Field of the invention

The invention relates to a scanning microscope for imaging a sample, comprising - imaging optics for generating an imaging light spot; a scan mechanism for translating the sample relative to the imaging light spot in a horizontal plane, a focusing mechanism for translating the sample relative to the imaging light spot on a vertical axis.

Background of the invention

In such fields as microbiology and digital pathology, scanning optical microscopes are used for producing digital images at microscope resolution of glass or plastic slides containing, for example, cells or tissue slices.

Figure 1 shows a schematic cross section of a typical pathology slide 12. On top of a glass substrate 18 about 1 mm thick, there is placed a slice 24 of the tissue under investigation. The slice 24 is generally contained in a horizontal plane 20 but exhibits a number of variations which may require re-focusing. The slice 24 is a few microns thick. Around and on top of the tissue 24, there is a specimen layer 16 of index matching fluid 22 of variable thickness, but normally in the range of about 8-15 μm. The refractive index of the fluid 22 is only slightly different from that of the glass 18. On top of the fluid 22 there is a layer of cover glass 14 (or transparent plastic with similar refractive index) about 0.14-0.17 mm thick. There are a total of four distinct interfaces: an interface 26 between air 34 and the cover glass 14, an interface 28 between the cover glass 14 and the specimen layer 16, an interface between the specimen layer 16 and the substrate glass 18, and an interface between the substrate glass 18 to air 34. The interfaces 26, 28, 30, 32 have been represented in the Figure in an ideal configuration in which they are all parallel to each other. In practice marked deviations from the parallel arrangement occur. Such deviations present a challenge for automatic focusing. In order to implement the imaging process in an automated manner, some form of auto focus is required. More specifically, a good focus position of a microscope objective lens relative to the sample needs to be known for each position in the scan. The accuracy at which the focus position needs to be known is typically of the order of 1 micron or less. WO 2004/095360 Al proposes producing a so-called focus map prior to the image acquisition scan. The best focus is determined for a limited number of positions on the slide, and from these the best focus for other positions is derived by interpolation or fitting of a function through the predetermined focus points.

Furthermore, there are a number of methods for ensuring a continuous auto focus. Many continuous auto focus methods known from optical storage device rely on the presence of a single, highly reflecting flat surface that the system can focus on.

US 7,071,451 B2 describes an auto focus control method that can be used for biological imaging. The method is implemented in a microscope system having as one of its distinguishing features a separate auto focus unit containing a lens with an adjustable location. The lens is used as a means for adjusting an offset between a position that is tracked via the auto focus branch and the location of the sample of interest. The position of this lens remains fixed during the scanning of the sample.

Summary of the invention

A disadvantage of the auto focus method disclosed in US 7,071,451 B2 is that it is not suitable for digital pathology, where the tissue of interest is embedded between two closely spaced glass-fluid interfaces. These interfaces would result in complicated interference patterns between the signals coming from both interfaces. Of the highly reflective flat surfaces, the top surface of the cover glass is nearest to the objective. However, this surface cannot be used for continuous autofocus since the distance between this surface and the tissue is variable in a range that is larger than the required auto focus accuracy of less than 1 micron.

It is therefore an object of the invention to provide a continuous auto focus system for a microscope for imaging a biological specimen. This object is solved by the features of the independent claim. Further specifications and preferred embodiments are outlined in the dependent claims. According to the invention, the scanning microscope further comprises auto focus optics for generating during an auto focus interval a set of auto focus light spots in the sample, the auto focus light spots having different vertical positions; a detector for receiving auto focus light from each of the auto focus light spots via a confocal aperture and for delivering a detector signal as a function of the received autofocus light; a controller for determining from the detector signal a vertical position of an object in the sample and for causing the focusing mechanism to adapt the vertical position of the imaging light spot to the vertical position of the object in the sample.

The reflected light may thus be detected substantially simultaneously from different depths in the sample in a confocal manner. The focus estimation may work in real time and can be used to continuously and automatically adjust the position of an objective lens relative to the sample to obtain an optimal focus. Preferably the imaging light spot is focused at the determined vertical position of the object in the sample. A high throughput may thus be achieved. The depth profiles may be generated at a rate larger than the highest frequency required for keeping the system in good focus. In the resulting intensity depth profiles, the positions of all interfaces of the sample, in particular both glass-to-fluid interfaces, relative to the current microscope focus position can be identified. The position of the tissue or cells relative to these interfaces can be identified as well, in case any such tissue or cells are present. The focus position of the objective lens relative to the sample can be adjusted accordingly to keep the system in best focus at all times. It is noted that above-mentioned US 7,071,451 B2 describes an autofocus system arranged so as to focus a beam on highly reflective interfaces. To the contrary, the system according to the present invention is arranged to focus the beam at any depth in the sample, including interfaces having a low reflectivity. An advantage of having a confocal arrangement is to prevent high intensity signals emanating from regions other than the focal region from reaching the detector. Detecting interfaces having a low reflectivity is thus facilitated. The microscope may use software for controlling the autofocus optics. The light from a dedicated light source (e.g. laser) may be focused to one or more near-diffraction limited spots inside the sample.

The microscope may be configured such that the sample is substantially immobile during the autofocus interval. Herein it is understood that the positions of the sample that are probed by the imaging light spot during the autofocus interval are separated by a distance shorter than the resolution of the microscope.

Or the microscope may be configured such that the sample is translated in the horizontal plane at a substantially constant speed during the autofocus interval. The sample may be moved horizontally by a distance larger than the resolution of the microscope. However, the vertical positions of the interfaces and tissues may be essentially constant during the autofocus interval.

The sample may comprise a substrate layer, a specimen layer containing organic matter, and a cover layer; wherein the organic matter is the object in the sample. The specimen layer may be liquid or solid, or partly liquid and partly solid. For example, the specimen layer may comprise a block of paraffin in which biological material has been embedded. A depth profile may be obtained by actuating a lens in the autofocus light path, or by using a grating to generate a linear array of laser foci at different depths inside the sample. The controller may be configured for determining from the detector signal a vertical position of the substrate layer and a vertical position of the cover layer. It may further be configured for determining any of the interfaces between the following successive layers: upper air or vacuum, cover layer (e.g. of glass of plastic), specimen layer, substrate layer (e.g. of glass or plastic), lower air or vacuum. The microscope may be configured for generating the autofocus light spots during the autofocus interval simultaneously. The auto focus optics may comprise a set of pinholes, each of the auto focus lights spots having associated with it one of the pinholes in a confocal arrangement, that pinhole providing the confocal aperture. The light reflected from the sample may thus be imaged in a confocal manner onto the one or more pinholes. A photodetector may be placed behind each of the pinholes. The confocal arrangement with the one or more pinholes may provide a high depth resolution as the auto focus light reflected from depths different from the depth at which the corresponding autofocus spot is generated are effectively filtered out.

The autofocus optics may comprise an optical element for generating a set of beamlets. More specifically, the autofocus optics may comprise an element for generating a set of beamlets from an incident beam, for example, a diffractive element or an array of micro lenses.

The microscope may be configured for generating the autofocus light spots during the autofocus interval successively. The autofocus optics may comprise a movable lens; and a lens actuator for shifting the movable lens on its optical axis.

The autofocus optics may further comprise a mirror, the mirror being arranged such that the autofocus light traverses the movable lens at least twice on its way from the sample and/or on its way to the sample. Alternatively or additionally to moving the movable lens, the mirror may be shifted such that the autofocus light traverses the movable lens at least twice on its way from the sample and/or on its way to the sample.

The autofocus optics may comprise a single pinhole, the single pinhole providing the confocal aperture for each of the autofocus light spots. Each of the autofocus light spots may have associated with it an optimal pinhole position and wherein the autofocus optics comprises a mechanism for shifting the single pinhole to any of the optimal pinhole positions associated with the autofocus light spots.

The imaging optics and the autofocus optics may share a common objective lens. Thus a compact and economic design may be achieved. The autofocus light path may be coupled into the main imaging light path via a dichroic mirror. Alternatively, a separate objective lens may be used for autofocus so that the imaging and autofocus light paths are completely separated. The auto focus light may be in the infrared. Thus disturbing the normal image acquisition may be avoided.

The microscope may be configured for performing the following steps in a cyclic manner: - generate the autofocus light spots; receive the detector signal; derive from the detector signal the position of the object; move the sample to the desired position; generate the imaging light spot; - translate the sample relative to the imaging light spot in the horizontal plane.

Brief description of the drawings

Figure 1 provides a simplified cross-sectional view of an example of a microscope slide;

Figure 2 schematically illustrates a scanning microscope according to a first embodiment. Figure 3 schematically illustrates a scanning microscope according to a second embodiment. Figure 4 schematically illustrates a scanning microscope according to a third embodiment. Figure 5 schematically illustrates a scanning microscope according to a fourth embodiment.

Figure 6 shows a flow chart of an example of an autofocus method.

Description of preferred embodiments

Unless specified otherwise, identical or similar reference numerals appearing in different Figures label identical or similar components. Schematically illustrated in Figure 2 is a microscope 10 according to a first embodiment. A beam of infrared light 38 (autofocus light) is generated by a laser diode 36. The infrared light 38 is transmitted by a polarizing beam splitter 58 and collimated by a movable lens 60. A λ/4 plate 66, located either in front of or behind the movable lens 60, converts the infrared light 38 from linear to circular polarization. The infrared light 38 is subsequently coupled into a main imaging light path of the microscope 10 by reflection on a dichroic mirror 72. The mirror 72 is chosen such that it transmits visible light 42 but reflects the infrared light 38. Both the visible light 42 of the main imaging path and the infrared light 38 are focused into the sample 12 by the objective lens 54. Both the visible light 42 and the infrared light are reflected by various features and interfaces inside the sample 12 and collected by the same objective lens 54. The reflected infrared light is labeled 40 in the Figure. The reflected infrared light 40 travels back along the same path until it reaches again the polarizing beam splitter 58. Since it has made a second pass through the λ/4 plate 66 it is again linearly polarized but has a polarization perpendicular to the polarization of the incoming infrared light 38. The infrared light 38 is therefore reflected by the polarizing beam splitter 58 and directed towards a pinhole 50 centred on a light spot 48 formed by the reflected infrared light 40. The infrared light 40 that is transmitted by the pinhole 50 is detected by a photodetector 52. The movable lens 60 can be shifted in a direction 62 along its optical axis by a lens actuator 64. The central position of the movable lens 60 is ideally such that the infrared light 38 in the main imaging path (between elements 56 and 54) is exactly collimated and the position of the infrared light spot 46 in the sample 12 coincides with the position of the imaging light spot 44 of the imaging system. Shifting the movable lens 60 along its optical axis around its central position causes the IR light spot 46 in the sample 12 to shift in depth (i.e. parallel to the z axis 6) relative to the imaging light spot 44. While the movable lens 60 is being shifted, a peak in a detector signal delivered by the photodetector 52 corresponds to the presence of an interface 26, 28, 30, or 32, or tissue 22 (see Figure 1) at the current position of the infrared light spot 46. Information on the relative positions of the glass-to-fluid interfaces 26, 28, 30, 32, and especially on the position of the tissue 22 relative to the current depth position of the imaging light spot 44 is used to control the position of the objective lens 54 relative to the sample 12.

Schematically shown in Figure 3 is a variant to the basic embodiment described above with reference to Figure 2. An additional lens 70 behind the beam splitter 58 creates an intermediate focus 71. The IR light 38 is re-collimated by the joint action of the movable lens 60 and a further additional lens 72. This arrangement allows the selection of a movable lens 60 with a much smaller diameter than is required in the basic embodiment described above with reference to Figure 2. Also the range over which the movable lens 60 needs to travel to produce a given shift of the infrared focus 46 in the sample 12 can be much smaller than in case of the basic embodiment of Figure 2, as the numerical aperture of the beam of infrared light 38 focused by the movable lens 60 can be increased. The travel range Ay _traV ei corresponding to a focus shift Azf _0CUS is given by:

ky travel where n is the refractive index of the specimen layer 16 (see Fig. 1), NA _o b is the numerical aperture of the objective lens 54, and NA _1n is the numerical aperture of the IR beam 38 that is focused by the lens 60. This latter numerical aperture NA _1n can be larger in the present embodiment as compared to the one discussed with reference to Figure 2 and hence the travel range Ay _traV ei can be reduced. Both modifications make it easier to actuate the movable lens 60 at high frequencies. Furthermore, two additional lenses 68 and 74 have been introduced between the laser diode 36 and the beam splitter 58, and between the beam splitter 58 and the pinhole 50, respectively. This allows the incoming light beam 38 and the reflected light beam 40 to be substantially collimated when passing through the polarizing beam splitter 58, for a more efficient performance of the beam splitter 58. Figure 4 illustrates, in a schematic manner, another variation to the basic embodiment discussed above with reference to Figure 2. In this embodiment the microscope 10 comprises a second beamsplitter 78. The second beamsplitter 78 is a polarizing beamsplitter, whereas the first beamsplitter 58 is now a normal non- polarizing beamsplitter, having for example a 50/50 split in transmission and reflection for both polarizations. The actuated lens 60 focuses the infrared light 38 on a mirror 80. The lens 72 shown in Figure 3 has been replaced by a telescope lens set 75 and 76 to match the diameter of the beam of infrared light 38 to the pupil size of the objective lens 54. Infrared light from the laser 36 and (partially) transmitted by the first beamsplitter 58 is (fully) transmitted by the second beamsplitter 78, passes through the λ/4-plate 66, and is focused by the movable lens 60 onto the mirror 80. The IR light 38 reflected by the mirror 80 is collimated by the movable lens 60, passes through the λ/4-plate 66 again and is (fully) reflected by the second beamsplitter 78. The telescope arrangement 75, 76 modifies the diameter of the beam 38 to match the pupil of the objective lens 54. The dichroic mirror 56 directs the beam of IR light 38 to the objective lens 54 which focuses IR light 38 into the sample 12. Backscattered infrared light 40 is collected by the objective lens 54 and retraces the path of the incident beam 38. The first beamsplitter 58 (partially) reflects the infrared light 40 into the confocal detection branch 74, 50, 52. A key advantage of this embodiment is the reduction of the travel range of the movable lens 60 by a factor of two, the lens being operated in reflection. A further advantage is that alignment requirements in the lateral direction of both the movable lens 60 and the lens 70 in the previous embodiment are reduced, if not eliminated. In the previous embodiment, the required accuracy is of the order of the field of the lens 70, which decreases with increasing numerical aperture NA _1n of the movable lens 60. As a consequence the numerical aperture NA _1n of the movable lens 60 can now be increased, which further reduces the required travel range of the movable lens 60. For example, an off-the shelf CD-actuator has an objective lens with NA _1n =

0.45, which gives for a typical objective lens (NA _o b ⁼ 0.75) and a sample with refractive index n = 1.5 a travel range that is about equal (up to a few percent) to the required focus shift.

The embodiment of Figure 3 has been built and tested in a lab setup. In this setup a pathology slide can be scanned in a lateral direction (i.e., parallel to the x y plane 4) relative to the imaging optical axis. At the same time the movable lens 60 can be actuated. Subsequent depth profiles of the sample reflectivity can be represented in the form of 2D maps similar to the representation of the sample 12 in Fig. 1. The length of the lateral scan was 2 mm; the depth profile corresponded to a depth range of 93 μm inside the sample 12. The two glass-to-fluid interfaces 28, 30 could be clearly identified as well as some tissue 22.

A fourth embodiment of the invention is represented schematically in Figure 5. The microscope 10 according to this embodiment does not comprise any moving parts in the auto focus part of the light path. Instead, the IR light 38 from an infrared laser diode 36 passes through a transmission grating 82 to generate a plurality of IR beamlets. The transmission grating 82 has concentric circular lines that together form a small off- axis part of a Fresnel zone plate. This ensures that the beamlets corresponding to the various diffraction orders have different directions as well as different amounts of divergence. The beamlets pass through a λ/4 plate 66 and are coupled into the main imaging light path by reflection off the dichroic mirror 56. The beamlets are focused to form light spots 46, 47 in the sample 12 by the objective lens 54. Only two IR light spots 46, 47 are represented in the Figure; in practice there are many more. Because of the differences in direction and divergence of the beamlets, the foci in the sample 12 are displaced relative to each other both laterally (i.e. in the x and y directions 4) and in depth (i.e. in the z direction 6). The mutual lateral and depth separations between neighbouring laser foci 46, 47 can be tuned by carefully choosing the values of the parameters of the transmission grating 82. Reflected IR light 40 from the sample 12 is collimated by the objective lens 54 and travels back along the same light path up to the polarizing beam splitter 58. It is reflected by the polarizing beam splitter 58 and focused onto a linear array of pinholes 50, 51. The array of pinholes 50, 51 is tilted relative to a plane perpendicular to the optical axis, such that each of the laser foci 46, 47 inside the sample 12 is imaged confocally onto one pinhole of the array 50, 51. Closely behind each pinhole 50, 51 there is a photo-detector 52, 53 that detects the light 40, 41 that is transmitted by the respective pinhole. Each of the photodetectors 52, 53 corresponds to a fixed depth inside the sample 12 relative to the current main imaging focus 44. A large signal from a given photodetector, e.g. from photodetector 52, occurs if one of the glass-to-fluid interfaces 28, 30 or tissue 22 (see Figure 1) is present near that depth in the sample 12. As in the embodiment described above with reference to Figure 2, information on the relative positions of the glass-to-fluid interfaces 28, 30 and especially on the relative position of the tissue 22 relative to the current focus position 44 of the main imaging system is used to control the position of the objective lens 54 relative to the sample 12.

In a further embodiment (not shown), the line arrays of pinholes 50, 51 and the photodetectors 52, 53 may be replaced by a pixelated line sensor, in which the function of the pinholes is provided by the limited extent of the individual pixels in the array. In yet another embodiment the microscope 10 comprises a 2D grating, for generating a two-dimensional array of spots, in order to detect the position of the organic matter 24 at a large number of positions. The infrared light reflected from the sample is then imaged with a 2D pixelated detector, again with the pixels acting as pinholes.

Referring now to Figure 6, there is illustrated a flow chart of a method of focusing a scanning microscope, the scanning microscope comprising an autofocus branch and a main imaging branch. The scanning microscope may be one of the microscopes described above with reference to Figures 1 — 5. In a first step 601, a lookup table is generated in order to calibrate a focusing mechanism. The lookup table may, for example, indicate for a set of positions of a movable optical element of the autofocus branch a corresponding set of positions of a movable optical element of the main imaging branch, or a corresponding set of vertical positions of a scan stage.

Alternatively the lookup table may indicate for each of a set of autofocus light spots a corresponding position of a movable element of the main imaging branch or a vertical position of the scan stage. Effectively the lookup table relates a coordinate system fixed to the autofocus branch to a coordinate system fixed to the main imaging branch. The lookup table may be stored in a digital storage element of the microscope. Using the lookup table, an imaging light spot generated in the sample under study via the main imaging branch may be positioned at a desired position determined by the autofocus branch. It may be sufficient to generate the lookup table only once, for example, when the microscope is manufactured. However, the microscope may be designed so as to allow for re-calibrating the focusing mechanism. In a subsequent step 602, the sample is shifted horizontally to a new x y position. At this position, a depth profile (i.e. a profile in the z direction) of the sample is generated by means of the autofocus branch. Autofocus light, which may be in the infrared, is used to generate a plurality of autofocus light spots, the autofocus light spots being distributed over different depths of the sample, i.e. having different z coordinates. Autofocus light reflected by the sample at the autofocus light spots is received by a detector to generate an intensity signal indicating an intensity of the reflected autofocus light for each of the autofocus light spots. Thus a depth profile of the sample is obtained. If the sample comprises a substrate glass, a specimen layer, and a cover glass, vertical positions of the interfaces between these layers (substrate glass to specimen and specimen to cover glass) are derived from the depth profile. The depth profile generally depends on the current x y position of the sample. This is in particular due to the fact that the substrate glass and the cover are generally not exactly parallel to each other or to the x y plane. Furthermore, a vertical position of organic matter contained in the specimen layer is derived from the depth profile. Analyzing the depth profile may comprise analyzing an intensity and/or an intensity gradient along the vertical direction, and possibly detecting peaks in the intensity and/or in the intensity gradient. In a subsequent step 605 the imaging light spot is placed at the position of the organic matter as determined from the depth profile. The step 605 comprises consulting the lookup table. The imaging light spot may be shifted relative to the sample by moving a scan stage of the microscope, the sample being held by the scan stage, or by moving at least one optical element on the light path of the imaging light. Transmitted, scattered, or fluorescent light emanating from the imaging light spot is detected in reflection or in transmission. The process then returns to step 602 to image another point of the sample.

In summary, the introduction of digital pathology in hospitals requires the ability to scan and digitize microscope slides at high speed. Automated digitization of slides requires an automated method to determine the best focus position of the imaging system at each point in the scan. The solution proposed here is to continuously track the position of two glass interfaces in the sample and the position of the tissue relative to those interfaces, during the image acquisition. This is done by confocally acquiring information from various depths in the sample (quasi-)simultaneously. To this end, the auto focus branch of the system may comprise an actuated lens. In a particular embodiment, an array of laser spots is generated in the sample at different depths. The invention thus provides a means of fast continuous auto focus in a digitizing scanning optical microscope for cell and tissue pathology applications. It can also be used to provide autofocus in any other type of scanning optical microscope. While the invention has been illustrated and described in detail in the drawings and in the foregoing description, the drawings and the description are to be considered exemplary and not restrictive. The invention is not limited to the disclosed embodiments. Equivalents, combinations, and modifications not described above may also be realized without departing from the scope of the invention. For example, according to the level of depth resolution desired, it may be possible to use a detector without a confocal configuration. For example, it is possible to use a configuration wherein the detector in the autofocus path receives autofocus light spots having different vertical depths, while the autofocus optics in said path generate an autofocus light spot at only one single vertical depth in the sample.

The verb "to comprise" and its derivatives do not exclude the presence of other steps or elements in the matter the "comprise" refers to. The indefinite article "a" or "an" does not exclude a plurality of the subjects the article refers to. Adjectives describing locations or directions, such as "top", "bottom", "horizontal", and "vertical", indicate the arrangement of components relative to each other and have no absolute meaning. It is also noted that a single unit may provide the functions of several means mentioned in the claims. The mere fact that certain features are recited in mutually different dependent claims does not indicate that a combination of these features cannot be used to advantage. Any reference signs in the claims should not be construed as limiting the scope.