GB2355354A - Auto-focus method - Google Patents

Abstract

A 'back-projecting' method for automatically focusing a reflective surface, for example in viewing a specimen on a microscope slide, comprises directing a plurality of (for example, three) illumination beams 64 through beamsplitter 14, and intermediate optics 20 towards the interface between a sample to be viewed and specimen plate 26. Beams 30, reflected at the interface 26, return via the optics to produce (in this example) three corresponding images on the surface of detector 28. An accurate position determination of the substrate surface with respect to the focal plane of the instrument can then be made by analysis of the spatial distribution of the reflected light beams (Figs 5A-C), where relevant parameters may include the maximum reflected value and the full-width at half maximum of the irradiance distributions at a series of axial positions of the interface with respect to the focal point of the lens. A device for implementing the method, and a similar method of auto-focusing using only a single illuminating beam are also disclosed in the specification. The method may also be useful in camera auto-focusing systems.

Description

2355354 BACK-PROJECTED BEAM FOR AUTOMATICALLY FOCUSING A MICROSCOPE

DETECTOR ONTO A SAMPLE INTERFACE SURFACE

RELATED APPLICATIONS

This application is based on U.S. Provisional Application Serial No. 60/147,052, filed on August 3, 1999.

BACKGROUND OF THE INVENTION

Field of the Invention

This invention is related in general to the field of electronic cameras and microscopes for studying biological specimens supported on a carrying substrate. In particular, the invention concerns an automatic focusing apparatus and procedure based on the detection and analysis of reflections of one or more beams of light back- projected from the direction of the detector toward the specimen substrate.

Description of the Related Art

The study of biological cells on a microscope involves placing the specimen on the surface of a carrying substrate such as a glass slide or coverslip, or the glass or clear- plastic bottom of a multi-well plate-type specimen carrier. In many such applications, including time-lapse imaging of live cells and serial screening of multiple wells, thermal drift in the microscope or specimen carrier, as well as manufacturing imperfections in the specimen carrier, affect the measurement and necessitate periodic refocusing of the specimen to obtain clear images. Because refocusing may be required very frequently, it is desirable to accomplish it quickly, as well as accurately and reliably.

Some prior-art microscopes include an automatic focusing 2 function derived from an electronic focus signal related to the distance between the target and the microscope objective lens. This distance is adjusted to produce an in-focus position. Other auto-focus microscopes measure the white-light contrast in the specimen image seen in a camera and determine the in-focus position on the basis of maximum contrast. U.S. Patents No. 4,595,271, No. 4,931,630, No. 5,228,987, No. 5,317,142, No. 5, 483,079, No. 5,530,237, No. 5,672,861, No. 5,714,749 and No.

5,925,874 describe several different approaches for producing an automated focusing feature in microscopes used for repetitive, sequential work, such as in biological applications. In all cases, repeatability, quickness and accuracy are the most important goals of the auto-focus apparatus.

In U.S. Patent No. 5,783,814, Fairley et al. disclose a microscope system that generates an electronic focus signal related to the magnitude of the light reflected from the target sample and recorded on the microscope's detector. An automated focus function is derived from the focus signal by performing a scan through focus along the optical axis of the microscope and identifying the position corresponding to the largest value of the signal recorded at the detector. Various algorithms are disclosed to determine the brightest layer of the target imaged through the scan. In alternative embodiments, the fixed-point reflection signal received at the detector during the axial scan is substituted with the largest value recorded by either a line or an area scan performed over the surface of the target at each axial position.

The back-projected illumination approach disclosed in U.S. Patent No. 5, 783,814 enables the simultaneous imaging of the target for auto-focus as well as for testing purposes, allowing for direct and relatively rapid identification of the in-focus position. an the other hand, because each 3 detector measurement is based on a single measure of illumination received at a f ixed point on the detector, its accuracy depends on having a sample surf ace sufficiently uniform to be all contained within the focal depth of the microscope. Alternatively, to ensure that all points on the sample surface are at a best in-focus position, the light source must be moved to scan through a line or an area of the sample surface, which complicates the system and materially reduces the speed of auto-focus operation.

The present invention is based on a back-projected illumination system that overcomes these shortcomings. In addition, the invention provides alternative illumination schemes for improving the precision and the speed of the auto-focus system.

4 BRIEF SUMMARY OF THE INVENTION

The primary objective of this invention is a fast and reliable auto-focus approach suitable for microscope applications, such as for sequential measurements of biological specimens contained in multi-well plates, including time-lapse imaging of live cells.

Another important goal of the invention is a method and an apparatus that can be used to ascertain the correct seating of each well plate in its holder and/or to detect plate-to-plate variations during a particular experimental run.

Another goal is a procedure that can be carried out during and in conjunction with normal testing of the samples.

Another object is an approach suitable for automated implementation.

Another objective is a procedure that can be implemented with very few data points, possibly with a single measurement, dedicated to the autofocus function.

Still another objective is a method and apparatus that are suitable for incorporation within existing instruments.

A final objective is a procedure that can be implemented easily and economically according to the above stated criteria.

Therefore, according to these and other objectives, the preferred embodiment of the present invention includes a special illuminating light source to project structured light toward the specimen/substrate interface in the direction opposite to the image forming beam. That is, light is back-projected toward the specimen substrate from the direction of the detector using a beamsplitter and other dedicated optics. One or more beams may be backprojected either simultaneously or sequentially, and multiple beams may differ in their lateral position with respect to the optical axis of the microscope and/or in the position of their focal point with respect to the microscope's in-focus plane. An actuator is used for varying the position of the sample substrate with reference to the instrument's focal plane along the microscope's optical axis.

According to one aspect of the invention, an analysis of the detected reflections from the back-projected beam or beams allows the fast, accurate and reliable positioning of the substrate surface in focus based on the reflection of the back- projected light toward the detector surface. The two-dimensional distribution of the reflected light detected through normal operation of the electronic detector provides information about the precise position of the substrate surface. The in-focus position is identified as the position producing the maximum peak value of intensity back-projected from the sample.

According to another aspect of the invention, its reliability is enhanced by using multiple light beams of back-projected light and/or by displacing some of the beams either laterally or axially, or both. Multiple, laterally displaced beams provide information about the condition of the specimen surface and system alignment.

Multiple, axially displaced beams provide information about the position of the specimen surface relative to the objective lens' focal plane. Thus, a combination of the two can be used for quality control as well as for autofocus purposes. In some cases, the entire set of measurements needed for automatic focusing may be performed through a single exposure and image acquisition operation. Finally, the invention includes a method for 6 automatic calibration of the entire system for optimal operation with a given objective lens and type of substrate and specimen medium.

Various other purposes and advantages of the invention will become clear from its description in the specification that follows and from the novel features particularly pointed out in the appended claims. Therefore, to the accomplishment of the objectives described above, this invention consists of the features hereinafter illustrated in the drawings, fully described in the detailed description of the preferred embodiment and particularly pointed out in the claims. However, such drawings and description disclose but one of the various ways in which the invention may be practiced.

7 BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1A is a schematic representation of an embodiment of the backprojection illumination and detection apparatus 5 of the invention.

Fig. 1B illustrates the paths of the illumination and detection beams produced by the system of Fig. IA when the sample surface of interest is in an above-focus position.

Fig. IC illustrates the paths of the illumination and detection beams produced by the system of Fig. 1A when the sample surface of interest is in an in-focus position.

Fig. 1D illustrates the paths of the illumination and detection beams produced by the system of Fig. 1A when the sample surface of interest is in a below-focus position.

Fig. 2A is an illustration of the light intensity distribution sensed at the surface of a two-dimensional multi-element detector for each condition shown in Figs. IB-D. The spatial distribution of the profile over the detector represent intensities measured at different detector pixels along an axis intersecting and perpendicular to the axis of the incident beam.

Fig. 2B is an illustration of how the peak intensity and the full-width at half-maximum of the profiles of Fig. 2A vary with the distance of the surface of interest from the focal plane of the instrument.

Fig. 3A is a schematic representation of an embodiment of the invention utilizing multiple, redundant back-projected illumination beams to enhance the reliability with which the position of the surface of interest with respect to the focal plane is determined.

a Fig. 3B illustrates the paths of the illumination and detection beams produced by the system of Fig. 3A when the sample surface of interest is in an in-focus position.

Fig. 4A is a schematic representation of an embodiment of the invention utilizing multiple, axially displaced backprojected illumination beams to reliably determine the position of the surface of interest with respect to the focal plane with a single exposure of the detector.

Fig. 4B illustrates the paths of the illumination and detection beams produced by the system of Fig. 4A when the sample surface of interest is in an in-focus position.

Fig. 5A is an illustration of the distinct intensity profiles for each axially displaced beam generated by the system of Fig. 4A at a single step during an auto-focus scan according to the invention. The condition shown here corresponds approximately to the focal position of Fig. 4.

Fig. 5B is an illustration of how the peak intensity values of profiles generated by the system of Fig. 4A can be used to determine the distance of the surface of interest from the focal plane of the instrument with a single measurement.

Fig. 5C is an illustration of how the full-width at halfmaximum profiles generated by the system of Fig. 4A can also be used to determine the distance of the surface of interest from the focal plane of the instrument with a single measurement.

Fig. 6 illustrates schematically the combination of two orthogonal beam multipliers to produce a two-dimensional beam array with the redundancy of the embodiment of Fig. 3A and the speed advantages of the embodiment of Fig. 4A.

9 Fig. 7A is a schematic representation of a simple beam multiplier constructed from three beam splitters.

Fig. 7B is a schematic representation of a simple offset beam multiplier constructed by adding three lenses differently positioned along the optical axis of each output beam.

Fig. 8 is a schematic representation of the invention where the back-projection beamsplitter replaces the window conventionally used to protect the CCD detector.

DESCRIPTION OF THE PREFERRED EMBODIMENTS OF THE INVENTION

In a wide range of applications, including many in light microscopy, the task of automatic focusing can be restated as the task of finding the surface of a substrate, such as a slide, coverslip or transparent well bottom to which the true specimen of interest is attached. Even if the specimen is not optimally focused precisely at the substrate surface, the point of optimal focus for the specimen is easily attained by moving some constant offset distance with respect to the optimal focus for the substrate surface. Therefore, focusing on the interface between specimen and substrate satisfies the function provided by an auto-focus device.

Optimum focus is defined for the purpose of this invention as the condition where the imaging of the sample onto the detector is optimized. This may or may not occur when the sample is coincident with the focal plane of the objective lens. For example, finite-conjugate microscope objective lenses operate by design with the sample displaced somewhat from the actual focal plane of the objective lens. This invention applies equally to this case as it does to cases where optimum imaging is obtained when the sample location is coincident with the focal plane of the lens (as found, for example, with infinite-conjugate microscope objective lenses).

The invention is based on the general concept of projecting light toward the specimen from the direction of the microscope's detector (commonly referred to in the art as "epi- illumination") and deriving information about the precise position of the substrate surface from the twodimensional distribution of the reflected light as it is detected during the normal operation of the detector. one or more beams may be back-projected either simultaneously or sequentially, and multiple beams may differ in the lateral and/or axial position of the light sources or images of the light sources.

It is noted that the embodiments of the present invention are described with reference to x, y and z orthogonal coordinates wherein x and y define a horizontal plane parallel to the light detector surface and z defines a vertical direction along which the optical axis of the microscope is aligned, but it is obvious that the structure and operation of the features detailed herein could be rotated in any direction with equivalent results. It is also understood that the various components shown in the drawings are not drawn to relative scale, but they are represented only schematically for illustration. For the purposes of this disclosure, a "reflective surface" is defined as any interface between two media with different (potentially complex) indices of refraction. Accordingly, such interfaces are intended to include those with a dielectric medium or a metal surface (such as in mirrors).

The term "back-projected" light is used throughout the disclosure to refer to epi-illumination in the context of the auto-focus device of the invention. The terms "multielement" and "multiple-element" detectors are used to refer to light detectors capable of measuring the spatial distribution of light intensity (irradiance distribution) received over the sensing surface; that is, detectors consisting of more than a single detection element, such as arrays of CCD and similar multi-element electronic detection devices, as opposed to single-element devices like photo- multiplier tubes. Finally, the terms "parfocal," "intrafocal" and "extrafocal" refer to light beams converging to a point coincident with, closer to the objective than, or further from the objective than the focal plane of the microscope objective lens, as defined by a corresponding image on the instrument's light detector.

12 In its simplest form, the invention is carried out using a single backprojected beam. Referring to the drawings, wherein like parts are designated throughout with like numerals and symbols, Fig. 1A illustrates the basic autofocus principle of the invention wherein a light source 10, such as a laser, produces a beam 12 directed in part by a beamsplitter 14 toward the back side of a samplecarrying substrate 16. It is noted that a diverging beam is illustrated, but a converging or collimated beam could be used in equivalent f ashion so long as focused at an appropriate design point. Because the beamsplitter reflects only a portion of the light (about 1- 10%), the remaining unreflected li ght is eliminated by a beam dump absorber 18. A low-level reflectance is generally desired, as the beamsplitter is also present in the conventional imaging path, where maximum transmission is important. It is also possible, as discussed below, to use a beamsplitter with spectral characteristics that allow implementation of the autofocus function in a spectral range that is otherwise unimportant to the operation of the instrument.

An objective lens 20 is used to focus the back-projected beam 22 on a spot 24 at a selected objective lens' design focal plane, such as the specimen plane 26. A preferably motorized actuator M is provided to change the position of the substrate along the optical axis of the instrument and implement the auto-focus function of the invention.

Equivalently, all or part of the objective lens 20, the source 10, and other microscope optics may be translated along the optical axis to implement the autofocus function of the invention.

As illustrated in Figs. 1B, 1C and ID, the back-projected light 22 is reflected at the substrate surface in a way that depends on the position of the surface 26 along the 13 optical axis of the microscope (that is, the distance between the point 24 and the lens 30 along the optical path of the instrument). As one skilled in the art would readily understand, reflections occur at interfaces between media with different refractive indices (known in the art as Fresnel reflections). The reflected light is transmitted through the optics 20 toward an electronic camera 28 (such as one based on a CCD or other 2dimensional multi-element electronic detection array).

For clarity, the path of the back-projected beam 22, also referred to herein as illumination or epi-illumination beam, is illustrated in dotted lines in the figures, while the path of the reflected beam 30, also called detection beam, is shown in solid lines.

The figures illustrate how information about the position of the substrate surface 26 relative to focus is available from the two- dimensional image pattern of the reflected detection beam 30 reaching the surface of the detector 28.

As shown unidimensionally in Figs. 1B-D, the diameter 32 of the illuminated spot 24 on the surface 26 (see Fig. 1B) varies and reaches a minimum when the reflecting surface 26 is at the design specimen plane (i. e., the design focal distance, as illustrated in Fig. 1C). While light may also be reflected from other optical surfaces (including the lower substrate surface 34), the relatively greater distances of those surfaces from the design specimen plane 26 typically ensure that this light is very broadly defocused at the detector, and thus it is readily neglected as a low-intensity and relatively constant background.

Figs. 2A and 2B illustrate how information about the focal condition of the instrument can be extracted according to the invention from images captured at the detector array 28 positioned as indicated in Figs. IA-D. Since the irradiance patterns produced by the detection beam 30 on 14 the surface of the detector 28 correspond to reflections from the in- focus or near-focal positions of the spot 24, the patterns will tend to be circularly symmetric about the incident beam axis. Therefore, the radial variations of these patterns can be analyzed simply and equivalently by measuring the irradiance variation along any diameter of the circular pattern. Such irradiance patterns at the detector surface are illustrated in Fig. 2 as intensity profiles (also referred to as irradiance distributions or patterns) along a single transverse axis on the detector (note that other profiles of spatial distribution of intensity measured across the detector surface could be used as well). Specifically, Fig. 2A shows three profiles of the light intensity distribution of the detection beam along an axis intersecting and perpendicular to the axis of the incident detection beam 30 (that is, a profile taken along a diameter of the circular light pattern on the surface of the detector 28).

As illustrated in Fig. IC and readily apparent to one skilled in the art, the exact in-focus position of the substrate surface 26 will produce a near point-like image 36 on the detector 28. Accordingly, the corresponding intensity profile (see curve 38 in Fig. 2A) will feature a sharp peak 40. On the other hand, the out-of-focus positions illustrated in Figs. 1B and 1D will produce profiles 42 and 44, respectively, that reflect the greater diameter 32 of the spot 24 formed on the surface 26. (Profile 42 is shown as broader than profile 44 for demonstrative purposes only, rather than as a means for distinguishing the above-focus and below-focus conditions.) In turn, correspondingly circular images with diameters 46 and 48 are formed on the detector surf ace by the detection beam 30. As seen in Fig. 2A, the intensity profiles 42 and 44 along each image's diameter will have much less pronounced peaks 50 and 52, respectively, thereby providing a means for readily ascertaining the correct focal position of the substrate surface.

For instance, it is clear from Fig. 2A that the in-f ocus position of the specimen surface 26 corresponds to the maximum "peak intensity" registered during a scan along the optical axis of the microscope from the intensityprofile data measured by a multiple-element detector along a diameter of the detection image. Similarly, a plot of the width of the detection image corresponding to half of its maximum intensity, normally referred to in the art as the "full-width at half- maximum," as a function of translation along the optical axis will show a minimum at the in-focus position. While full-width at half-maximum is being used as a criterion for descriptive purposes, the technique can be equally applied using other ratios or portions of the width and height of the irradiance distribution at the detector. Thus, the minimum of any parameter related to the width or the maximum of any parameter related to height of the irradiance patterns can be used equivalently to practice the invention. Representative curves of these analytical tools are illustrated in Fig. 2B, where the inflection point of each curve can be detected and readily used to ascertain the perfect in-focus position of the substrate interface surface 26. Thus, the peak-intensity curve 54 will have a maximum inflection 56 at a zero distance z from the focal position; correspondingly, the full -width-at- hal f -maximum curve 58 will have a minimum inflection point 60.

Given the ability to extract focus measures like those illustrated in Figs. 2A and B and to control a f ocus axis actuator, various different strategies may be used to set f ocus precisely with reference to a substrate surf ace. By interactive movements of the specimen plane through focus and capture of corresponding images formed on a multipleelement detector, one can simply f ollow the gradient of a 16 focus measure (such as either curve 54 or 58) to an inflection point as shown in Fig. 2B. More efficient strategies can be developed based on empirical calibration data (as discussed below) or analytical approximation.

Based on such calibration or approximation criteria, inflection points of focus-measure curves can be determined with a minimal number of actual measurements of reflected light. For instance, since it is known that both curves 54 and 58 in Fig. 2B are approximately parabolic near their inflection point (i.e., at z=0), any three measured points near inflection suffice to determine a parabola that would in turn predict the exact inflection point. The curves shown in Fig. 2B are for descriptive purposes, and are not meant to precisely describe the distribution of light in the region about focus. Such distributions are well known (see Born and Wolf, "Principles of Optics," Pergamon Press, 6th Ed., 1980).

With sufficient calibration data and under optimally repeatable conditions, it would be possible to find the in-focus position of the substrate surface with measurements taken from even fewer than three focal-axis positions. Since defocus of the reflected spot will generally occur asymmetrically with respect to deviations from focus above and below the substrate surface, accurate calibrations and reasonably constant conditions may allow the exact focal position to be determined from careful analysis of just one single image captured at a single z position. For instance, even though the curves 54 and 58 appear symmetrical in the illustration of Fig. 2B, they in fact would always show a certain degree of asymmetry depending on materials and other system parameters such as aberrations in the objective lens 20. Accordingly, each curve may be skewed in distinctly different ways with respect to its inflection point, thereby providing sufficient information, with appropriate calibration for a given system and specimen, to determine the direction as 17 well as the absolute distance from surface focus with a single image- capture operation. (This is plausible, for example, if curves 54 and 58 are asymmetric in such a way that each pair of values on the curves corresponds to a 5 single focal position.) According to another embodiment of the invention illustrated in Figs. 3A and 3B, the reliability of the focusing procedure can be enhanced by using multiple back- projected beams. Using a beam multiplier 62 or other equivalent apparatus (such as multiple laser sources), multiple parallel light beams 64 are focused on a single specimen plane 26 through a common beamsplitter 14 and appropriate optics 20 to provide redundancy of measurement. Thus, the system is made more tolerant of localized defects or irregularities in the substrate surface or adjacent specimen. Using the same dotted-line and solid-line convention described above, Fig. 3B illustrates the formation of three point-like spots on the substrate interface surface 26 and correspondingly of three point-like images on the surface of the detector 28 when the surface 26 is in focus. It is clear that Figs. 3A and B illustrate the use of three beams, but a different number of beams could be used in equivalent fashion.

With the arrangement shown in Figs. 3A-B, the three reflected beams 30 should yield approximately identical local irradiance patterns. Therefore, the similarity of such multiple patterns with respect to a single image can be used as a quality-control parameter for the focusing operation. As few as a single beam can be used to identify the in-focus position of the specimen plane by the maximum-peak or the full-width at half-maximum methods, and the rest can be utilized advantageously to identify misalignments or other measurement problems. Deviations could signify that the substrate is warped or 18 angled with respect to the desired normal specimen plane; or they could indicate the presence of localized optical interference between the substrate surface and the specimen or contamination materials. Once such anomalies are detected, detrimental effects on the focusing process can be minimized, or slower but more robust focusing strategies can be implemented.

According to yet another embodiment of the invention, the speed of the procedure can be increased by using multiple, axially displaced parallel beams. As illustrated in Fig. 4A, an offset beam multiplier 66 can be utilized to produce three illuminating beams 68, 70 and 72, two of which (68 and 70) are axially displaced from the focal plane of the system. Accordingly, the focal points of beams 68 and 70 are extrafocal and intrafocal, respectively, with respect to the focal point of the parfocal position of the main measurement beam 72. As a result, the three spots formed by the illumination beams on the substrate surface 26 will vary in size and produce correspondingly distinctly different images on the detector 28, as illustrated in Fig. 4B for the in-focus case. Based on these images, a fast, robust and accurate determination of substratesurface position can be made with a single exposure and readout of the detector array.

Figs. 5A and 5B illustrate how an analysis of a detector image representing the reflection of three axially displaced beams provides unambiguous information about the substrate surface position over a substantial axial range. At each step during the auto-focus scan along the optical axis of the microscope, three distinct intensity profiles are measured, as illustrated by curves 74, 76 and 78 in Fig. 5A for the case when the specimen plane 26 is in focus with respect to the parfocal beam. Thus, a peakintensity plot as a function of displacement from 19 substrate- surface focus can be empirically derived for the extrafocal, intrafocal and parfocal beams 68, 70 and 72, respectively, as shown in Fig. 5B, and used to identify the z displacement of the substrate surface 26 from its in-focus position. For example, a measurement of the peak intensity of each beam, taken from a single detector frame, would provide three values, such as illustrated by points 80, 82 and 84 in Fig. 5B, that could be used to identify the focal displacement 86 of the surface 26 from its in-focus position. That is, the values of the three measures indicated in Fig. 5B for a single surface position (i.e., at a single displacement from focal position) provide a unique and redundant indication of the surface's displacement from its perfect in-focus position.

Full -width- at - half -maximum data can be similarly used, as shown in Fig. 5C, to place the system in focus with one measurement. After empirically deriving curves 88, 90 and 92 for the intrafocal, extrafocal and parfocal beams, respectively, a measurement of the full-width at halfmaximum for each beam, taken from a single detector frame, would provide three values, such as illustrated by points 94, 96 and 98 in Fig. 5C, that could be used to identify the focal displacement 100 of the surface 26 from its infocus position.

Figs. 4 and 5 show beams that have been laterally, as well as axially displaced. The technique of axially displaced beams may also be applied without lateral displacement. While the reflected beams will overlap at the detector, it will still be possible to distinguish the optimum focal position, albeit with somewhat greater difficulty than in the case shown here.

Obviously, the correctness of such a single -measurement focusing operation depends on constant or predictable system parameters and on the proper calibration of the instrument with a given type of specimen. For some applications, as few as two axially displaced beams might provide adequate focus precision and reliability enhancement. It is also understood that increasing the number and/or the total displacement of the beams could be used to increase the range of defocus distance over which the focusing procedure can be carried out reliably with a single detector measurement.

In yet another embodiment of the invention, the redundancy of light beams illustrated in Figs. 3A-B is combined with the axial displacement of beams shown in Figs. 4A-B to optimize both speed and reliability. As shown schematically in Fig. 6, the reliability advantages of multiple redundant beams can be combined with the speed advantages of multiple axially displaced beams by creating beam arrays with both redundancy and axial displacement. For example, a two-stage beam multiplier is obtained by first producing three beams laterally displaced along one specimen-plane axis (shown as the y axis in the figure) using an offset beam multiplier 66, and then by multiplying each laterally displaced beam along another specimen-plane axis (shown as the x axis) using a beam multiplier 62. Thus, a two-dimensional beam array is created as illustrated by the progression shown in boxes 102, 104 and 106 in Fig. 6. This particular example shows three redundant beams placed at each of three axial positions, for a total of nine beams, but other numbers of beams could be used as needed or preferable to achieve a desired level of reliability, speed and substrate defect tolerance.

Note that the procedure of the invention differs from the technique disclosed by Fairley et al. (U.S. Patent No. 5,783,814) in the fact that multiple intensity data points (i.e., the intensity profile across the sample surface) are recorded at each position during the axial translation of the microscope relative to the specimen plane. The 21 peak value of the profile, that is, the value recorded at the detector pixel that measured the highest intensity, is taken as the peak for that axial scan position, regardless of its location on the surface of the specimen. Thus, at each scan position the maximum light intensity seen by the detector is used to identify the focal plane.

Fairley et al., on the other hand, utilize a single-spot measurement representative of the light-intensity integral over the portion of the detector element illuminated by the reflected light. In essence, a total -illumination measure is used. In cases where the surface of the specimen is irregular, this can produce an in-focus position that is sub-optimal for parts of the specimen surface outside of the spot imaged during the auto-focus procedure. This shortcoming is in part remedied by scanning the light source over a line or an area of the specimen surface at each axial-scan position, and then using the highest reading so produced as the value for each position. This technique approaches the procedure of the present invention, but at considerably greater mechanical complexity and cost, and much slower data collection speed. In addition, the present invention is much less sensitive to misalignments and to intensity fluctuations in the light source.

In various embodiments of the invention, a beam multiplier is utilized to split the energy of one or more input beams into some larger number of output beams. The variant identified as an offset beam multiplier performs that function while also displacing the focal waist of an output beam by a predetermined amount along the optical axis. As such, these beam multipliers can be implemented in many alternative fashions simultaneously using static optical components; or using scanning components to generate beams in multiple positions sequentially; or using some combination of those two approaches.

22 Regardless of whether the multiple beams are generated simultaneously or sequentially, though, because of the detector's finite and often adjustable integration time, the resulting multiple reflections can be captured by a single exposure and image readout of the detector array.

Beam multiplication can be accomplished in simple fashion with static optical components. Partially reflecting beamsplitter surfaces on mirrors or prisms, microlens arrays, fiber optic components, one-dimensional or twodimensional gratings, and holograms may be used. In fact, a single hologram could provide for all of the functions of one- or two-dimensional beam multiplication and focus offset required for the invention. Suitable devices for a scanning approach to beam multiplication could include mirror galvanometers, other electromechanically activated devices, or acousto- optic deflectors.

A rudimentary implementation of a simple beam multiplier 62 is provided as an example in Fig. 7A. A two-thirds reflecting beamsplitter 108 is used to divert 66-0. of the energy of an input beam 110, thereby providing a first output beam 112. A one-half reflecting beamsplitter 114 is then used to divert 500s of the light it receives from the first beamsplitter 108 and provide a second output beam 116. Finally, a totally reflecting beamsplitter 118 (i.e., a mirror) is used to divert 1000s. of the light transmitted from the second beamsplitter 114 and provide a third output beam 120.

Similarly, Fig. 7B is an example of a simple implementation of an offset beam multiplier 66. The same combination of reflecting beamsplitters 108, 114,118 is utilized; in addition, corresponding output lenses 122,124,126 are used to provide the desired axial offset of the focal positions of the output beams 112,116,120. As one skilled in the art would readily understand, 23 relatively large axial displacements may be -required from the offset beam multiplier 66 because in the application of the invention it is positioned near a primary image plane, such that the axial magnification of the microscope equals the square of its nominal lateral magnification (e.g., the axial magnification is 100 with a 10x objective) Such displacements might be generated by differential placement of elements in an array of small lenses, as shown in Fig. 7B, or by having different beams pass through different thicknesses of plane glass following a single lens.

All of the embodiments of the invention have been described as using a beamsplitting surface 14 to place the back-projected illumination beam or beams 22 into alignment with the optical path leading in the opposite direction from the objective lens 20 to the camera's detector array 28. Accordingly, it may be convenient to include the beamsplitter 14 and all the other back- projection illuminating components within the housing of the system camera, as illustrated in Fig. 8. Solid-state sensors such as CCD arrays usually use a transparent window to protect the array from environmental contamination and perhaps to facilitate cooling of the array. By replacing the normal transparent window with a beamsplitter cube, as illustrated in the figure, it is possible to avoid any additional insertion loss in the detection pathway that might otherwise be associated with the beamsplitter placement.

A range of wavelengths of light can be used for the backprojected focusing beams of the invention. However, it may be advantageous and therefore preferable to choose a spectral band outside that of primary interest for normal operational imaging detection. Since the insertion of a focal position detection pathway may produce losses, they can be minimized by using a spectrally selective 24 beamsplitting surface within the beamsplitting cube 14. Thus, the partial reflections at the back-illumination wavelength necessary for auto-focus can be reduced to lower values over the spectral range of interest for operational detection. For optimal performance, the image quality at the autofocus wavelength must be good, and any axial chromatic aberration between the autofocus and operational wavelengths must be minimized or well understood.

The use of spectral filters in fluorescence microscopy may also provide some guidelines for preferred back-projection illumination wavelengths. For example, red or near infrared wavelengths for focusing may be most advantageous because high transmission between specimen and detector array is most common at these longer wavelengths. Presently available sources of laser light at red and near infrared wavelengths are also particularly robust and inexpensive.

As mentioned above, it is clear that the amplitudes and shapes of the focus-measure curves utilized for the invention, such as those illustrated in Figs. 2A-B and 5AC, vary substantially depending on the objective lens type and the optical properties (i.e., indices of refraction) of the substrates and specimen media. Therefore, the reliability and the accuracy of the auto-focus approach of the invention can be enhanced by a programmed calibration procedure in which these curves are measured empirically with a given objective lens and given substrate and medium types. Such a procedure can be accomplished by stepping the axial-distance actuator through a sequence of numerous regularly spaced axial steps bracketing the focal position of the interface surface 26 of the substrate 16. The results of each set of calibration curves can be stored in computer memory as a table of discrete values and interpolated as necessary to accomplish routine focusing operations. A calibration procedure might be performed at the start of each working session or whenever relevant conditions (i.e., objective lens, substrate or medium) are changed. When working with multi-well plates upon which dozens or hundreds of auto-focus operations are to be performed, one might start each plate with a fresh calibration run. This procedure would also suffice to ascertain the correct seating of each well plate in its holder as well as to automatically detect plate-to-plate variations under any pertinent condition.

Various changes in the details, steps and components that have been described may be made by those skilled in the art within the principles and scope of the invention herein illustrated and defined in the appended claims. Therefore, while the present invention has been shown and described herein in what is believed to be the most practical and preferred embodiments, it is recognized that departures can be made therefrom within the scope of the invention, which is not to be limited to the details disclosed herein but is to be accorded the full scope of the claims so as to embrace any and all equivalent processes and products.

26

Claims (29)

I claim:

1. An auto-focus method for automatically focusing a reflective surface positioned within an optical path of a 5 lens, comprising the steps of:

(a) producing a plurality of illumination beams; (b) directing the illumination beams toward the reflective surface along the optical path of the lens; (c) measuring a plurality of irradiance distributions of reflection beams spatially distributed over a multielement light detector, said reflection beams corresponding to reflections of the illumination beams from the reflective surface; and (d) determining a current position of the reflective surface with respect to a f ocal point of the lens as a function of said irradiance distributions.

2. The method of Claim 1, further including the step of repeating step (c) at each of a series of axial positions of the reflective surface with respect to the focal point of the lens.

3. The method of Claim 2, wherein said plurality of illumination beams comprises laterally displaced beams; and wherein said step (d) is carried out by identifying a parameter related to a peak of each of said plurality of irradiance distributions measured at said series of axial positions, and by finding an axial position of the reflective surface corresponding to a maximum parameter value for at least one of said irradiance distributions.

4. The method of Claim 3, wherein said parameter is said peak of each of said plurality of irradiance distributions measured at said series of axial positions.

27

5. The method of Claim 2, wherein said plurality of illumination beams comprises laterally displaced beams; and wherein said step (d) is carried out by calculating a parameter related to a width of each of said plurality of irradiance distributions measured at said series of axial positions, and by finding an axial position of the reflective surface corresponding to a minimum parameter for at least one of said irradiance distributions.

6. The method of Claim 5, wherein said parameter is a full-width at halfmaximum of each of said plurality of irradiance distributions measured at said series of axial positions.

7. The method of Claim 1, wherein said plurality of illumination beams comprises axially displaced beams.

8. The method of Claim 7, wherein said plurality of illumination beams comprises an intrafocal beam, a parfocal beam and an extrafocal beam generating corresponding irradiance distributions with corresponding peaks and widths at half-maximum.

9. The method of Claim 8, wherein said step (d) is carried out by matching said peaks of irradiance distributions with predetermined peak values generated by calibration of the lens as a function of axial position of the reflective surface.

10. The method of Claim 8, wherein said step (d) is carried out by matching said full-widths at half-maximum of irradiance distributions with predetermined full-width at half-maximum values generated by calibration of the lens as a function of axial position of the reflective surface.

28

11. The method of Claim 7, further including the step of repeating step (c) at each of a series of axial positions of the reflective surface with respect to the focal point of the lens. 5

12. The method of Claim 1, wherein said plurality of illumination beams comprises laterally displaced beams and axially displaced beams.

13. The method of Claim 12, wherein said axially displaced beams comprise an intrafocal beam, a parfocal beam and an extrafocal beam generating corresponding irradiance distributions with corresponding peaks and widths at half-maximum-

14. The method of Claim 12, wherein said step (d) is carried out by matching said peaks of irradiance distributions with predetermined peak values generated by calibration of the lens as a function of axial position of the reflective surface-

15. The method of Claim 13, wherein said step (d) is carried out by matching said widths at half-maximum of irradiance distributions with predetermined width at halfmaximum values generated by calibration of the lens as a function of axial position of the reflective surface.

16. An auto-focus method for automatically focusing a reflective surface positioned within an optical path of a lens, comprising the following steps:

(a) producing an illumination beam; (b) directing the illumination beam toward the reflective surface along the optical path of the lens; (c) measuring an irradiance distribution of a reflection beam spatially distributed over a multi-element light detector, said reflection beam corresponding to a reflection of the illumination beam from the reflective 29 surface; and (d) determining a current position of the reflective surface with respect to a focal point of the lens as a function of said irradiance distribution. 5

17. The method of Claim 16, further including the step of repeating step (c) at each of a plurality of axial positions of the reflective surface with respect to the focal point of the lens.

18. The method of Claim 17, wherein said step (d) is carried out by identifying a peak of each irradiance distribution measured at said plurality of axial positions, and further by finding an axial position of the reflective surface corresponding to a maximum peak value.

19. The method of Claim 18, wherein said step (d) is carried out by calculating a width at half-maximum for each irradiance distribution measured at said plurality of axial positions, and by finding an axial position of the reflective surface corresponding to a minimum width at half-maximum value.

20. A device for automatically focusing a reflective surface positioned within an optical axis of a lens, comprising:

a light source for producing a plurality of illumination beams; means for directing the illumination beams toward the reflective surface along the optical path of the lens; a multi-element light detector adapted to detect a plurality of reflection beams corresponding to reflections of the illumination beams from the reflective surface; means for varying a position of the reflective surface along the optical path of the lens; and means for determining a current position of the reflective surface with respect to a focal point of the lens as a function of a plurality of reflection-beam profiles measured at multiple elements of said detector.

21. The device of Claim 20, wherein said plurality of illumination beams comprises laterally displaced beams, and said means for determining a current position of the reflective surface comprises means for measuring a peak of each of said reflection-beam profiles and for finding an axial position of the reflective surface corresponding to a maximum peak value for at least one of said irradiance distributions.

22. The device of Claim 20, wherein said plurality of illumination beams comprises laterally displaced beams, and said means for determining a current position of the reflective surface comprises means for measuring a width at half-maximum of each of said reflection-beam profiles and for finding an axial position of the reflective surface corresponding to a minimum width at half-maximum value for at least one of said irradiance distributions.

23. The device of Claim 20, wherein said plurality of illumination beams comprises axially displaced beams, and said means for determining a current position of the reflective surface comprises means for measuring a peak of each of said reflection-beam profiles.

24. The device of Claim 23, further comprising means for matching said peaks of irradiance distributions with predetermined peak values generated by calibration of the lens as a function of axial position of the reflective surface.

25. The device of Claim 20 wherein said plurality of illumination beams comprises axially displaced beams, and said means for determining a current position of the reflective surface comprises means for measuring a width 31 at half -maximum of each of said ref lection-beam prof iles.

26. The device of Claim 25, further comprising means for matching said widths at half-maximum of irradiance distributions with predetermined width at half-maximum values generated by calibration of the lens as a function of axial position of the reflective surface.

27. The device of Claim 20, further comprising means for storing reflection-beam profiles at each of several axial positions of the reflective surface with respect to the lens during a calibration process.

28. An auto-focus method for automatically focusing a reflective surface positioned within an optical path of a lens adapted to operate substantially as hereinbefore described with reference to the accompanying drawings.

29. A device for automatically focusing a reflective surface positioned within an optical axis of a lens constructed and adapted to operate substantially as hereinbefore described with reference to the accompanying drawings.