WO2003003057A2

WO2003003057A2 - System and method for capturing multi-color fish images

Info

Publication number: WO2003003057A2
Application number: PCT/IL2002/000450
Authority: WO
Inventors: Opher Shapira; Yuval Harari; Eran Kaplan
Original assignee: Bioview Ltd.
Priority date: 2001-06-27
Filing date: 2002-06-11
Publication date: 2003-01-09
Also published as: WO2003003057A3; AU2002309241A1

Abstract

A method of obtaining multi-color FISH images is provided. The method is effected by: (a) using a color imager for capturing a dark image of a magnified biological sample stained with as least two distinct FISH probes; (b) determining dark intensity (Id) for each pixel of the dark image; (c) using the color imager for capturing a fluorescently illuminated image of the magnified biological sample; and (d) correcting an intensity (I) of each pixel of the fluorescently illuminated image of the biological sample according to its dark intensity (Id) to thereby obtain a multi-color FISH image of the biological sample.

Description

SYSTEM AND METHOD FOR CAPTURING MULTI-COLOR FISH

IMAGES

FIELD AND BACKGROUND OF THE INVENTION

The present invention relates to a system and method for capturing high resolution multi-color images from biological specimen. More particularly, the present invention relates to a novel imaging system which utilizes a color camera for capturing multi-color fluorescence in-situ hybridization (FISH) images.

Imaging systems employing microscopes and computer controlled image-capturing devices such as CCD cameras are routinely used to analyze and characterize biological specimen.

In addition to being used for morphological characterization of biological components, such imaging systems are also used for biochemical and genetic characterization of biological components.

For example, the Fluorescence In Situ Hybridization (FISH) imaging method is routinely used to genetically evaluate complex biological specimens, such as blood, amniotic fluid, and solid tumors. In FISH, fluorescently labeled DNA probes specific to a target DNA of interest are imaged via fluorescence microscopy. FISH enables in-situ analysis of native cell chromosomes of individual cells enabling detection of genetic abnormalities associated with prenatal disorders, cancer, and other genetic diseases. FISH is a low light application and as such, FISH imaging systems typically employ monochrome cameras that are efficient at collecting almost every photon at the spectral range used.

A fluorescence microscope equipped with appropriate filter sets, a strong lamp, which illuminates the sample in the required spectrum (e.g. a mercury arc lamp), and fluorescence objectives with numerical apertures > 0.75 is typically required to view the results of a FISH assay. The probes and a counterstain can be visualized separately or simultaneously depending on the filter set used. A single-band-pass filter allows one fluorophore to be viewed, while a multi-band-pass filter allows viewing of several different fluorophores. Use of multicolored probes allows for simultaneous detection of multiple genetic events within a single nucleus. Thus, multi-color screening can be used to identify the clonal nature of a specimen as well as to detect chromosomal rearrangements present in metaphase and interphase cells. By using probes that hybridize to control loci (versus test loci), multi-color screening can also be used to determine hybridization efficiency for each nucleus, thus greatly increasing the accuracy of the assay.

There are several approaches for generating multi-color or multi-event FISH detection. One approach employs several spectrally distinct fluorophores each capable of hybridizing with a different target. At least seven different chromosomes have been detected simultaneously by using this approach. Another approach, termed color-coding, utilizes different fluorophores singly and/or in combination to identify more chromosome targets than the number of fluorophores used. By combining fluorophores, 2^N-1 targets can be detected (N = number of fluorophore labels). When each combination consists of different proportions of each fluorophore label, the number of targets detectable is limited only by the ability to distinguish the different label proportions. For example, eight chromosomes have been reproducibly detected using only two-fluorophore labels, while all 24 chromosomes have been detected simultaneously in metaphase chromosomes by use of only five fluorophore labels.

Although widely utilized, FISH imaging systems employing monochrome cameras suffer from several inherent limitations.

To compose a single color image, several images each representing a different color plane must be stacked into a composite image which accurately represents the objects of interest. Generating such a composite image is difficult when the objects of interest are small, as is the case with most FISH applications. Shifts of several microns between the image planes can result in a composite color image which does not accurately represent the true image.

In addition, since each fluorophore image is taken separately using a different filter, the need to mechanically change filters leads to an inherently slow imaging sequence.

Furthermore, since all color planes are obtained using a monochrome camera, color must be artificially added according to the filter used. For example, an image captured through a red filter is artificially painted in a reddish hue, while an image captured through a green filter is artificially painted with a greenish hue. Inconsistency between captured and ocular- viewed images can lead to misinterpretation of results.

While reducing the present invention to practice, the present inventors have devised a system and method which can be used to capture true multi-color FISH images, thus traversing the limitations inherent to prior art FISH imaging systems and methods.

SUMMARY OF THE INVENTION

According to one aspect of the present invention there is provided a method of obtaining multi-color FISH images. The method is effected by: (a) using a color imager for capturing a dark image; (b) determining a dark intensity (I_d) for each pixel of the dark image; (c) using the color imager for capturing a fluorescently illuminated image of a magnified biological sample stained with as least two distinct FISH probes; and (d) correcting an intensity (I) of each pixel of the fluorescently illuminated image of the biological sample according to its dark intensity (I_d) to thereby obtain a multi-color FISH image of the biological sample.

According to further features in preferred embodiments of the invention described below, step (d) is effected according to I_corr= (I- I_d) According to still further features in the described preferred embodiments the method further comprising capturing a uniformly illuminated bright image of a known intensity (I_ref) of said magnified biological sample stained with as least two distinct FISH probes and calculating a gain factor (G) according to G= I_ref /(I_b- I_d) for each of the pixels prior to step (d), where I_b is the bright intensity for each pixel of said bright image.

According to still further features in the described preferred embodiments step (d) is effected according to I_corr= G*(I- I_d).

According to still further features in the described preferred embodiments the biological sample is magnified by a microscope having a numerical aperture of at least 0.75.

According to still further features in the described preferred embodiments the method further comprising the step cooling the color imager prior to the step of capturing the fluorescently illuminated image of the magnified biological sample.

According to still further features in the described preferred embodiments the method further comprising staining the biological sample with the at least two distinct FISH probes prior to step (a).

According to another aspect of the present invention there is provided an imaging system for obtaining multi-color FISH images comprising: (a) optical components defining an optical path being capable of generating a magnified image of at least a portion of a biological sample stained with at least two distinct FISH probes; (b) a color imager being for capturing a color image of the biological sample magnified by the optical components; wherein the optical components are selected so as to enable resolving of the at least two distinct FISH probes in a fluorescently illuminated image captured by the color imager from the biological sample stained with the at least two distinct FISH probes.

According to still further features in the described preferred embodiments the system further comprising a computing platform, the computing platform executing at least one software application being for: (i) determining dark intensity (I_d) for each pixel of a dark image captured by the color imager; and (ii) correcting an intensity of each pixel (I) of the fluorescently illuminated image of the biological sample stained with the at least two distinct FISH according to its dark intensity I_d. According to still further features in the described preferred embodiments step (ii) is effected according to I_corr= (I- I_d).

According to still further features in the described preferred embodiments the at least one software application also serves for calculating a gain factor (G) according to G= I_ref /(l_b- I_d) for each of the pixels prior to step (ii), where I_ref is a uniformly illuminated bright image of a known intensity of said magnified biological sample stained with as least two distinct FISH probes and l_b is the bright intensity for each pixel of said bright image.

According to still further features in the described preferred embodiments step (ii) is effected according to I_oorr= G*(I- I_d). According to still further features in the described preferred embodiments the system further comprising a cooling device, the cooling device being capable of cooling the color imager below room temperature.

According to still further features in the described preferred embodiments the color imager is a color CCD camera. According to still further features in the described preferred embodiments the color CCD camera is a three CCD camera.

According to still further features in the described preferred embodiments the optical path includes an objective with a magnification larger than x40 and an NA larger than 0.75, at least one fluorescence filter set including an excitation filter, a beam splitter and an emission filter.

According to yet another aspect of the present invention there is provided a color CCD camera comprising: (a) a body being configured for attachment to a magnifying device; (b) a lens; (c) a CCD array for capturing photons incident on the lens; and (d) a cooling device being for cooling the CCD array thereby decreasing the probability of thermal noise generated in the camera.

According to still further features in the described preferred embodiments the CCD array is a three CCD array. According to still further features in the described preferred embodiments the CCD array includes three monochrome array each being capable of receiving photons of a single color channel.

According to still further features in the described preferred embodiments the cooling device employs Peltier cooling. The present invention successfully addresses the shortcomings of the presently known configurations by providing a system and method, which can be used to capture a multi-color FISH image.

BRIEF DESCRIPTION OF THE DRAWINGS The invention is herein described, by way of example only, with reference to the accompanying drawings. With specific reference now to the drawings in detail, it is stressed that the particulars shown are by way of example and for purposes of illustrative discussion of the preferred embodiments of the present invention only, and are presented in the cause of providing what is believed to be the most useful and readily understood description of the principles and conceptual aspects of the invention. In this regard, no attempt is made to show structural details of the invention in more detail than is necessary for a fundamental understanding of the invention, the description taken with the drawings making apparent to those skilled in the art how the several forms of the invention may be embodied in practice. In the drawings:

FIG. 1 is a diagram illustrating the system of the present invention; FIG. 2 is a multi-color FISH image as obtained by one configuration of the imaging system of the present invention; and FIG. 3 is the multi-color FISH image of Figure 2 following image correction according to the teachings of the present invention;

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention is of an imaging system and method which can be used to obtain a multi-color FISH image. Specifically, the present invention can be used to capture a single multi-color image of biological specimen stained with a plurality of distinct FISH probes, thus enabling simultaneous detection of several distinct FISH probes within a single biological sample. The principles and operation of the present invention may be better understood with reference to the drawings and accompanying descriptions.

Before explaining at least one embodiment of the invention in detail, it is to be understood that the invention is not limited in its application to the details of construction and the arrangement of the components set forth in the following description or illustrated in the drawings. The invention is capable of other embodiments or of being practiced or carried out in various ways. Also, it is to be understood that the phraseology and terminology employed herein is for the purpose of description and should not be regarded as limiting.

FISH imaging - Overview

A comparison of the light energy in conventional vs. FISH images:

Conventional, Bright-Field (BF) imaging of biological specimen employs a Halogen lamp, which produces a continuous light spectrum centered at » 900nm. In order to view cells, a CCD camera set to an exposure time of 1 mSec is typically used.

FISH imaging uses a mercury lamp, which generates several bands, which pass through two filters on route to the camera. The camera is typically set to an exposure time of approximately 1 second. This setting enables one to view the fluorescent information at a gray level similar to that of BF imaging. This large difference between the exposure times used in BF and FISH imaging is due to the low light level emitted by the FISH sample. In FISH, the probability that a photon incident on the sample will excite the fluorophore is very small, and as such, the number of photons which are collected by the camera surface is typically very small.

Efforts to overcome this low light level are mainly directed at improving the light energy density (better matching of the filters and probes to the light source), developing probes with a higher excitation probability and/or optimizing the camera. Using CCD camera for FISH imaging: In principle, any CCD camera is suitable for FISH imaging, provided enough photons are allowed to accumulate in the CCD pixels allowing sufficient signal levels. As with most low-level measurements, noise signal must be discounted or minimized in order to provide reliable measurements of true signals. The number of photons incident per pixel can be increased by optimizing the CCD efficiency (so that every photon is detected), increasing the pixel area (either by improving imager geometry or by decreasing the magnification) and/or increasing the integration time of the signal.

Noise sources in FISH imaging: The noise sources in FISH imaging may be divided into optical noise and imager noise.

Optical noise sources include:

(i) Illumination noise: This noise relates to the photon flux variations and is dependent on the type of lamp used.

(ii) Photon noise: This noise relates to the number of photons emitted into the CCD and converted into photoelectrons per unit time. Noise arises from the fundamentally statistical nature of photon production. The number of photons impinging upon a given pixel, during two consecutive but independent observation intervals of length T, will vary. Since Photon production is governed by the laws of quantum physics one can only assume the average number of photons within a given observation window. The probability for counting p photons during an observation window of T seconds is Poisson distributed. It is critical to understand that even if there were no other noise sources in the imaging chain, the statistical fluctuations associated with photon counting over a finite time interval T would still lead to a finite signal-to-noise ratio (SNR) since the noise level is relative to the square root of the signal level.

The noise fluctuations due to photon statistics may be ignored in bright signals where the number of photons exceeds 10⁵.

(iii) Probe decay: The intensity emitted by the probes decreases exponentially with time. The timescale of decay is tens of seconds. While not a source of noise, it certainly affects the reliability of measurements.

There are five sources of imager noise:

(i) Photoelectron noise: not every photon incident on the CCD surface will generate a photoelectron. The process of generating photoelectrons is a

Poisson process, and is very similar to the photon noise described hereinabove. (ii) Thermal Noise: thermal energy is a stochastic source of electrons in a CCD well. Electrons which are freed from the CCD material itself through thermal vibration and then retrapped in the CCD well are indistinguishable from "true" photoelectrons. By cooling the CCD chip it is possible to significantly reduce the number of "thermal electrons" that give rise to thermal noise (dark current). Thermal noise is independent of the signal, and is a characteristic of the CCD itself. As the integration time T increases, the number of thermal electrons increases. Thermal electrons are Poisson distributed, where the rate parameter is an increasing function of temperature.

(iii) On-chip Electronic Noise: This noise originates in the process of reading the signal from the sensor, in this case tl rough the field effect transistor

(FET) of a CCD chip. The general form of the power spectral density of readout noise is:

where o and β are constants and α is the (radial) frequency at which the signal is transferred from the CCD chip. At very low readout rates (α<c_min) the noise displays a II f character. Readout noise can be reduced to manageable levels by appropriate readout rates and proper electronics. At very low signal levels however, readout noise can still become a significant component in the overall SNR.

(iv) Amplifier Noise: The noise originating from the amplifier circuits of the camera is additive, Gaussian, and independent of the signal. In modern well-designed electronics, amplifier noise is generally negligible. (v) Quantization noise: This noise is generated by an error in the analog to digital conversion circuits. It is usually on the order of one gray level.

Technical considerations Analysis of the dominant noise factors: In a modern CCD camera, having carefully designed electronics, one may ignore the on chip, amplifier and quantization noise. At sufficiently high light levels, the main noise source is the photon noise. In order to get a satisfactory signal to noise ratio at low light levels, the inherent Poisson process makes long exposure times mandatory. For example, in order to have a SNR of 100, one needs to collect 10000 electrons (approximately 20000 photons). For long exposure times, the number of photoelectrons becomes comparable with the number of thermal electrons. The number of photons incident on the CCD surface, N, increases linearly with time: N= T (Eq. 2), Where f is the number

7/9 of photons per unit time and T is the time. The photon noise is equal to ( T) . On the other hand, thermal electrons accumulate with a rate of χT. Thus, the overall noise level, σ, can be represented by: c²=_fT+₂T=( +2)T (Eq. 3) while the SNR can be represented by:

SNR=(_fT)^1/2/[l+(λ/f)^]/2J (Eq. 4) where χ is the coefficient of thermal noise, in "photons" per second per pixel. χ reduces the SNR of the camera; for example, using a commercial uncooled camera that has ^5000 "thermal photons'Vsecond and collects light photons at a rate of 2500 photons per second, (^=2500), reduces the SNR by a factor of 2.4. In low light level application such as FISH, exposure times of 1 second or even longer are used for image acquisition. For such applications, the main sources of noise are photon noise, photoelectron conversion noise and thermal noise.

Achieving satisfactory SNR:

The optical system is made of a light source, optics and the collection system. To maximize the energy flux emitted by the sample and incident on the camera surface the light source must be optimized with respect to intensity stability and spectral characteristics.

Optimizing the intensity stability assures a constant flux of photons on the sample, while optimizing the spectral characteristics concentrates as much energy as possible at wavelengths that excite the sample. If suboptimal illumination is utilized, higher energy is needed to obtain the same signal. However, when higher energy is used, parasitic radiation may cause crosstalk in the sample, and may even damage the sample. Optimizing optical components: The illumination path should be designed such that it will deliver most of the light energy into the sample. On the other hand, the illuminated field of view should be as large as the CCD area. These requirements are oftentimes compromised by a need to support more than one magnification.

The filters should pass as much energy as possible in the excitation band, block non relevant wavelength, and pass back as much energy as possible in the emission band. Filters should be designed carefully to match the FISH probes used. Since manufacturers usually match modern filters with probe types, a large variety of filters and filter combinations are available.

The collection optics should employ a high numerical aperture (NA) objective, in order to maximize light collection (the higher the NA, the larger the spatial collection angle), and to image the light onto a small area. The best objectives available have an NA of 1.4, and enable a spatial resolution of 0.17μ m within the blue (λ=400nm) wavelength.

The collection system must also be optimized to better convert incident electrons into an image. Exposure times must be minimized in order to minimize noise resultant from a dark current. To minimized exposure times, the collection system must be designed such that each pixel receives the maximal possible photon flux. This can be achieved by:

(i) maximizing the fill factor (making the entire surface area of the camera sensitive to light);

(ii) maximizing quantum efficiency (the ratio of electrons to incident photons);

(iii) optimizing spectral response (making the camera most sensitive to light at relevant wavelengths). (iv) matching the pixel size to the needed resolution, if the pixel size is too small, the same number of photons would divide between several pixels, thus lowering photon counting; and

(v) minimizing the dark current (e.g. by cooling).

Integrating over time: For some applications, the amount of light incident on the collection system surface is so small, that one needs either to amplify the signal or to increase exposure times. Long exposures (time integration) enable to collect more photons thus increasing a signal, although at a cost of increasing the noise. However, while a signal increases linearly with time integration, noise increases as a square root only. Thus, in principle, detection of any weak signal is possible provided long enough exposure times are used. Such time integration is limited, however, by the accumulation of thermal noise and limited dynamic range well before an acceptable signal to noise ratio is achieved. For example, assume that an incoming photon flux is 2500 photons per second, and that every photon is converted to a photoelectron. Assume now that the rate of thermal electrons is 10⁴ per second, and that the system saturates at 10⁵ electrons. According to this example, a maximal integration time of 8 seconds is available. During this time, 20000 electrons will be collected from the signal, and 80000 electrons will be collected from thermal electrons. Only 20% of the dynamic range is left for the signal; however, the thermal noise will only be 80000 ^" , or 283. Thus, the resultant SNR is approximately 60 which is satisfactory for most applications. Now assume that the photon rate drops to 1000 photons/second In this case, the maximal integration time will be 10 seconds, and the best signal to noise ratio will be only 10.

Cooling the camera:

In extreme low light applications, very large integration times may be needed; for such applications it essential to reduce the thermal noise.

The probability for thermal release of an electron can be represented by e^{~ c} , where T is the temperature, k is the Boltzman constant (k=l .38* 10 ) and E is the energy needed to release the electron. Thus, reducing the temperature reduces the probability of a thermal release event. For example, if the temperature is reduced from 300K to 25 OK, the probability of electron release from a typical gap of 0.5eV decreases by a factor of 58. Thus, cooled cameras play a very important role in low light applications, enabling integration times of hours for a single exposure.

Image enhancement: Capturing the fluorescent image onto a computer memory, enables analysis and/or manipulation of image information, including all color information. Computer analysis/manipulation enables enhancement of one color component over the others, identification of points of interest, color comparison, color analysis at a different focal planes and the like. Most computer analysis/manipulation techniques are better executed on a true color image rather than a composite.

For example, relative color intensity can be more accurately determined from a true color image as opposed to a composite image assembled from several color planes each captured under unique conditions. In addition, the need to analyze relative color intensity from various focal planes introduces another variation factor thus reducing the integrity of the comparison even further.

A true multi-color image also enables analysis of colors that do not coincide with filters used in monochrome capturing. Such colors may result from overlapping colors, generated by, for example, juxtaposed chromosomes.

Unlike monochrome FISH, a true multi-color FISH image, can be easily manipulated and analyzed thus enabling the user to derive more information from the image.

The system of the present invention

The technical considerations described hereinabove were used as a basis for deriving parameters for constructing the system of the present invention.

As is further described in the Examples section which follows, the system constructed according to the teachings of the present invention enabled

"one-shot" capturing of multi-color FISH images of high quality.

Referring now to the drawings, Figure 1 illustrates the system of the present invention which is referred to herein as system 10.

System 10 includes a color camera 12 which is preferably a 3-chip CCD color camera. Examples of commercially available color cameras suitable for use with the present invention include, but are not limited to, the DXC9000 manufactured by Sony Inc. Camera 12 preferably includes a cooling device 13, utilizing, for example, Peltier cooling. Cooling device 13 serves to cool camera 12 thus dramatically decreasing thermal noise therefrom. As is further described hereinbelow, camera 12 is designed for capturing images magnified by microscope 14. Microscope 14 can be a fluorescent microscope such as that manufactured by, for example, Zeiss, Olympus, or others. Configurations of microscope 14 which are suitable for multi-color FISH imaging according to the teachings of the present invention are further described in the Examples section which follows.

System 10 further includes a computing platform 16, which communicates with camera 12 and preferably microscope 14. Computing platform 16 can be, for example, a personal computer such as PC or a Macintosh computer or a work station such as that manufactured by Sun

Computers, Silicon Graphics and the like.

Computing platform 16 operates software application(s) which enable a user of system 10 to vary the exposure time, gain and/or any other image capturing parameters of camera 12. In addition computing platform 16 preferably also controls image capturing, thus ensuring that images are captured under optimal conditions (e.g., that the specimen is correctly positioned and that there are no vibrations that may effect the image quality).

Computing platform 16 may also control the specimen holder, the objective (i.e. magnification), the focus and the illumination intensity of microscope 14.

As detailed herein, noise patterns which hinder data extraction from real images limit the effectiveness of low light imaging applications such as, FISH.

As such, system 10 of the present invention preferably employs a novel image correcting approach which is effected as follows: (i) a dark image is captured and the dark intensity (I_d) of each pixel is determined;

(ii) a uniformly illuminated bright image of a known intensity I_ref is optionally also captured and the bright intensity (I_b) of each pixel is determined; and (iii) when snapping a fluorescent FISH image, system 10 of the present invention automatically corrects the intensity of each pixel (I) of the fluorescent image according to its dark intensity as determined in (i).

Alternatively, in cases where a uniformly illuminated bright image is also captured, then the following steps are also effected by system 10:

(iii) the gain factor, G, as G= I_ref /(l_b- I_d) is computed for each pixel; and (iv) When snapping a fluorescent FISH image, system 10 of the present invention automatically corrects the intensity of each pixel, I, according to the formula: I_C01T=G*(I- I_d) (Eq. 5)

A detailed description of a FISH imaging system constructed and calibrated according to the teachings of the present invention is provided in the Examples section which follows.

The system and method of the present invention employ an RGB color camera instead of the conventional monochrome camera typically used in FISH systems. In order to understand the advantages of using an RGB camera for

FISH imaging, the two camera types are compared below for factors affecting image quality.

Camera quantum efficiency: Quantum Efficiency (QE) defines how many electrons are generated from each photon incident on the CCD surface. The value of QE depends on the properties of the pixel, as well as on the part of the chip that actually collects the light (fill factor). Most CCD chips achieve a peak QE of 0.3-0.4, although some CCD chips may achieve QE of up to 0.8. There are two types of color cameras: Mosaic CCD and 3 chip CCD. In the Mosaic CCD type, each pixel has a color filter above it, which transfers only the correct color to the filter. This method implies that for each color, only one third of the CCD area collects the light. Thus, the actual QE for this type of CCD is reduced by a factor of 3. The 3 chip CCD splits the incoming light to three color bands, and deflects each color onto a separate CCD chip (hence the name 3 chip CCD). Using this method, the QE is reduced only by the efficiency of the band splitting process, and is close to that of a monochrome camera. Camera SNR and spectral response:

The noise characteristics of a monochrome CCD and a color CCD are similar. However, when considering the SNR of a color camera, one needs to consider the matching between the incident light spectrum and the Red, Green and Blue spectrum in the camera. Ideally, the spectral response of a monochrome camera is uniform for all visible light spectrum. For a color camera, the ideal response is uniform, but divided into 3 bands (R, usually centered at λ_mid=630nm, G (λ_mi_d=540nm), B (λ _mi_d ⁼440nm)). In practice, the responses are smoother, and fall off gradually between the bands. If the incident light corresponds to the R, G, B bands, then the SNR of a color camera will be similar or even somewhat better to that of a monochrome camera. However, when the incident light is at λ=585nm (between the red and green wavelengths) approximately half of the photons will fall on each CCD. This will lead to a double exposure time, and an increase in thermal noise by a factor of 2.

Camera spectral resolution

Modern FISH applications often employ several probes (color markers). Each probe is characterized by a unique excitation and emission spectrum. If a monochrome camera is used together with a perfectly fitting filter, the only restriction on the resolution is the overlapping of different probes. Commercial probes have Full Width Half Maximum (FWHM) of approximately 30nm-50nm, both for the excitation and the emission wavelength, this means that by proper choice of filters, one may easily distinguish up to 10 colors in the visible light. However, if the filters allow partial excitation and emissions of other colors, information may be lost. A color camera, such as the one used by the present invention, allows simultaneous imaging of several colors. In a color camera, color information is collected in all color channels, so actual color resolution is limited only by the

SNR of the system. Overlapping probes are detectable since direct identification thereof is possible. An RGB camera allows direct resolution of 3 different colors; if a trivial algorithm for color weighting is employed, the resolution may be doubled.

Lateral accuracy:

Fluorescent imaging employs high numerical aperture (NA) objectives, in order to collect as much light as possible. Such objectives also have high resolution. For example, a common xl00/NA=1.3 objective may resolve objects separated by 0.23μm. When using a monochrome camera, separating two colors implies changing a filter, which involves a mechanical motion, which can reduce resolution. FISH systems employ special registration algorithms to overcome this problem. In order to increase the accuracy of such algorithms, one needs to increase magnification, which means that the amount of light per pixel decreases. This, in turn, reduces the effective SNR, and leads to a small and limited field of view. Using a color camera eliminates this problem since the mechanical offset between the color channels is constant and very small. Since color imaging does not involve mechanical motion, the pixel size may be chosen by the objective resolution only. This leads, in addition to an improved SNR, to an ability to view a larger area, a feature which is important in many applications. Time of imaging:

Color cameras snap a single image. In contrast, monochrome cameras snap an image for each filter used with the overall time needed for image acquisition given by:

(Eq. 6), where T is the overall time per image, N is the number of filters used, and t_map and t_βter are the time per snap and time needed to switch filters respectively. As is clear from Equation 6, image acquisition in prior art monochrome systems is time consuming and thus less suitable for large scale screening applications, which require rapid screening.

Thus, the present invention provides a novel multi-color FISH imaging system and method which traverse the limitations inherent to prior art monochrome FISH imaging systems while providing FISH imaging of a quality comparable to that provided by prior art systems. Table 1 below provides a comparison between prior art systems and the present invention further illustrating the advantages of the system of the present invention.

Table 1 Comparison between prior art monochrome FISH imaging systems and the system of the present invention

Additional objects, advantages, and novel features of the present invention will become apparent to one ordinarily skilled in the art upon examination of the following examples, which are not intended to be limiting. Additionally, each of the various embodiments and aspects of the present invention as delineated hereinabove and as claimed in the claims section below finds experimental support in the following examples.

EXAMPLES

Reference is now made to the following examples, which together with the above descriptions, illustrate the invention in a non limiting fashion.

The present invention discloses a novel FISH imager, which uses an RGB camera, instead of the customary monochrome camera. As shown hereinbelow, RGB imaging, which provides several inherent advantages, can be efficiently used in FISH assays.

EXAMPLE 1 System considerations

The following is a description of a color system suitable for FISH imaging. Illumination path components:

Illuminator: For most applications, a 100W mercury lamp is optimal since it has spectral emission peaks that match most standard FISH fluorophores in the market. Thus, loss due to radiation which does not excite the probes is minimized. Alternatively, a xenon lamp may be used wherever non-standard probes are used (e.g. with multiple (>5) color FISH or spectral imaging). In cases where the number of fluorophores to be used is limited to 3 or less, laser sources, which maximize the incident light intensity can be utilized.

Illumination lenses: The illumination optical path is selected so as to be able to illuminate the camera Field of View uniformly. The standard method of microscope illumination (termed Kohler illumination) produces the best results for non point source illuminators (e.g., lamps). It is important that the illuminator NA will match the objective NA, in order to minimize light lose; thus, the system of the present invention employs an NA as close to 1 as possible.

Filters: Standard commercially available fluorescent filters which have a typical transmission efficiency of 0.5 for the relevant bandwidth, and only 0(1%) for non-relevant bandwidth are utilized by the system of the present invention. Objectives: Fluorescent objectives are selected so as to maximize light transfer. The signal quality of objectives having an NA < 0.5 are almost useless for FISH imaging; thus, the NA utilized is preferably larger than 0.75.

In practice, the system of the present invention utilizes a standard Zeiss microscope (AxioPlan II). This microscope incorporates a 100W mercury lamp coupled with a Koehler illuminator, standard filters supplied by Zeiss, and the following objectives:

? x40 (NA=0.75),

? x63 (NA=1.4), oil immersion.

? X63 (NA=0.95) ? xl 00 (NA=1.3), oil immersion.

This illumination path is capable of delivering an excitation light of 35mW directly into the sample, for every fluorophore used.

Detection path components:

The sample itself is stained with commercial fluorophores (e.g., Dapi, Texas Red and FITC). The scattering cross section (the probability to emit a o photon for an incident one) of these fluorophores is approximately 3*10^" μm .

Objectives: Same as described above. Note that an objective with NA equal to 0.75 collects approximately 15% of the scattered light, while an objective with NA=1.4 collects approximately 30% of the scattered light. Filters: Same as described above. Camera lens: A standard camera lens, produced by, for example, Zeiss, is utilized by the system of the present invention.

Camera: the following parameters are considered when selecting a camera: (i) the maximal allowed time for an image; and

(ii) the signal to noise ratio.

The following example describes the considerations when selecting a camera suitable for acquiring images with a resolution of O.lμm/pixel under a maximal exposure time of 1 second and a noise level less than 1% from a green (λ=550nm) FISH probe of length lOKbp (having 2500 fluorescent molecules).

Preferably, a xlOO objective, with a numerical aperture of 1.3 is used for image viewing. Since the incident light level in the relevant band is 35mW, an influx of 10 photons/seconds impinges upon a Field of View (FOV), having an area 4.9* 10⁴ μm²; as such, the resulting flux density is 2* 10¹² photons/sec*μ m².

The rate of scattered photons equals the number of scatterers multiplied by the scattering cross section and the rate of incident photons. In this example, the rate of scattered photons equals 2* 10¹²*2500*3* 10^"8=50* 10⁶ photons per second. Assuming in this case, that the objective's transmittance is 0.9, that the filter's transmittance is 0.8 and that the camera lens transmittance is 0.9, the resulting rate of photons impinging upon the camera surface would be 9.7* 10⁶ photons per second, due to the limited NA. Using a pixel size of 0.1 μm implies that the probe described above will be covered by approximately 50 pixels, with each pixel receiving 2* 10⁵ photons per second. Assuming a QE of 0.3, and a color response of 0.5, translates to 30000 electrons per second per pixel. Since the conversion of photons to electrons is Poisson distributed, the count noise will be 175 electrons. If the camera is maintained at room temperature (RT), the thermal noise will add approximately 100 electrons per second. Thus, the overall noise level in this case will be 200 electrons per second, or 0.67%.

According to the above considerations, using an un-cooled 3-CCD camera having a lOμm pixel size and a full well capacity of 30000 electrons (standard for this pixel size), will be sufficient for capturing a color image with a signal to noise ratio better than 100, within one second of integration.

The system of the present invention preferably utilizes a commercial, low cost 3-CCD camera (e.g., Sony DXC9000), which meets the above considerations. The camera is operated through an exposure time of 0.5-2.5 seconds, and depending on the probe size and quality of the sample, achieves an SNR better than 100.

Two more configurations, which are extensions of the above considerations, are also considered herein. Working with a camera that has a smaller pixel size (e.g., 6.7μn), with a full well capacity of 13000 electrons and QE of 0.5 (e.g., the pixelFly camera, available from PCO Computer Optics), would maintain a probe image size of 50 pixels, providing an objective of x63 is utilized. In this case, however, due to an enhanced quantum efficiency, an integration time of 0.3 seconds is sufficient since only 13000 electrons are needed for a full signal. Although the noise level will increase to a level of 1.2%, it is still within working limits.

Using a cooled CCD chip reduces the thermal noise to a negligible level. Using a cooled camera with the above setting enables to improve the SNR to better than 100 while retaining a low integration time. Use of 3 monochrome cameras instead of one RGB camera:

A novel "3 chip CCD" made of connecting 3 B/W imager chips via custom dichroic filters, which match the actual colors of the probes can be constructed and used by the present invention. Such a camera configuration exhibits all the advantages of B/W camera, having an ability to image all color chamiels, at optimal matching, simultaneously. The registration between images may be performed once using a well-defined target, in order to achieve sub pixel accuracy. While more expensive than a single color camera, such a camera configuration may provide optimal imaging per application. With respect to all other considerations, such a novel camera design is identical to any other color camera.

Intensified camera:

An intensified camera may also be used by the system of the present invention. Such a camera will add sensitivity to the system, and will enable fast imaging. The advantages and disadvantages of intensified cameras are known in the art, and need to be taken into account when considering this configuration.

Cooled RGB camera:

The system of the present invention may also employ a cooled camera. As mentioned hereinabove, cooling the camera dramatically decreases noise levels, hence improving the signal to noise ratio.

Designing probes that match camera response:

To overcome color matching problems, the present invention preferably utilizes probes which optimally match the peaks of the color chamiels of the camera. Color image enhancement:

Image averaging: Image averaging may be used in order to reduce signal to noise ratio without saturating the imager. Image averaging reduces the random noise by a factor of " N, where N is the number of images taken. The technique is limited by the time of processing used to generate a single image, which may, in turn, be limited by the illumination, probe properties, mechanical properties of the system etc. Alternatively, other "averaging" techniques such as the median filter may also be used.

Spatial and spatio-temporal filtering: Another technique of noise reduction, which may be applied to the images captured, is spatial filtering. Spatial filtering refers to "neighborhood" calculations performed on a certain pixel to try and find the information embedded within the noise, by comparing the contents of the pixel of interest to neighboring pixels. Such algorithms are very strong in the sense that they may reveal information embedded in the signal even when such information is not detectable via image viewing. In machine vision applications, this means that the integration (averaging) time may be reduced, since even in cases where the image looks "bad", the information is there, and significantly above the noise and as such can be extracted.

A trivial expansion of spatial filtering is spatio-temporal filtering, which compares the content of any pixel both to its neighboring pixels and to images of the same pixel taken at a different time. For example, one may use the averaging filter followed by a "closing" filter, in order to find the contents of a given pixel.

"Chromo "filtering: The use of a color camera by the present invention enables use of yet another type of filters, the chromo filters, which add a chromatic dimension to the filtering process. Chromo filtering allows determination of a difference between the green and the red colors for a certain pixel, and how such pixels are different from neighboring pixels. Extracting color information from images: removing limitation of 3 colors:

Since the system of the present invention utilizes a color camera, it is not limited to the low color resolution of prior art monochrome systems. The use of a color camera by the present invention greatly enhances the ability to resolve colors, since the color information is recorded simultaneously, and is free of spatial and temporal variations.

Color is viewed as mapping of relative intensities of color channels into a color space.

The ability of an eye to easily resolve a plurality of colors is exemplified by the color test bar of monitors. The test bar features seven colors, which are very easily resolved by the eye. In fact, anyone who tried to view a computer monitor set at 16 colors notices the superior resolution of three-color channels. For FISH imaging, it is the bandwidth of the probes that limits the color resolution. A regular fluorophore has a FWHM bandwidth of 30-100nm, which is significantly higher than the resolving power of any color camera.

Extracting quantitative information:

Relative intensity measurements: In principle, if the transfer functions of the optical path are known (including lamp spectrum, filters response, camera response, as well as the scattering cross section of the probes), one can compare and evaluate the information present from the intensity of the signals. Color imaging facilitates the quantitative analysis of image color content, since all of the information is provided by a single image. In addition, information may be integrated over the z (focal) dimension, independently from the height of the measurement, in order to get a fuller measurement. While this feature of the present invention can be effected by B/W imaging as well, the time it takes makes it impractical.

Probe positions: The fact that images obtained by the system of the present invention are snapped simultaneously enables spatial analysis of data with a far better resolution than the composite images obtained by prior art monochrome systems.

For example, when analyzing chromosomal translocations a difference of a single pixel at microscope resolution (0.15μm) will be enough to negate a translocation. This is hard to achieve using monochrome systems, due to the registration problems inherent to such systems. Thus, simple spatial analysis tools enable resolving of closely positioned probes thus significantly improving results.

Probe colors: Loss of information arising from probe colors which are spectrally close_? may be overcome by introducing a simple transformation of the intensity and color space. This task an be achieved by modifying the Hue, saturation and Luminance (HSL) curves, or by using any other convenient transformation technique which separates colors. Overcoming depth of field problems:

Using a large NA objective enables the system of the present invention to collect as much light as possible from the fluorescent source. However, this feature is limited by the depth of field (the range of height change for which the image remains in focus).

The use of a color camera for fluorescent imaging enables to overcome this problem, since imaging is fast and does not require registration. Two possible approaches can be used to overcome this problem:

Sequence of images: Instead of imaging a single image, a "snap" will be defined as a sequence of images, each at a different height around the "best' focal point. Viewing through this acquired image sequence can be effected via keyboard or mouse control. Such an image "scrolling" method is advantageous since it produces three-dimensional information which can be easily viewed in a manner similar to manual FISH observation.

Although this method requires large amounts of storage memory image compression can be utilized to overcome such limitations. "Confocal imaging": According to this approach, a sequence of images is obtained and every pixel is taken from an image within which this pixel is at maximal local contrast. The combined image includes a set of pixels each at maximum contrast. While seemingly trivial to generate, special care should be taken such that artifacts are not introduced into such images. This approach is advantageous in that it enables generation of sharp high-quality images, which include all of the information available in a single image.

However, this method may be limited by the loss of 3D information during processing of some images. For example, two stacked chromosomes, one transmitting a red signal and the other a green signal at two different heights within the image plane, would "fuse" to provide an interim color. In such cases, "confocal imaging" may lead to erroneous results.

EXAMPLE 2 FISH imaging

To test the viability of multi-color FISH imaging according to the teachings of the present invention, an imaging system employing a standard, low cost, 3-CCD color camera was constructed. The system was used to image a sample stained with commercial fluorophores (Dapi, Spectrum orange and Spectrum green; all from Vysis Ltd.) each having a scattering cross section of approximately 3* 10^"8μm².

The system was configured as follows:

Illumination path components:

(i) Illuminator: 100W mercury lamp (Osram HBO103 W2). (ii) Illumination lenses: standard Kohler illuminator (Zeiss)

(iii) Filters: Standard fluorescent filters, compatible with fluorophore colors including Zeiss filters set 25 and 31, as well as a Vysis aqua/orange/green filter set.

(iv) Objectives (Zeiss): ? x40 (NA=0.75),

? x63 (NA=1.4), oil immersion.

? X63 (NA=0.95)

? xlOO (NA=1.3), oil immersion.

The resulting illumination path of the system was capable of delivering 35mW of excitation light into the sample for every fluorophore used.

Detection path components:

(i) Camera lens: A standard camera lens (Zeiss) was used. ^•

(ii) Camera: The following parameters were considered when selecting a camera: -The maximal allowed time for an image. -The signal to noise ratio -CCD quality

Depending on the sample type, a conventional 3 CCD camera may capture a satisfactory color image using an integration time of 0.5-2.5 seconds. The system of the present invention utilized a commercial, low cost,

3-CCD camera (Sony DXC9000). This camera has a maximal exposure time of up to 16 seconds which is more than sufficient for imaging.

The system described above was used to acquire multi-color FISH images. Figure 2 illustrates a typical dark image obtained using an exposure time of 2 seconds. Clearly, at such an exposure time the CCD uniformity becomes a major consideration.

Due to a non-uniform pixel response, a noise pattern is generated; extracting data from a real image becomes virtually impossible. In order to overcome this problem, uniformity must be corrected by:

(i) capturing a dark image (i.e., an image taken with the optical path closed) and determining the dark intensity (I_d) for each pixel;

(ii) (optional) capturing a uniformly illuminated bright image, with a known intensity I_ref and determining the bright intensity (I_b) of each pixel; (iii) when snapping a fluorescent FISH image, correct the intensity of each pixel (I) of the fluorescent image according to its dark intensity as determined in (i).

Alternatively, in cases where a uniformly illuminated bright image is also captured, then the following steps are effected: (iii) For each pixel compute its gain factor, G, as G= I_ref /(I_b- I )

(iv) When snapping a fluorescent FISH image, correct the intensity of each pixel, I, according to the formula:

I_corr=G*(I- I_d) (Eq. 5) Applying this correction to the image shown in Figure 2 results in a vastly enhanced image (Figure 3). As seen in Figure 3, the upper left area contains 2 cells, which would not have been seen without the correction.

Thus, this algorithm enables to correct uniformity problems, which lead to image degeneration. Although high uniformity CCD cameras are available, such cameras are expensive. Such expenses are multiplied for high uniformity 3-CCD cameras, which at present are not commercially available.

Since one-CCD color imaging will need exposure times which are relatively long, a 3-CCD camera is essential for high resolution, fast color imaging.

Although the invention has been described in conjunction with specific embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art. Accordingly, it is intended to embrace all such alternatives, modifications and variations that fall within the spirit and broad scope of the appended claims. All publications, patents and patent applications mentioned in this specification are herein incorporated in their entirety by reference into the specification, to the same extent as if each individual publication patent, or patent application was specifically and individually indicated to be incorporated herein by reference. In addition, citation or identification of any reference in this application shall not be construed as an admission that such reference is available as prior art to the present invention.

Claims

WHAT IS CLAIMED IS:

1. A method of obtaining multi-color FISH images comprising:

(a) using a color imager for capturing a dark image;

(b) determining a dark intensity (I_d) for each pixel of said dark image;

(c) using said color imager for capturing a fluorescently illuminated image of a magnified biological sample stained with as least two distinct FISH probes; and

(d) correcting an intensity (I) of each pixel of said fluorescently illuminated image of said biological sample according to its dark intensity (I_d) to thereby obtain a multi-color FISH image of said biological sample.

2. The method of claim 1, wherein step (d) is effected according to Icorr⁼ (I- Id)-

3. The method of claim 1, further comprising capturing a uniformly illuminated bright image of a lαiown intensity (I_ref) of said magnified biological sample stained with as least two distinct FISH probes and calculating a gain factor (G) according to G= I_ref /(l_b-I_d) for each of said pixels prior to step (d), where I_b is the bright intensity for each pixel of said bright image.

4. The method of claim 3, wherein step (d) is effected according to

5. The method of claim 1, wherein said biological sample is magnified via a microscope having a numerical aperture of at least 0.75.

6. The method of claim 1, further comprising the step cooling said color imager prior to said step of capturing said fluorescently illuminated image of said magnified biological sample.

7. The method of claim 1, wherein said color imager is a color CCD camera.

8. The method of claim 7, wherein said color CCD camera is a three CCD camera.

9. The method of claim 1, further comprising staining said biological sample with said at least two distinct FISH probes prior to step (a).

10. An imaging system for obtaining multi-color FISH images comprising:

(a) optical components defining an optical path being capable of generating a magnified image of at least a portion of a biological sample stained with at least two distinct FISH probes; and

(b) a color imager being for capturing a color image of said biological sample magnified by said optical components; wherein said optical components are selected so as to enable resolving of said at least two distinct FISH probes in a fluorescently illuminated image captured by said color imager from said biological sample stained with said at least two distinct FISH probes.

11. The system of claim 10, further comprising a computing platform, said computing platform executing at least one software application being for:

(i) determining dark intensity (I_d) for each pixel of a dark image captured by said color imager; and

(ii) correcting an intensity of each pixel (I) of said fluorescently illuminated image of said biological sample stained with said at least two distinct FISH according to its dark intensity I_d.

12. The system of claim 11, wherein step (ii) is effected according to corr U-" idy-

13. The system of claim 11, wherein said at least one software application also serves for calculating a gain factor (G) according to G= I_ref /(l_b-Id) for each of said pixels prior to step (ii), where I_ref is a uniformly illuminated bright image of a lαiown intensity of said magnified biological sample stained with as least two distinct FISH probes and I_b is the bright intensity for each pixel of said bright image.

14. The system of claim 13, wherein step (ii) is effected according to I_C01T= G*(I- 1„).

15. The system of claim 15, further comprising a cooling device, said cooling device being capable of cooling said color imager below room temperature.

16. The system of claim 10, wherein said color imager is a color CCD camera.

17. The system of claim 16, wherein said color CCD camera is a three CCD camera.

18. The system of claim 10, wherein said optical path includes an objective with a magnification larger than x40 and NA larger than 0.75, at least one fluorescence filter set including an excitation filter, a beam splitter and an emission filter.

19. A color CCD camera comprising:

(a) a body being configured for attachment to a magnifying device;

(b) a lens;

(c) a CCD array for capturing photons incident on said lens; and

(d) a cooling device being for cooling said CCD array thereby decreasing the probability of thermal noise generated by the camera.

20. The color CCD camera of claim 19, wherein said CCD array is a three CCD array.

21. The color CCD camera of claim 19, wherein said CCD array includes three monochrome array each being capable of receiving photons from a single color channel.

22. The color CCD camera of claim 19, wherein said cooling device employs Peltier cooling.