JP2014164004A - Fluorescence microscope - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a fluorescence microscope configuration capable of correcting aberrations in the whole visual field and image degradations such as blurring of an image, and a correction process.SOLUTION: The fluorescence microscope comprises: a light source 102; an objective lens 109; a two-dimensional sensor 116; and a computer 101 connected to the two-dimensional sensor. The computer stores two or more point distribution functions obtained at different positions in the visual field, and corrects an image by, using the point distribution functions, performing deconvolution on a fluorescence image of a corresponding region in the visual field.

Description

  The present invention relates to a fluorescence microscope, and is used, for example, when separating and analyzing nucleic acids and proteins.

  The next-generation DNA sequencer is made on the basis of a wide-field fluorescence microscope (Non-Patent Document 1, Non-Patent Document 2). In order to detect fluorescence in parallel with a larger number of DNA fragments into which the phosphor has been incorporated, it is necessary to acquire a fluorescent image with less image blur and aberration due to defocusing. In the fluorescence image as described above, since the fluorescence emitted from the DNA fragment can be detected as a clearer fluorescent spot, the recognition accuracy of the DNA fragment is improved.

  As a means for correcting image blur and aberration of a fluorescence microscope, there is a method of deconvolution of a point spread function and a measurement image of the fluorescence microscope. Patent Document 1 discloses a method of obtaining a three-dimensional image of an observation target by correcting the blurred image obtained by moving the objective lens in the optical axis direction (shifting the focus) by deconvolution with a point spread function. Disclosure. Patent Document 2 discloses a method for correcting image blur in a confocal fluorescence microscope by deconvolution with a point spread function. The confocal fluorescence microscope has a drawback that the observation field of view is narrow. Patent Document 3 discloses a method of correcting by repeating a process of deconvolution of a blurred image by changing a point spread function using a brand deconvolution technique.

As a method for acquiring a point spread function, as shown in Non-Patent Document 3, there is a method of acquiring from a fluorescent bead image having a size equal to or smaller than the diffraction limit that is randomly fixed on a substrate. Similarly, Non-Patent Document 3 describes a method of acquiring a point spread function from an image of a laser spot focused on a sample substrate to a diffraction limit. In Patent Document 1, Patent Document 2, and Patent Document 3, the same point spread function is used in the same field of view.

Special Table 2006-518050 JP 2010-39323 A Japanese Patent Laid-Open No. 11-272866

D. R. Bentley et al., Accurate whole human genome sequencing using reversible terminator chemistry, Nature 456, 53-59 (2008) R. Drmanac et al., Human genome sequencing using unchained base reads on self-assembling DNA nanoarrays, Science 327, 78-81 (2010). James B. Pawley, Handbook of Biological Confocal Microscopy, Third Edition, Springer, Chapter 23.

  In the next-generation DNA sequencer, due to the aberration and image blur of the wide-field fluorescence microscope, the fluorescence emitted by the fluorescent substance of the DNA fragment cannot be recognized as a fluorescent spot, or it is erroneously recognized as the fluorescent spot of the wrong DNA fragment. There is. In particular, the above problem becomes more conspicuous at the edge of the field of view as the field of view becomes wider. The aberration is mainly caused by the performance of the objective lens and the imaging lens used, and the deviation of the spatial arrangement of these lenses from the optical axis. Aberrations vary from device to device. Image blur occurs due to misalignment between the focal plane of the objective lens and the light emitting surface of the sample substrate, in addition to aberrations.

  Aberrations and image blur can be corrected by deconvolution using a point spread function. However, since aberration and image blur differ depending on the position in the field of view, the aberration and image blur of the entire image cannot be satisfactorily removed unless a point spread function suitable for the position in the field of view is used.

  The present invention is for solving the above-described problems, and provides a fluorescence microscope configuration and a correction process capable of correcting image deterioration such as aberration and image blur of the entire field of view.

  The present invention employs the configurations described in the claims. A specific example of the fluorescence microscope includes a light source, an objective lens, a two-dimensional sensor, and a computer connected to the two-dimensional sensor, and the computer is acquired at two or more observation positions. The image distribution function is stored, and the fluorescence image acquired by the two-dimensional sensor is corrected by deconvolution using the two or more point image distribution functions acquired in the vicinity of the observation position.

  According to the present invention, by using a point image distribution function acquired at two or more different points in the field of view, the measured fluorescence image is deconvoluted, so that fluorescence superior to that obtained when deconvolution is performed using a single point image distribution function. Demonstrate image correction effect.

The block diagram of the fluorescence microscope of Example 1 integrated in the next-generation DNA sequencer. Point image array substrate. (a) Configuration viewed from above, (b) AA 'sectional view. 2 is a flowchart of a point spread function acquisition method according to the first embodiment. A part of the content of the flowchart of FIG. 3 is illustrated. 5 is a flowchart of an aberration / image blur correction method using the point image distribution function according to the first exemplary embodiment. A part of the content of the flowchart of FIG. 5 is illustrated. Example of correction of sample fluorescence image. 3 is a flowchart of a coordinate conversion coefficient acquisition method for inter-color pixel shift correction according to the first exemplary embodiment. 3 is a flowchart of an inter-color pixel shift correction method according to the first exemplary embodiment. An example of a point image array substrate. (a) Configuration viewed from above, (b) AA 'sectional view. An example of a point image array substrate. (a) Configuration viewed from above, (b) AA 'sectional view. An example of a point image array substrate. (a) Configuration viewed from above, (b) AA 'sectional view. A description example of the point spread function in the definition file 1301 saved in the control PC 101. The block diagram of the fluorescence microscope of Example 2 integrated in the next-generation DNA sequencer. 10 is a flowchart of a method for acquiring a point spread function according to the second embodiment. 9 is a flowchart of an aberration / image blur correction method using a point image distribution function in the second embodiment. 10 is a flowchart of a coordinate conversion coefficient acquisition method for correcting inter-pixel displacement in the second embodiment. Flowchart of inter-color pixel shift correction method in Embodiment 2 FIG. 5 is a configuration diagram of a fluorescence microscope when measuring a point image array substrate in Example 3. Flowchart of correction method using blind deconvolution in embodiment 4 Part of the point image distribution image in the present embodiment in the fifth embodiment (a) A pixel intensity profile in the X-axis direction for the four point images on the straight line BB 'in FIG. (b) Plot in which the intensity profile of the point image in (a) is overlaid with the X-coordinate shift amount corrected (a) The form of the point image array substrate in Example 5. (b) Point image distribution image of point image array substrate 201e

  Hereinafter, the novel features and benefits of the present invention will be described with reference to the drawings. However, the drawings are for explanation only, and do not limit the scope of the present invention.

(Configuration of microscope)
FIG. 1 is a configuration diagram of a fluorescence microscope incorporated in a next-generation DNA sequencer. The configuration of the fluorescence microscope is divided into an irradiation system and a detection system.

  The configuration of the irradiation system will be described in the order of the light traveling path. The light emitted from the light source 102 is converted into a parallel light flux by the condenser lens 107 and enters the filter cube 108. Only a specific wavelength band is transmitted by the band pass filter 103, and the other bands are reflected. The transmitted light is reflected by the dichroic mirror 113 and then collected by the objective lens 109 onto the flow chip 114 in a region larger than the detection visual field region.

Next, the configuration of the detection system will be described. Of the four types of phosphors fixed on the flow chip 114, one type of phosphor having an excitation wavelength band in the specific wavelength band is excited by the light from the condensed light source. Light emitted from the phosphor on the excited reaction spot passes through the dichroic mirror 113, and after the scattering component of the excitation light is removed by the bandpass filter 120, the imaging lens 121 causes the fluorescence spot on the two-dimensional sensor 116. To form an image. Four sets of filter cubes 108 are prepared. By automatically switching the filter cube 108 and changing the excitation and detection wavelength bands, the four types of phosphors can be sequentially detected. There are four types of filter cube 108, F ₁ -F ₄ . As the filter cube F _i detects phosphor Fl _i a (i = 1-4) in the most sensitive, the transmittance characteristic of the dichroic mirror 113 and the band pass filter 103 and the band-pass filter 120 in the filter cube is designed ing.

(DNA sequencing method)
Explain how to decode DNA sequences. On the flow chip, reaction spots in which the same DNA fragments are densely arranged are arranged in advance at a high density. A method for amplifying DNA fragments to consolidate identical fragments is described in DR Bentley et al., Nature 456, 53-59 (2008). DNA sequencing is performed as follows:
1) A nucleobase labeled with four kinds of phosphors is allowed to flow on a flow chip. A nucleobase complementary to the DNA fragment on the reaction spot is incorporated by the polymerase extension reaction.
2) Fluorescence detection of the four types of phosphors incorporated into the reaction spot with a fluorescence microscope. Here, four sample fluorescent images are generated. The sample fluorescence image is analyzed to identify the extended base species for each reaction spot.
3) Move the stage 115 to move to the next detection visual field, and repeat 2).
4) Repeat 3) for the entire flow chip area.

1)-By repeating the operation 4) 100 times, a base sequence of 100 bases per reaction spot can be determined. The extension reaction method is described in Patent Document WO2004 / 018497A2. In the above method, a liquid feeding mechanism for introducing the reagent into the flow chip is necessary, but this is omitted in FIG. For the operation 2), it is necessary to save the four sample fluorescence images in the control PC 101, correct the fluorescence images, and then extract the fluorescence spots. A fluorescent spot is a fluorescent image of a reaction spot. When the fluorescent substance Fl _i exists on the reaction spot, the fluorescent spot of the reaction spot can be seen in the sample fluorescent image acquired by the filter cube F _i . Hereinafter, a method for acquiring a point spread function and an image correction method will be described.

(Acquisition method of point spread function)
FIG. 2 shows a form of a point image array substrate 201a for obtaining a point image distribution function. The point image array substrate 201a has a light-shielding thin film 207 provided with an opening 204 having a diameter of 200 nm on a transparent substrate 202, and a phosphor layer 203 is coated on the light-shielding thin film 207. In the phosphor layer 203, four kinds of phosphors used for DNA sequence decoding are mixed at a high concentration.

  A method of creating the structure of the nano-aperture array substrate excluding the phosphor layer 203 will be described. A light-shielding thin film 207 is formed on the surface of the transparent substrate 202 by sputtering so that chromium is deposited to a thickness of about 100 nm. For the light-shielding thin film 207, a metal film such as aluminum is used. Silver, gold, chromium, silicon carbide, or the like may be used as a material other than aluminum. On the light-shielding thin film 207, a plurality of openings 204 having a diameter of 200 nm are formed in a lattice pattern at intervals of 10 μm by Electron beam lithography technology. The diameter of the aperture may be less than the resolution of the microscope (˜500 nm), and is preferably about half of the resolution (200-300 nm).

FIG. 3 is a flowchart of a method for acquiring a point spread function. FIG. 4 shows a part of the contents of the flowchart. 4A shows the relationship between the entire observation area on the flow chip 114 and each observation field 401, FIG. 4B shows a method of acquiring a point image distribution image in each field, and FIG. 4C shows a point image distribution image. (D) is a schematic diagram of the averaged image distribution function. According to the flowchart of FIG. 3, a point image distribution image is acquired for each observation visual field 401, and a different point image distribution function is acquired depending on the position in the image. The larger the division number N, the more accurate the image correction. The observation field 401 area on the flow chip is 1mm ^ 2, the imaging magnification is 20 times, the size of the two-dimensional sensor is 2048 x 2048 pixels, the pixel size of the sensor is 6.5μm (therefore, one pixel size on the flow chip is 325nm) For example, N is preferably 8 or more (that is, 8 × 8 divisions or more). In the next-generation DNA sequencer of the present invention, the entire observation area of the flow chip is 80 mm in the X direction and 60 mm in the Y direction. Point spread functions were acquired at a total of 25 locations, 5 in the X direction with ΔX = 15 mm and 5 in the Y direction with ΔY = 10 mm. At the time of image correction, a point spread function acquired in the visual field closest to the XY coordinates is used for the sample fluorescence image. The point spread function may be measured at intervals smaller than those described above. In this embodiment, a value obtained by standardizing the pixel value of the averaged point image distribution is used as the point image distribution function. This is shown in FIG. In the definition file 1301 saved in the control PC 101, standardized array values (pixel values) representing the point spread function are input together with the values of the coordinates of the sections. In addition, the coefficient obtained by fitting the point spread with a normal distribution function, a Bessel function, or the like may be stored in the definition file 1301 as the point spread function information. In this case, a common function expression used for fitting needs to be described in the definition file. The feature of the definition file is that a plurality of sets of X, Y coordinate values and point spread functions having a constant interval are stored as shown in FIG. Such a set of point spread functions is different for each DNA sequencer apparatus. Therefore, the numerical value of the point spread function stored in the definition file of the control PC varies from device to device. A method for correcting the sample fluorescence image using the point spread function will be described below.

(Image correction method)
The image correction process using the point image distribution function is divided into an aberration / image blur correction process and an inter-color pixel shift correction process.

  FIG. 5 is a flowchart of an aberration / image blur correction method using a point spread function. FIG. 6 shows a part of the contents of the flowchart. 6A shows the relationship between the entire observation region on the flow chip 114 and the observation field 401, FIG. 6B shows a schematic diagram of the reaction spot 601 in each field, and FIG. 6C shows a corrected image from the sample fluorescence image 602. A conceptual diagram for acquiring 604 is shown. FIG. 7 shows an example in which the sample fluorescence image is corrected according to FIG. The fluorescent spot 603 of the sample fluorescent image 602 is blurred throughout the image due to the focus shift of the objective lens. At the edge of the image, the amount of blur is large due to field distortion. In addition, at the edge of the image, the fluorescent spot is elliptical due to the influence of distortion, coma, astigmatism, and the like. FIG. 7 also shows an averaged point image distribution image divided by 4 × 4 and obtained in the same field of view. The result of correcting the sample image using the point spread function of the point spread image is shown at the bottom of FIG. In the corrected image, a clear fluorescent spot is obtained. Two graphs shown next to the sample fluorescent image and the corrected image are intensity profiles of the cross section A-A ′ at the same position. Two peaks (two black arrows in the figure) seen in the profile of the sample fluorescence image before correction are separated into three peaks (three black arrows in the figure) in the profile of the corrected image. As described above, the above-described image correction method makes it possible to separate and correctly recognize fluorescent spots that are not separated in the sample fluorescent image. In the example of FIG. 7, 4 × 4 division is used, but finer division may be used. Note that the Lucy-Richardson method, the Iterative Back Projection method, or the like can be used as an image correction method by deconvolution of the point spread function. Also, if the image contains a lot of noise due to the sensor used, exposure conditions, etc., noise removal processing is performed before image correction.

  The inter-color pixel shift is a coordinate shift when sample fluorescent images having the same field of view are acquired using different filter cubes 108. This is caused by the fact that the wedge angle of the dichroic mirror 113 and the bandpass filter 120 is not 0 degrees, and chromatic aberration. When there is an inter-color pixel shift, the detected coordinates of the fluorescent spots of different phosphors on the same reaction spot are different. For this reason, there is a possibility that a reaction spot is erroneously recognized between colors.

  FIG. 8 is a flowchart of a coordinate conversion coefficient acquisition method for correcting inter-pixel displacement. In the present embodiment, coordinate conversion coefficients (a, b, c, d, e, f) for correcting inter-pixel displacement are acquired according to the flowchart of FIG. The first step is to obtain the peak position of the point spread function. The peak position of the point spread function is obtained by fitting the point spread with a normal distribution function or a Bessel function. In addition, the pixel position indicating the maximum intensity in each point image distribution may be used as the coordinate of the peak position. The coordinate conversion coefficient acquired according to the flowchart is stored in the definition file.

FIG. 9 is a flowchart of the inter-color pixel shift correction method. In this embodiment, a corrected image obtained by coordinate-converting all the pixels of the sample fluorescent image can be created according to this flowchart.

(Shape of point image array substrate)
The point image array substrate may have a structure shown in FIGS. 10 to 12 in addition to the structure shown in FIG.

  FIG. 10 shows a point image array substrate 201b in which luminescent particles 205 are confined in the openings 204 of the point image array substrate excluding the phosphor layer 203 of FIG. As the luminescent particles 205, fluorescent beads having a diameter of 50 nm were used. A plurality of fluorescent beads enter a 200 nm diameter opening. Fluorescent beads are mixed with 4 types of phosphors used for DNA sequencing. The luminescent particles 205 may be particles that are smaller than the aperture diameter and emit light in the observation wavelength band. In addition to fluorescent beads, quantum dots may be used.

  FIG. 11 shows a point image array substrate 201c in which a light emitting layer 1102 is coated on an adhesive layer 1101 arranged in a grid. For the light emitting layer 1102, four kinds of phosphors used for DNA sequence decoding are used, but any molecule that emits light in a quantum dot or an observation wavelength band may be used. A method for manufacturing the point image array substrate 201c will be described. Synthetic quartz was used for the transparent substrate 202, and a titanium layer was formed in a circular pattern with a diameter of 200 nm by a semiconductor technique in a lattice shape. After the titanium was changed to titanium oxide by ozone ashing, Poly (vinylphosphonic acid) (PVPV) was applied to the titanium oxide surface. PVPA specifically adsorbs to titanium oxide. It does not adsorb to quartz. Thereafter, PVPA was biotinylated, and streptavidin was bound to biotin to form an adhesive layer 1101. Streptavidin is bonded to the surface of the adhesive layer 1101. Thereafter, the four types of biotinylated phosphors were immobilized on the surface of the adhesive layer 1101 by a biotin-streptavidin-biotin sandwich structure, whereby a light emitting layer 1102 was produced. The method for creating the adhesive layer 1101 and the light emitting layer 1102 may be other than the above.

FIG. 12 shows a point image array substrate 201d in which luminescent particles 205 are fixed on an adhesive layer 1101. FIG. The adhesive layer 1101 is formed in the same manner as the point image array substrate 201c in FIG. By using biotinylated fluorescent beads having a diameter of 200 nm as the luminescent particles 205, the luminescent particles 205 can be fixed to the adhesive layer 1101. In addition to fluorescent beads, quantum dots or particles that emit light in the observation wavelength band can be used. The size of the fluorescent particles 205 may be less than the resolution of the microscope, and is preferably 200-300 nm.

This embodiment is characterized in that four types of phosphors are detected simultaneously using four two-dimensional sensors. FIG. 14 is a configuration diagram of the fluorescence microscope of the second embodiment. Using the three dichroic mirrors 1401, 1402, and 1403 and the four two-dimensional sensors 116a, 116b, 116c, and 116d, the fluorescence of the four types of phosphors can be detected simultaneously. The band pass filter 103 has a transmittance characteristic that transmits light in the excitation wavelength band of the four types of phosphors. That is, the band pass filter 103 transmits the excitation light in two or more separated wavelength bands. The four types of phosphors Fl ₁ , Fl ₂ , Fl ₃ , and Fl ₄ on the reaction spot on the flow chip 114 are simultaneously excited by excitation light to emit four types of fluorescence signals. Of the four types of fluorescence signals, the fluorescent components Fl ₁ and Fl ₂ are reflected by the dichroic mirror 1401, and the phosphors Fl ₃ and Fl ₄ are transmitted. Of the reflected light path, the fluorescent component of the phosphor Fl ₁ is reflected by the dichroic mirror 1402, passes through the bandpass filter 120b that transmits only the emission wavelength band of the phosphor Fl ₁ , and then is reflected by the imaging lens 121b. An image is formed as a fluorescent spot on the dimension sensor 116b. Similarly fluorescent component of the fluorescent Fl ₂ is on a two-dimensional sensor 116a by the dichroic mirror 1402 and the phosphor Fl ₂ alone imaging lens 121a is transmitted through the band-pass filter 120b which transmits the fluorescence component of the phosphor Fl ₃ is reflected by the dichroic mirror 1403, on a two-dimensional sensor 116d by the imaging lens 121d passes through the band-pass filter 120d which transmits only phosphor Fl _3, the fluorescence component of the phosphor Fl ₄ is transmitted through the dichroic mirror 1403 Te, on a two-dimensional sensor 116c by the imaging lens 121c is transmitted through the band-pass filter 120c which transmits only phosphor Fl _4, respectively imaged as fluorescent spots. Here, the emission wavelength bands of the four types of phosphors have longer wavelengths in the order of Fl ₁ , Fl ₂ , Fl ₃ , and Fl ₄ . Since the four two-dimensional sensors 116a, 116b, 116c, and 116d acquire images while being synchronized by the control PC 101, the fluorescent spots of the four types of phosphors are simultaneously detected on the four images.

  15 shows a method for acquiring a point spread function in the second embodiment, FIG. 16 shows an aberration / image blur correction method using the point spread function in the second embodiment, and FIG. 17 shows an inter-color pixel in the second embodiment. FIG. 18 shows a method for correcting inter-pixel displacement in the second embodiment. Image correction according to the second embodiment is performed according to the flowcharts of FIGS. The DNA sequencing method is the same as in Example 1.

In this example, since signals of four kinds of phosphors can be detected simultaneously, there is an inherent effect of shortening the DNA sequence decoding time.

  FIG. 19 shows a configuration when the point image array substrate 201 is subjected to fluorescence measurement. The configuration and method for decoding the DNA sequence are the same as in Example 1. A point image distribution image is obtained by a configuration in which an XY driving unit 1902 is added to a two-dimensional sensor 1901 having a pixel size smaller than that of a two-dimensional sensor used for DNA sequence decoding.

  The resolution (the size of one pixel on the flow chip surface) of the two-dimensional sensor of the fluorescence microscope used for DNA sequence decoding may be from half the reaction spot size to the same size. For example, if the reaction spot size is 1 μm in diameter, the pixel size and imaging magnification of the two-dimensional sensor are designed so that the size of one pixel of the two-dimensional sensor on the flow chip surface is about 0.5 μm to 1 μm. good. If the size of one pixel of the two-dimensional sensor is too large (for example, 5 μm), the spatial resolution is lowered, so that it becomes more difficult to separate two adjacent spots. When the imaging magnification is increased and the size of one pixel of the two-dimensional sensor is reduced, the observation field of view is reduced, and the number of reaction spots that can be detected in one field of view is reduced. Therefore, the throughput decreases.

  On the other hand, it is desirable that the two-dimensional sensor for measuring the point image array substrate 201 has a resolution of about twice the optical resolution of the fluorescence microscope. For example, if the optical resolution is 0.3-0.4 μm, the size of one pixel on the flow chip surface of the two-dimensional sensor is preferably about 0.15-0.20 μm. Thereby, since the spread of the point image distribution can be detected by a plurality of pixels, a more accurate point image distribution function can be obtained.

  As in the above example, the optimal value of the size of one pixel of the two-dimensional sensor may be larger in the case of DNA sequence decoding than in the case of measuring a point image array substrate. In such a case, it is better to use a two-dimensional sensor having a smaller pixel size than the two-dimensional sensor used for DNA sequence decoding when measuring the point image array substrate 201. In general, since a two-dimensional sensor with a small pixel size has a small area of the sensor portion, the visual field area when capturing a point image distribution image is smaller than the visual field area in DNA sequence decoding. Therefore, by using the XY driving unit 1902 to move the two-dimensional sensor 1901 on the XY plane and image adjacent plural visual fields, a point spread function of the entire visual field region observed by the DNA sequence decoding can be acquired. Here, the Z axis is parallel to the optical axis of the detection system (the objective lens 109 and the imaging lens 121). The XY plane is a plane perpendicular to the Z axis. The X axis and Z axis are shown in FIG.

  In this embodiment, the point spread function is acquired according to the flowchart of FIG. However, it differs in that a plurality of point image distribution images are acquired at one time. For example, if the field of view in the DNA sequence decoding is 1 mm x 1 mm and the field of view when acquiring the point spread image is 0.5 mm x 0.5 mm, the XY drive unit 1902 doubles it in the X and Y directions in increments of 0.5 mm. The dimension sensor 1901 is moved to acquire four adjacent point image distribution images. An image in which four point image distribution images are arranged is referred to as a point image distribution image in FIG. However, the number of pixels of the averaged point spread image obtained is different from the number of pixels of the image acquired by the two-dimensional sensor used for DNA sequence decoding. Therefore, after the averaged point image distribution image is compressed by signal processing such as binning so that the two images have the same number of pixels, the point image distribution function is stored in a definition file as shown in FIG.

  In the present embodiment, the four two-dimensional sensors of the second embodiment can be changed to the two-dimensional sensor 1901 with the XY driving unit 1902.

The present embodiment has an effect of improving the accuracy of determining the point spread function by using a higher-resolution two-dimensional sensor when acquiring the point spread function.

  In the image correction methods described in the first, second, and third embodiments, a point spread function is acquired for each apparatus, and image distortion caused by the optical system is corrected. When there is little performance variation of the optical system for each apparatus, an image acquired by another apparatus can be corrected based on a point spread function measured by a certain apparatus. Thereby, the point image distribution measurement work for each apparatus can be omitted, and the production efficiency can be increased.

  FIG. 20 shows a flow of image correction. The apparatus A measures the point spread function at each point in the measurement range using the method of FIG. Next, the image acquired by the device B is subjected to blind deconvolution using the point spread function acquired by the device A as an initial value, and the point spread function of the device B is estimated and corrected. Using the point spread function acquired by the device A as an initial value, in the process of performing the iterative process, z when the point spread function was acquired by the device A and the sample fluorescence image were acquired by the device B Differences in z in time can also be corrected. Although it is possible to implement this method without prior information, such as using a Gaussian function as the initial value, there is a drawback that the possibility of falling into a local solution increases with point image distribution estimation in deconvolution. .

The image correction method described in this embodiment can be carried out using any of the optical systems and point image array substrates described in Embodiments 1, 2, and 3.

  The present embodiment is characterized in that the point image array substrate is tilted with respect to the observation field and the point image distribution image is acquired.

  FIG. 21 is a part of a point image distribution image in the present embodiment. A point image distribution image was obtained by tilting the point image array substrate 201 by θ with respect to the observation field. For this reason, the lattice arrangement of the point image is inclined by θ degrees with respect to the XY axis of the observation field. In the figure, θ = 4.1 degrees. Sections separated by white grid lines represent pixels, and numerical values around the image represent coordinates. FIG. 22A is a pixel intensity profile in the X-axis direction for four point images on the straight line BB ′ in FIG. However, the pixel value is a moving average value of three points in the Y direction. The Y coordinate of the point image is a pixel position where the pixel value shows the largest value on the straight line BB 'after the point image distribution image is subjected to moving average processing of three points in the Y direction. In effect, this coordinate position is the pixel position (or center position) of the peak of the point image. In the intensity profiles of the four point images, as the value of the Y coordinate increases, the intensity peak shifts to the right of the X coordinate (in the direction of increasing the X coordinate). This is because the lattice arrangement of the point image is inclined by θ = 4.1 degrees. FIG. 22 (b) is a plot in which the intensity profile of the point image of FIG. 22 (a) is overlaid by correcting the shift amount of the X coordinate. In this way, an intensity profile with many data points can be obtained by correcting the X coordinate shift of the tilted point image and superimposing the intensity profiles. By performing the same operation in the X-axis direction, an intensity profile of the Y-axis cross section obtained by superimposing a plurality of point images in the X-axis direction can be acquired. By increasing the number of data points of the intensity profile in this way, there is an effect of increasing the accuracy of determining the point spread function obtained by fitting. This method can be combined with any of the previous embodiments. The inclination angle θ is preferably 0 <θ <5 °. Any angle other than 0 and 45 degrees and 90 degrees or less can be selected.

  The correction method of X coordinate shift is shown. It is assumed that the point image distribution image has been subjected to moving average processing in the Y direction in advance. In the above, a moving average image of 3 points is acquired, but a moving average of 2 points or less or 4 points or more may be used. The Y coordinate position of the point image is obtained as a coordinate indicating the maximum value of the pixel region that is equal to or greater than the threshold value by providing a threshold value for the pixel value of the moving average image. The coordinate position of the point image may be obtained by calculating the point image distribution interval in the image from the lattice interval of the point image array substrate, the imaging magnification, and the tilt angle θ. The peak position obtained by fitting each point image with a function may be used as the coordinate position. Using the value of the Y coordinate thus obtained, the X coordinate X ′ after correction is obtained as X ′ = XY × tan θ (X is the X coordinate before correction, Y is the Y coordinate of the peak position). ). Assuming that X is Y and Y is X, the Y coordinate shift in the Y direction can be similarly corrected. In FIG. 22, the X-axis value in (a) is converted into X ′, and then four intensity profiles are plotted on the same graph to obtain (b).

A method for obtaining the inclination θ of the point image array substrate 201 will be described. There are cases where θ is obtained by measuring the tilt when the point image array substrate 201 is placed on the stage 115, or θ is obtained by analyzing a point image distribution image. A case where the point image distribution image is obtained by analysis will be described. The point spread function is filtered by a general differential filter in the X or Y direction, and the zero cross point coordinates in the point image area are set to the peak position (or center position) of the point image and the first decimal place It is calculated with the accuracy of (0.1 pixel accuracy). Here, as the differential filter, a Prewitt operator, a Sobel operator, a Savitzky-Golay smoothing differential filter, or the like can be used. The obtained XY coordinates of a plurality of points are fitted with a linear function (y = ax + b) for each column in the X direction or Y direction, and the inclination (a) is obtained as tan θ. The calculation of θ is preferably performed for each divided section by dividing the point image distribution image into N × N. Since the image is distorted due to the influence of aberration, the inclination θ may be different for each section. FIG. 23A shows a form of the point image array substrate 201e in the present embodiment. As shown in the drawing, in addition to the openings 204, a rectangular opening 231 parallel to the lattice arrangement of the openings is provided. A phosphor layer 203 is applied to the rectangular opening 231 as shown in FIG. As shown in FIG. 23 (b), the rectangular opening 231 is detected as a rectangular fluorescent image on the point image distribution image. The inclination angle θ is obtained by detecting the edge of the rectangular aperture image. The position of the edge is obtained from the zero cross point when the filtering process is performed twice by the differential filter. Thereafter, tan θ is obtained from the inclination when the coordinates of the edge position are fitted with a linear function in the same manner as described above, and the inclination angle θ can be determined. By using the rectangular opening, data points at more edge positions can be used, so that the angle θ can be obtained more accurately.

  As mentioned above, although the example of this invention was demonstrated, this invention is not limited to this, It is understood by those skilled in the art that various changes are possible in the range of the invention described in the claim. The It is also within the scope of the present invention to appropriately combine the embodiments.

101 Control PC
102 Light source
103 Band pass filter
107 condenser lens
108 Filter cube
109 Objective lens
113, 1401,1402,1403 Dichroic mirror
114 flow chip
115 stages
116, 1901, 116a, 116b, 116c, 116d Two-dimensional sensor
120 band pass filter
121 Imaging lens
201, 201b, 201c, 201d, 201e Point image array substrate
202 Transparent substrate
203 Phosphor layer
204 opening
205 luminous particles
207 Light-shielding thin film
401 Observation field
402 Point distribution image
403 Averaged point spread image
601 reaction spots
602 sample fluorescence image
603 fluorescent spot
1101 Adhesive layer
1102 Light emitting layer
1301 Definition file
1902 XY drive unit
231 rectangular opening

Claims (15)

A light source;
An objective lens;
A two-dimensional sensor,
Flow chip,
A computer connected to the two-dimensional sensor;
With
The computer stores point spread functions acquired at two or more observation positions, and the fluorescence image of the flow chip acquired by the two-dimensional sensor is used using the two or more different point spread functions. A fluorescence microscope characterized by deconvolution and correction.
In claim 1,
Two or more point spread functions stored in the computer are:
A coordinate value of an observation field, position information in the optical axis direction of the objective lens, and two or more numerical arrays defining a point spread function,
A fluorescence microscope characterized in that the interval between the coordinate positions of the point spread function is substantially constant.
In claim 1,
When the computer obtains the point spread function,
While continuously moving the focus position of the objective lens and the flow cell, acquire a point image distribution image of the substrate,
Divide the acquired image into two or more,
Create an averaged luminescent spot image by averaging multiple luminescent spot images in the division,
Normalize the intensity of the averaged light emission point image and use this as a point spread function,
A fluorescent microscope characterized by controlling the coordinate value of the point spread function and the focus position.
In claim 1,
When the computer corrects the fluorescent image,
Define the focus position of the objective lens and the flow cell when the fluorescent image is acquired,
Dividing the fluorescent image into two or more sections;
Deconvolution for each section using the point spread function acquired at the same focus position as the specified focus position,
Re-image the section image after the deconvolution to obtain a corrected fluorescence image,
A fluorescence microscope characterized by being controlled as follows.
In claim 1,
The point spread function is
A fluorescent microscope characterized in that a substrate in which two or more light emitting points having a size equal to or smaller than a diffraction limit are arranged in a lattice pattern is obtained from a pixel intensity distribution of a fluorescent image obtained by the microscope.
In claim 5,
The substrate is
A transparent substrate;
A light-shielding curtain thin film formed on a transparent substrate;
Two or more gratings provided in the thin film, an opening having a size less than the diffraction limit,
A luminescent material coated on the transparent substrate in the opening;
And a fluorescent microscope characterized by comprising:
In claim 6,
The fluorescence microscope characterized in that the size of the opening is 200-300 nanometers.
In claim 5,
The fluorescent microscope, wherein the luminescent material is a phosphor used for microscopic observation.
In claim 1,
It has 4 kinds of filters,
The fluorescence microscope, wherein the fluorescence image is four fluorescence images acquired by the two-dimensional sensor through transmission of light collected by an objective lens.
In claim 1,
Four two-dimensional sensors,
With three dichroic mirrors,
The fluorescence microscope is characterized in that the light collected by the objective lens is divided into four optical paths by the dichroic mirror and is acquired by the two-dimensional sensor.
In claim 1,
A fluorescence microscope characterized in that a pixel size of a two-dimensional sensor when acquiring a point spread function is smaller than a pixel size of a two-dimensional sensor when acquiring a fluorescence image of a flow chip.
In claim 12,
A fluorescence microscope comprising a stage for driving a two-dimensional sensor in parallel with an observation field plane in a two-dimensional sensor for acquiring a point spread function.
In claim 1,
Two or more point spread functions stored in the computer are:
It was acquired with a different fluorescence microscope of the same configuration,
A fluorescence microscope comprising a step of estimating, as an initial value, a point spread function acquired by a different fluorescence microscope in the step of correcting an image.
In claim 5,
A fluorescence microscope characterized in that the substrate is subjected to fluorescence detection in a state where a lattice direction of the light emitting point is inclined with respect to a pixel lattice of the two-dimensional sensor.
In claim 15,
The tilted state is a fluorescence microscope characterized in that the tilt angle is greater than 0 and 5 degrees or less.