 in Figure 1. There will also be a zeroth order component of diffraction that is simply a continuation of the input laser beam 16, i.e. the zeroth order of diffraction is an undeflected laser beam.

The laser beam 16 typically has a width of 10 to 15 mm and the plural beams illustrated in Figure 1 are merely illustrative of a single wide laser beam. The equation governing the angle of diffraction is:

where Φ_∑ — Φj is the angle of diffraction, λo is the wavelength of the laser beam, fa_c is the frequency of the acoustic wave propagating in the crystal and V_ac is the velocity of the acoustic wave propagating in the crystal. In Figure 1, the acoustic wave has a constant frequency f_ac.

By manipulating the acoustic wave propagating in the crystal, special effects can be achieved. For example, the acoustic wave can be "chirped" such that its frequency linearly increases or decreases with time, for example by giving it the form:

In this equation the constant a is known as the "chirp rate" and is measured in MHz per second. It is clear from this equation that the frequency of the ultrasonic wave is a linear function of time. Figure 2 shows the situation where the chirp rate a is negative, i.e. the frequency of the acoustic wave linearly decreases with time. As the angle of diffraction is proportional to the frequency of the acoustic wave, those parts of the laser beam that are deflected by the high-frequency portion of the acoustic wave will be deflected more than those parts which are diffracted by the low frequency portion. This is illustrated in Figure 2 and it can be seen that the effect is to focus the laser beam at a position in the general direction of the dotted arrow 20 in Figure 2. The distance D to the focal position in the vertical direction is given by the following equation:

VJ

D = (3) λoa

As illustrated in Figure 3, the acoustic wave moves in the direction of arrow 24 at the acoustic wave velocity V_ac. The focal position 22 created by the converging laser beam will thus also move in the direction of arrow 26 at the acoustic velocity. Accordingly, one AOD can be used to focus a laser to a line perpendicular to the page that is moving horizontally in the plane of the paper at the acoustic velocity V_ac.

It is also pertinent to point out that the range of acoustic frequencies that may be propagated through the crystal 10 is limited because the diffraction efficiency drops rapidly outside the design range of the AOD. Figure 4a shows the frequency of the acoustic wave as it varies with time and Figure 4b shows the frequency of the acoustic wave as it varies with distance.

As can be see from Figure 4a, it is necessary to keep the frequency of the acoustic wave between the limits f_m\n andf_max. It is therefore not possible to indefinitely chirp the frequency of the acoustic wave and, once the frequency reaches f_min it is necessary to very quickly change the frequency tof_max such that the chirping can continue. This creates a "saw-tooth" graph in Figure 4a. This same saw-tooth pattern occurs in Figure 4b, but it is reversed because the frequencies present in the acoustic wave on the right-hand side of the crystal represent frequencies at an earlier time point than the frequencies present in the acoustic wave at the left-hand side of the crystal.

For one design of AOD₅ typical values ϊoxf_min are 50-60 MHz and typical values fotfmax are 90-100 MHz. However, a special design of AOD may be provided that is more efficient at lower frequencies, for example 20-50 MHz, more preferably 25-45 MHz, more preferably still 30-40 MHz and more preferably still 32-37 MHz. f_min and T^rnay thus be chosen in accordance with these lower and upper limits. A low range of acoustic frequencies is useful because although they reduce the deflection provided by each AOD they reduce the need to provide AODs that have large acceptance angles. This allows the efficiency to be kept high. The trade off between AOD acceptance angle (and hence useful deflection angle) and AOD efficiency is thus a system design trade off.

For those points in time where the "fly-back" portion of the graph is present in the central region of Figure 4b or, in other words, for those points in time where the discontinuity between the highest and lowest frequency exists in the crystal of the AOD, proper focussing cannot be achieved. There are therefore certain periods of time for which the AOD cannot be used for focussing. In two-photon applications, it is therefore important to measure signals induced by the laser pulses only at points in time where there is minimal discontinuity in chirped frequency across the AOD. There is therefore a "duty cycle" limitation on the AOD which duty cycle is the amount of time, expressed as a percentage, that the AOD may be used for useful focussing. It is apparent that this duty cycle will be reduced by increasing the gradient of frequency increase/decrease in Figures 4a and 4b.

The focal spot 22 can be made stationary by utilising a second AOD, as described by Kaplan et al (supra) and as illustrated in Figure 5.

In this configuration, a second AOD crystal 10 and ultrasonic transducer 12 is utilised and the ultrasonic waves in the AODs are made to propagate in substantially opposite directions. In Figure 5, the first (upstream) AOD has an ultrasonic wave propagating from the right to the left and the second AOD has an ultrasonic wave propagating from the left to the right. The first AOD modifies the input laser beam 16 to be a focussed laser beam 18 with the focal spot moving substantially from the right to the left and the second AOD modifies the laser beam 18 to be a stationary focussed laser beam 28. As illustrated in Figure 5, resultant focal spot 22 does not move. Figure 6 shows the same set up as Figure 5 but additionally shows the undiffracted beam (known as the "zeroth order component of diffraction") that is transmitted through the first AOD. Due to the offset positioning of the AODs, the undiffracted beam passes well to the right of the focal spot 22 and so does not interfere with the light reaching the focal spot 22. Baffles or other mechanisms may be used to cut the undiffracted beam out of the system altogether.

For the sound wave direction and diffraction order illustrated, utilising a chirp rate of zero (as shown in Figure 1) provides a parallel laser beam. Utilising a negative chirp rate (as shown in Figure 2) provides a converging laser beam. Utilising a positive chirp rate provides a diverging laser beam. These three possibilities are illustrated in Figures 7a, 7b and 7c. In any practical system the AODs will be followed by one or more lenses 70 which serve to provide further focussing. Thus, whether the laser beam leaving the AOD system is converging (Figure 7a), parallel (Figure 7b) or diverging (Figure 7c) the subsequent lens system brings the laser beam to a real focus. The system is preferably calibrated such that when the laser beam leaving the AOD system is parallel (Figure 7b) the point at which the subsequent lens system 70 focuses the beam is designated the Z = O point. Then, for this configuration, applying a positive chirp rate moves the resultant focal point upwards (see Figure 7a) and applying a negative chirp rate moves the focal point downwards (see Figure 7c). In practice, the laser beam passes through several lenses before reaching the physical target.

Figure 8 illustrates how the focal spot 22 can be moved within the target volume. The following and subsequent explanations ignore the effect of subsequent lens systems (such as the lens 70 in Figure 7) in order to provide clarity. In any practical embodiment, such a lens system will be present and the principles below apply equally to the case when the AODs themselves provide a diverging laser beam (in which case there is a virtual focus above the laser beams that is relayed by the subsequent lens optics to a negative Z position). In order to assist in understanding this aspect of the invention the following Figures take the example when the chirp rate is positive which in this configuration produces a converging laser beam. As explained above, the distance to the focal position is inversely proportional to the chirp rate a. Increasing the chirp rate therefore brings the focal position upward in the Z direction and decreasing the chirp rate brings the focal position downward in the Z direction. As explained in Figure 8, varying the slope of the frequency time graph (i.e. modifying the chirp rate a) serves to move the focal position 22 in the Z direction. As also illustrated in Figure 8, the focal position 22 may be moved in the X direction by varying the separation between the two ramps in the frequency-time graph. When the two AODs are excited with acoustic waves that are identical and without any chirp, the resultant focal position is defined as the X = O₅ Z = 0 position. When a chirp is introduced, this moves the focal position in the Z direction. When the absolute frequency of the waves applied to the two AODs differs, this causes the focal position 22 to be moved in the X direction.

As can be derived from the above, two AODs can be used to focus the laser beam in the X-Z plane. Two further AODs₅ rotated 90° about the Z-axis compared to the X-Z AODs₅ can be used to focus the laser beam in the Y-Z plane. Thus, with four AODs₅ the beam can be arbitrarily focussed within a target volume of about 250 x 250 x 250 μm.

It is possible with an AOD-based system to cause the focal position to "jump" from one point to another. This can be achieved by filling the AOD with acoustic waves of a chirped frequency as to cause the laser beam to be focussed at a first position and then filling the AOD with acoustic waves of a different chirped frequency so as to cause the laser beam to be focussed to a different position. The time it takes to make this "jump" is limited by the so called "AOD fill time". This time is typically equal to the width of the AOD divided by the speed of the acoustic wave in the AOD. For example, if the AOD is 15mm wide and the acoustic waves travel at 600 m/s, then the AOD fill time is 25 μs.

In addition to an AOD fill time, there is also a "reset time", which is the time it takes to reset the frequency in the electronics. This is typically 4 μs. Accordingly, the time that it takes to "jump" from one focal point to another using an AOD is around 29 μs.

Galvanometer Mirror Deflectors

It is highly desirable in neuroscience applications to cause a beam of light to sequentially visit a number of spots selected within the target volume. The position of the spot will be determined by the structure of the tissue in question and, usually, the selected spot positions are seemingly randomly spread throughout the volume. Accordingly, it is usually not possible to define a smooth path through the volume that directly visits each spot of interest. A system which used a galvanometer mirror to sequentially visit each point would be very slow indeed. The reason for this is that the speed at which a galvanometer mirror can be adjusted, so as to deflect a laser beam from being focussed at a first point to being focussed at a second point is very low. Galvanometer mirrors are thus not suitable for jumping from point to point at high speeds. However, galvanometer mirrors can be used to produce a quite fast moving spot along a smooth curve, for example, a circle, an ellipse or Lissajou figure.

In principle, a X-Y galvanometer mirror scanner can be used to set the laser beam moving in a circle or an ellipse in the X-Y plane. The cycle time (i.e. the time required for the laser beam to complete a single circle or ellipse) is of the order of lms. Two possible laser beam routes 102 in the X-Y plane comprising a target volume of 25 x 25 points are illustrated in Figure 9.

For any three randomly selected points 120 in the target volume, it will be possible to define an ellipse that passes through all three points. Accordingly, with a galvanometer mirror based system, it is possible to guarantee hitting three points within the volume during each cycle. Or, in other words, it is possible to hit three points every 1 ms.

As will be noted from Figure 9, the great majority of the time the beam is not pointing to a point of interest. In fact, the beam is only pointing to one of the three points of interest 120 for approximately 5% of the time. Furthermore, the "dwell time" on each spot is defined by the speed of the laser beam and the size of the spot. For a cycle time of 1 ms, the beam would be pointing to a spot of interest for a dwell time of approximately 17 μs.

As the number of points in the target volume increases (i.e. the total number of points increases so that the size of each point reduces), the less dwell time there is at each point and the lower the duty cycle, because the laser beam spends even less time pointing to each point of interest compared to the time that is spent pointing in between the points of interest. To very roughly estimate how duty cycle varies with the number of points in the target volume, consider a square of nxn points. Consider a circle of diameter 0.7x length of the side of the square. For this approximation this is close enough to the average path length of an ellipse through any three randomly chosen points averaged over many points. It is clear that the size of the point is inversely related to n . Simple geometry shows that the duty cycle is 3/(0.7 x π x n) . Thus, for instance, with a target area of 3000 x 3000 points, the duty cycle reduces to 0.05%. Two-Photon Microscopy System

Figure 10 shows a two-photon microscopy system in accordance with the present invention. An input laser beam 16 is passed through four acousto-optic deflectors 30, 40,

50, 60, a tube lens 70, a galvanometer scanner 75, a further tube lens 80 and an objective lens 90. The laser beams forms a focal spot 22 in the first image field which has Cartesian axes XiI, YiI, ZiI. This image is projected through other relay optics (not shown for clarity) which can create a second image field Xi2, Yi2, Zi2. This is projected by a tube lens 80 through a microscope objective lens 90 to form a focal spot 32 in the third image field Xi3, Yi3, Zι3. This third image field is the target field and, in two- photon applications, the target is placed in this field. Such a target might be a slice of brain tissue or other biological material with a fluorescent dye that requires imaging. The input laser beam 16 in two-photon applications takes the form of an ultra- short femtosecond or picosecond pulse in order to get sufficiently intense electric fields at the focal point. The pulses are typically spaced in time by a duration very much larger than the pulse length. Typical pulse lengths are 2-5 ps or less, preferably 500 fs or less, even more preferably 50 to 200 fs. The pulses are typically repeated at a frequency of 50 to 200 MHz (e.g. 80 MHz). Two distinct experiments can be carried out with a two-photon microscopy system. The first experiment is to image fluorescent materials and such experiments typically require powers of 10 mW to be focussed to an area of just over 1 μm² (corresponding to a power density of around 600,000 W/cm²). Typical laser wavelengths of 650 - 1100 nm (e.g. 800 - 1000 nm or more preferably 850 nm) are utilised. The second experiment is photolysis in which the laser is used to uncage biologically active compounds. Lasers having a wavelength of 650-750 nm (e.g. 720 nm) are often used and the power requirement is much higher, there being a need for in excess of 100 mW.

In a preferred embodiment of the invention, the laser is supplied by a mode locked Ti sapphire laser tuneable in the near infrared region having an average power of 1 to 10 W and supplying 100 fs pulses at 80 MHz.

Sensitive collection photomultipliers are utilised near to the target area to pick up any fluorescence from the two-photon excitation of fluorophores in the target. This enables a 3D image to be constructed in imaging applications and further enables any sequence of points in 3D space to be interrogated by the laser beam for repeatedly monitoring the state of tissue at each point during dynamic biological processes.

The AODs used in the present invention are preferably shear-mode anisotropic AODs. Suitable materials for the AOD crystal are TeO₂ crystals. Such AODs rotate the polarisation of incoming laser light by 90°. The AODs 30, 40, 50, 60 are schematically illustrated in Figure 10 (and in other Figures of the present application) with no intervening components between them. However, in practice, such components will be present. Typically, these components may include half-wave plates and polarisers (the reason will be explained later). Furthermore, a telecentric relay can be used between each AOD (as disclosed by Reddy & Saggau, "Fast Three-Dimensional Laser Scanning Scheme Using Acousto-Optic Deflectors", Journal of Biomedical Optics, 10(6), November/December 2005) to properly couple the AODs together. If such a telecentric relay were not used, then it would be difficult to achieve a stationary focal position, without taking other measures. The measures to achieve a compact configuration disclosed in PCT/GB07/003455 may be used.

The light emitted by the fluorophores is picked up by a photomultiplier (not shown) coupled to the system by a dichroic mirror in the standard fashion.

Figure 11 shows two orthogonal views of the AOD configuration. The first AOD 30 and second AOD 40 are used to provide focussing in the X-Z plane. The third AOD 50 and fourth AOD 60 are used to provide focussing in the Y-Z plane. As is apparent from Figure 11, the AODs are configured in the order first, third, second, fourth starting from the laser beam entry end and finishing at the laser beam exit end. This configuration is preferred because it avoids the need to utilise half- wave plates. Not shown in Figure 11, but preferably present in a practical embodiment, are first to fourth polarisers. A polariser is located subsequent to each AOD. Laser light 16 entering the first AOD 30 will be converted into a zeroth order component of X polarisation and a first order component of Y polarisation. It is desirable to only transmit the first order component. A Y polariser is therefore located after the first AOD to block the zeroth order component. This Y polarised light is suitable for input into the third AOD 50 in which a zeroth order component of Y polarisation and a first order component of X polarisation is produced. A X polariser is therefore located after the third AOD 50. Such X polarised light is suitable for input into the second AOD 40 which produces a zeroth order component having X polarisation and a first order component having Y polarisation. A Y polariser is therefore located after the second AOD 40. This serves to block the zeroth order component. Such Y polarised light is suitable for acceptance by the fourth AOD 60 which produces a zeroth order component having Y polarisation and a first order component having X polarisation. An X polariser is therefore located after the AOD 60 to block the Y polarised zeroth order component. As a result, all light reaching focal spot 22 is the result of properly diffracted first order components and no undiffracted zeroth order components can filter through the system. Furthermore, this configuration does not require a half- wave plate to adapt the polarisation at various stages.

As is well known to those skilled in the art of AODs, the precise degree of polarisation of the first order diffracted wave, although close to linear and at 90 degrees to the direction propagation of the acoustic wave, is not exact. Particularly if the AOD crystal is cut with less than 2 or 3 degrees deliberate misorientation of the optic axis from the direction of propagation of the acoustic wave, the optimised input beam and the diffracted and zero order output beams of light can be slightly elliptically polarised so the configurations described here, which use linear polarisers would not maximally transmit the diffracted wave nor perfectly suppress the undesired undiffracted zero order components of the light. In such cases, to further improve performance, small rotations of inserted half wave plates or insertion of appropriate phase plates with small fractions of a wave correction (e.g. 1/4 or 1/20 wave) may fine tune the performance of the configuration concerned. The key point is for the polariser after each AOD to maximally transmit the wanted diffracted first order beams and maximally suppress the unwanted zero order beam. If the polariser is before another AOD₃ then there may be more polarisation state adjustment before the next AOD to optimise its performance. The present invention proposes a system and method in which a first deflector, such as an inertia-limited galvanometer mirror or piston mirror, is used to deflect a laser beam and a high speed second deflector, such as a diffractive deflector (for example an AOD configuration) is used to at least partially counteract the effect of the inertia- limited deflector. This may provide that the laser spot is quickly moved from one position of interest to the next position of interest and ensures that the laser beam is focussed at the point of interest for the necessary "dwell time". The result is a series of jerky movements of the laser beam in which the laser beam stops and "stares" at a succession of selected points with very fast movement of the beam from one point to another. This is optimal for neuroscience applications as a useful signal can only be obtained while the laser is properly focussed at the positions of interest and any time period for which the laser is not so focussed (for example time periods inbetween the "staring" while the laser spot is moving) cannot be utilised to obtain measurements. Any type of mechanical deflector can be used to provide the first amount of deflection. Mechanical deflectors are preferred because they are able to deflect the beam over relatively large angles. A mechanical deflector is one that has moving parts that have a mass, such that the mechanical deflector is limited in speed by its inertia. Any type of deflector can be used to provide the second counter-deflection. Non-mechanical deflectors (i.e. those which do not have inertia limited moving parts) such as diffractive deflectors are preferred. AODs are the most preferred type of diffractive deflectors. Although the mechanical deflector (e.g. comprising a galvanometer mirror) is referred to as a "first" deflector and the diffractive deflector (e.g. comprising the AODs) is referred to as a "second" deflector, this does not imply any positional limitation. In fact, the preferred embodiment has the diffractive deflector first in the laser beam path compared to the mechanical deflector. The system and method of the present invention solves two of the above discussed problems. Firstly, the system allows the laser to spend more of the time pointing at the points of interest. This allows the "dwell time" to be increased. Secondly, the system allows the laser to spend less time pointing at positions other than the points of interest. This allows the duty cycle to be increased. This is achieved by using both mechanical deflectors (e.g. galvanometer mirrors) and high speed deflectors (e.g. AODs) in combination to deflect the laser beam. The galvanometer mirrors can be set to some smoothly varying path that passes near the points of interest. The AODs, thanks to their very fast response time, can be used to deflect the beam from the standard smooth path dictated by the galvanometer mirror to the point of interest. The AODs can be programmed to then deflect the beam in the exact opposite way to the deflection provided by the galvanometer mirrors so as to cause the beam to stop over the point of interest for a predetermined dwell time. After that, the AODs can cause the beam to be deflected directly to the next point of interest. The present invention can provide that the focussed beam undergoes a series of jerky movements in which the beam spends more time at the point of interest and less time moving in between points. The resulting movement is analogous to the saccade scanning performed by the human eye. The principle is illustrated in Figure 12 for one spot 100 that does not lie on the path 102 defined by the galvanometer mirror deflector. Figure 13 shows a close up of the beam diversion.

When the AOD (or system of AODs) is programmed so as to not deflect the laser beam (i.e. no acoustic waves are provided in the AODs)₅ then the path of the laser beam would be determined only by the galvanometer mirror system, as depicted by path 102 in Figure 12. In order to point the laser beam at a spot not lying on this path 102, the AOD deflector can be used to cause the laser beam to "jump" to the point 100, stay there a predetermined time before jumping back to the path dictated by the galvanometer mirrors .

The box 130 defined by the dashed line in Figure 12 describes the range of deflection possible with the AODs. Accordingly, this box 130 rotates around the diagram in Figure 12, always having its centre on the current laser beam focal point defined by the galvanometer mirror. The arrows attached to the corners of the box 130 in Figure 12 depict the fact that this box is moving and the vector of movement is identical to the vector of movement of the focussed laser spot due to the galvanometer mirror.

In the method of the present invention, the AOD configuration is firstly programmed so as to deflect the beam through the vector 104 shown dotted in Figures 12 and 13. Once the AODs have been filled with the appropriate chirped frequency (which takes an amount of time called the "AOD fill time"; typically 24 μs), the beam will be pointing at the point of interest 100. During the AOD fill time, the laser beam may be inadvertently momentarily focussed to a different position or there may be no focussed point anywhere in the target volume. For such a time, the laser can be switched off or simply no signal recording takes place. Now that the laser beam is pointing at the point 100, to ensure a predetermined dwell time, the AOD configuration is programmed to smoothly deflect the beam at the same speed and with the opposite direction to the deflection provided by the galvanometer mirrors. This is shown by the bold arrow 106 in Figures 12 and 13. Once the laser has been focussed at the point of interest for the predetermined dwell time (e.g. 20 μs), the AOD can be refilled so as to cause the laser beam to jump along the vector 108 to the next point, which in the case of Figure 12 again lies on the path of the galvanometer mirror. In this example, the vector 108 can easily be achieved by centring the drive frequencies to each AOD so that the 4 AODs have no net deflection. Figure 14 shows the path sequence of the focussed light spot without showing the vectors of the AOD deflection. As can be seen from the close-up view of Figure 15, the combined effect of the galvanometer mirrors and the AODs is to cause the laser beam to "jump" to the point of interest 100 and, after the predetermined dwell time, to "jump" back again to the path of the galvanometer induced deflected beam.

Rather than simply doing this once for each cycle, the system can be arranged to cause the laser beam to "jump" directly from point to point. For a cycle time of 1 ms around 20 such points can be visited and the concept is illustrated in Figure 16 (although less than 20 spots are shown for clarity). The dotted arrow 112 shows the laser beam jumping from point to point. The smoothly varying deflection produced by the galvanometer mirrors is illustrated by the line 102 in Figure 16. As can be seen, the AODs induce additional deflection so as to jump the laser beam from the path determined by the galvanometer mirrors to the point of interest and to cause the laser beam to stop there for the predetermined dwell time. After this dwell time, the AODs are filled with acoustic waves which cause the laser beam to jump to the next point of interest, where it again remains stationary for the predetermined dwell time, and so on. Performing scanning with both the galvanometer mirrors and AODs in this way allows about 20 points to be visited during each lms cycle, assuming a 20 μs dwell time, and a 30 μs combined reset time and AOD fill time. Furthermore, the dwell time can be chosen almost arbitrarily and is not fixedly related to the speed of galvanometer mirror scanning. Comparing Figure 9 with Figure 16, using galvanometer mirrors alone (Figure 9) allows three points to be sampled per cycle with a dwell time of approximately 17 μs per point. With the combined galvanometer/AOD system, 20 points can be visited in the same cycle time with a dwell time of 20 μs. The duty cycle (i.e. the percentage of time spent looking at the points of interest) is about 5% in Figure 9 and about 40% in Figure 16.

As shown in the graph of Figure 17, if the system of Figure 9 was scaled up, such as for example to look at 3 points selected from a target area of 3000 x 3000 points (as opposed to 25 x 25 points), then the duty cycle would fall to 0.05%. However, with the combined galvanometer/AOD system, the duty cycle remains at 40% thanks to the ability to jump the laser beam quickly (i.e. in 30 μs) from point to point and to cause the laser beam to stay looking at one point by having the AOD configuration create a deflection that is equal and opposite to the deflection created by the galvanometer mirror. It is not necessary that the continuous deflection provided by the AOD configuration completely counteracts the deflection provided by the galvanometer mirrors. When such complete counteraction takes place, a stationary laser spot is achieved. However, a stationary laser spot is not always necessary. For example, some applications may require a target volume to be scanned with a laser beam moving at a constant speed. This can be performed adequately with an AOD system alone, provided that the target volume is 250 x 250 x 250 μm or less. For larger target volumes, the system of the present invention can be used to create a scan configuration known as "dynamic tiling". Here, the galvanometer mirror system is set to cause the laser beam to be focussed along a smooth path passing around the area of volume. The AOD deflector is then used to scan a small section of the 3 dimensional space in the vicinity of the path dictated by the galvanometer mirror. This is illustrated in Figure 18. The desired resultant scan path is labelled 110 in Figure 18. The dotted lines represent AOD "jumps". The scan path that would be achieved with the galvanometer mirror alone is shown as 102. As can be seen, the AOD deflector is programmed so as to jump the laser focal point from the path dictated by the galvanometer mirrors to the beginning of the desired scanning line 110. Thereafter, the AOD causes a deflection equal to the vector sum of the opposite of the deflection caused by the galvanometer mirrors (to counteract this deflection) added to the required deflection according to the desired scan pattern. With this system, the laser beam runs along the desired scan line 110. Once scanning of this "tile" 130 has been completed, the laser beam jumps to the beginning of the next tile 132 to scan that. In one revolution of the galvanometer mirror, several (e.g. 4 in Figure 18) mini-scans can be achieved with the result that the whole area is scanned, preferably at constant speed. This can of course be extended to 3 dimensions using exactly the same principle. For example, a 3D Lissajou figure or ellipse can be traced out in space using the diffractive deflector(s) so as to scan a volume. Alternatively, a fast 3D raster scan can be carried out in space using the diffractive deflectors such that the entire volume to be scanned is broken down into "cubes" or, with the mechanical deflectors tracing a path through or near to each of the cubes. As described above, the present invention allows fast point-to-point movement over a large target area in the X-Y plane or a large target volume in X-Y-Z with a very good duty cycle. An improvement allowing greater movement in the Z-direction can be obtained by incorporating a component, such as a piston mirror, that mechanically moves the focus point in the Z direction. Botcherby et al "Aberration-free optical refocusing in high numerical aperture microscopy" , Optics Letters, Vol. 32, No. 14, 15 July 2007, disclose a method for moving the focus of a beam in the Z direction whilst avoiding spherical aberration. This involves moving a mirror in the Z direction. Accordingly, a piston mirror arrangement similar to that disclosed by Botcherby et al can be incorporated into the system of the present invention so as to provide the capability to move the focus point over larger ranges of Z with reduced spherical aberration.

Figure 19 illustrates the problem of spherical aberration.

The objective lens 90 used to focus the laser beam to the target point is optimised for a particular focal position. This is illustrated in Figure 19a where it is shown that the AODs do not affect the incoming laser beam such that the laser path remains parallel as it travels to the objective lens 90, whereupon it is focussed to the focal point 32.

When the AODs are programmed so as to increase the divergence of the incoming laser beam (the bold line in Figure 19b) or decrease the divergence of the incoming laser beam (see dotted line in Figure 19b) this has the effect of moving the focal point 32 downward or upward in the Z direction respectively. However, with a high numerical aperture (NA) (e.g. NA>1) this also introduces spherical aberration which is illustrated in Figure 19b by the various beams not all crossing the focal point at exactly the same position. Accordingly, the focal point 32 will be blurred. This reduces 2-photon excitation and reduces resolution. Botcherby et al overcome this problem by moving a piston mirror so as to alter the position of the focal point in Z without introducing spherical aberration. That disclosure is incorporated herein by reference. Figure 20 illustrates one system, representing the best mode, in accordance with the present invention. The laser beam 16 is first routed to the XYZ AOD scanner and focuser 45 which typically comprises the four AODs shown in Figure 11. Thereafter, the laser beam is routed to the optional XY galvo scanner 75 for providing a continuous beam deflection in the X-Y plane and then on to the piston mirror Z-focussing apparatus disclosed in Botcherby et al. This comprises a piston mirror 85 which, when moved along the path of the laser beam, changes the focus achieved at the objective lens 90. The piston mirror 85 can be moved at speeds of 10-1000 Hz so as to provide large scale Z deflections with no spherical aberration. The AOD system 45 can be used to provide complementary deflection in the Z direction such as to cause the focal point 32 to move to the selected point of interest and to remain there for the predetermined dwell time, or to start a smooth scan different to that dictated by the mechanical deflectors. Accordingly, the concept of using an inertia-limited mechanical deflector in combination with fast diffractive deflectors can be applied in this way.

As shown in Figure 20, the AOD system 45 can be combined with both the X-Y galvo scanner 75 and the piston mirror 85 so as to achieve a system that can quickly, and at high duty cycle, sequentially visit a series of points (e.g. 30 points of interest/ms) within a larger volume (e.g. 500 x 500 x 500 μm) than would be possible using AODs alone and at speeds that are much quicker than when using galvanometer mirrors alone.

Claims

1. A method of deflecting a beam using at least two beam deflectors, said method comprising: using a first mechanical deflector to introduce a first, continuously changing, deflection to said beam so as to move said beam in a first direction; simultaneously using a second deflector to introduce a second, continuously changing, deflection to said beam; wherein the movement of the beam that would be provided by said second deflector comprises a component in a direction opposite to said first direction so as to at least partially counteract the movement of the beam provided by said first deflector.

2. A method according to claim 1, wherein the movement of the beam that would be provided by said second deflector is substantially equal and opposite to said first direction so as to provide a stationary focal position at a point of interest.

3. A method according to claim 1, wherein the result of the first and second deflections is to provide a focussed laser beam that moves along a predetermined path.

4. A method of saccade scanning, said method comprising: using a first deflector to introduce a first, continuously changing, deflection to a beam so as to move said beam in a first direction; simultaneously using a second deflector to introduce a second deflection that causes the beam to be focussed at a point of interest; using the second deflector to introduce a third, continuously changing, deflection to said beam such that said beam remains substantially stationary at said point of interest.

5. A method according to any one of the preceding claims, wherein said first deflector is one or more galvanometer mirrors.

6. A method according to any one of claims 1 to 4, wherein said first deflector is a piston mirror.

7. A method according to any one of the preceding claims, wherein said second deflector is a diffractive deflector.

8. A method according to claim 7, wherein said second deflector is one or more acousto-optic deflectors.

9. Apparatus for deflecting a beam, said apparatus comprising: a first mechanical deflector arranged to introduce a first, continuously changing, deflection to said beam so as to move said beam in a first direction; a second deflector arranged to introduce a second, continuously changing, deflection to said beam; said apparatus being arranged such that the movement of the beam that would be provided by said second deflector comprises a component in a direction opposite to said first direction so as to at least partially counteract the movement of the beam provided by said first deflector.

10. Apparatus according to claim 9, wherein said second deflector is arranged to provide a second deflection that provides a stationary focal position at a point of interest.

11. Apparatus according to claim 9, wherein said first and second deflectors are arranged such that the result of the first and second deflections is to provide a focussed laser beam that moves along a predetermined path.

12. Apparatus for saccade scanning, said apparatus comprising: a first deflector to introduce a first, continuously changing, deflection to a beam so as to move said beam in a first direction; a second deflector, said second deflector being arranged to:

(i) introduce a second deflection that causes the beam to be focussed at a point of interest; and

(ii) thereafter introduce a third, continuously changing, deflection to said beam such that said beam remains substantially stationary at said point of interest. I

13. Apparatus according to any one of claims 9 to 12, wherein said first deflector is one or more galvanometer mirrors.

14. Apparatus according to any one of claims 9 to 12, wherein said first deflector is 5 a piston mirror.

15. Apparatus according to any one of claims 9 to 14, wherein said second deflector is a diffractive deflector.

10 16. Apparatus according to claim 15, wherein said second deflector is one or more acousto-optic deflectors.

17. A method or apparatus according to any one of the preceding claims, wherein said beam is a laser beam and the system further comprises a laser for supplying said

15 laser beam.

18. A method or apparatus according to claim 17, wherein said laser is arranged to supply a pulsed laser beam.

20 19. A method or apparatus according to claim 18, wherein said pulsed laser beam comprises pulses having a length of 5ps or less.

20. A method or apparatus according to any one of the preceding claims, wherein the beam passes through a system comprising microscope optics.

25

21. A method or apparatus according to any one of the preceding claims, wherein there is provided an objective lens for directing said beam onto or into a target volume.

22. A method or apparatus according to any one of the preceding claims, wherein 30 the method is for implementing a non-linear optical process.

23. A method or apparatus according to claim 22, wherein the non-linear process is a multi-photon process.

4. A method or apparatus according to claim 23, wherein the multi-photon process a two-photon process.