Method and apparatus for super-resolution imaging of a biological sample

Abstract

The application discloses a kind of biological sample super-resolution imaging method and imaging device, belong to optical imaging technical field.In the imaging method, active light and excitation light are emitted to biological sample, and the fluorescence light generated by fluorescent probe is reflected to detector by objective lens, and detector generates fluorescence image.Visible light generator periodically emits first imaging light and second imaging light, and the drift data of biological sample is predicted according to the generated reference image.According to the drift data, a plurality of fluorescence images are synthesized into super-resolution image.The application determines whether to enter calibration state according to the size of drift data, and moves the workbench in calibration state.The intensity of excitation light is adjusted by the transmittance of optical filter, the luminescence state of fluorescent probe is controlled, and the photobleaching rate of fluorescent probe is delayed.

Description

Technical Field

[0001] This invention relates to the field of optical imaging technology, and in particular to a super-resolution imaging method and imaging device for biological samples. Background Technology

[0002] Super-resolution fluorescence imaging technology excites fluorescent probes (fluorescent molecules) in a sample, illuminating specific areas to form a fluorescence image. Multiple fluorescence images are then synthesized into a super-resolution image, overcoming the resolution limitation imposed by the diffraction limit. However, super-resolution fluorescence imaging is time-consuming, generating only one fluorescence image at a time. Over the imaging period of several minutes, the cumulative effects of spot deformation and sample drift cause gradual distortion of the fluorescence image. CN201811589999.2 discloses a super-resolution imaging device where quenching light from a first laser and excitation light from a second laser are incident on a multimode fiber. The light spot at the fiber's output end is imaged onto a camera in a calibration system. The modulation signal on a spatial light modulator is continuously changed, and the light spot intensity information acquired by the camera is used as the data basis for multimode fiber calibration. This technical solution avoids using the excitation light path as an error correction standard, but it can be used as a reference. However, this technical solution can only be used to correct errors caused by spot deformation.

[0003] CN202010723855.2 discloses a three-channel fluorescence-guided super-resolution biological microscopy system. During imaging, this system uses three laser channels to simultaneously illuminate the sample, causing three types of fluorescent molecules labeled on the sample to flash simultaneously. A detector acquires raw images containing the "flash" signals, and the center positions of the flash points from all the raw images are superimposed to obtain a three-channel ultra-high-resolution image. This technique addresses the sample drift problem in three-dimensional space by registering each fluorescence image. Image registration can eliminate minor sample drift; however, as drift accumulates and the similarity of the center positions of the fluorescence images decreases, image registration becomes insufficient to reduce distortion. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention proposes a super-resolution imaging method and device for biological samples, resolving the image distortion caused by biological sample drift during fluorescence imaging. By adding a calibration state, the intensity of the excitation light is adjusted by controlling the transmittance of the filter.

[0005] The technical solution of this invention is implemented as follows:

[0006] A super-resolution imaging method for biological samples includes the following steps:

[0007] Step 1: Inject fluorescent probes into biological samples, fix the biological samples on the worktable, and the visible light generator sequentially emits the first imaging light and the second imaging light to the biological samples.

[0008] Step 2: After the first imaging ray and the second imaging ray pass through the biological sample, they generate first contrast data and second contrast data in the detector, respectively. Then, a reference image is synthesized based on the first contrast data and the second contrast data.

[0009] Step 3: Entering the imaging state, the first laser emits an activation light and the second laser emits an excitation light. After the activation light and the excitation light are coupled, they are emitted to the biological sample through the objective lens. The fluorescent light generated by the fluorescent probe is reflected to the detector through the objective lens, and the detector generates a fluorescent image.

[0010] Step 4: Enter the detection state. The visible light generator emits the first imaging light and the second imaging light again. A reference image is synthesized based on the first comparison data and the second comparison data generated in the detector.

[0011] Step 5: Calculate the displacement of the reference image when the cross-correlation value between the reference image and the baseline image reaches its maximum, and determine the drift data of the biological sample based on this displacement.

[0012] Step 6: Store the drift data using the sequence number of the fluorescence image as an index. If the drift data is greater than the baseline value, proceed to step 7; otherwise, proceed to step 8.

[0013] Step 7: Enter calibration state. The control unit moves the stage according to the drift data. The excitation light is coupled to the activation light through a filter. The transmittance k of the filter is adjusted according to the duration Δt of the calibration state. t0 is the total time for acquiring N frames of fluorescence images;

[0014] Step 8: If the number of frames of the obtained fluorescence image is less than the preset value N, return to step 3; otherwise, synthesize multiple fluorescence images into a super-resolution image based on the drift data.

[0015] In this invention, in step 2, the reference image I0 = 2(I1-I2) / (I1+I2), where I1 is the first comparison data and I2 is the second comparison data.

[0016] In this invention, in step 5, the cross-correlation value between the reference image and the baseline image is... Where A(x,y) is the pixel value at coordinates (x,y) on the reference image, B(xu,yv) is the pixel value at coordinates (x,y) on the reference image after the displacement (u, v), and μ A μ is the pixel mean of the reference image. B σ is the pixel mean of the reference image. A σ is the pixel variance of the reference image. B Let m be the pixel variance of the reference image, and m and n be the number of pixels at the horizontal and vertical coordinates of the reference image, respectively.

[0017] In this invention, the wavelength of the activation light is matched with the transition wavelength of the fluorescent probe, and the wavelength of the excitation light is matched with the absorption wavelength of the fluorescent probe.

[0018] An imaging apparatus for implementing the super-resolution imaging method for the biological sample, comprising:

[0019] A workbench for placing biological samples;

[0020] A visible light generator is used to emit first and second imaging rays onto biological samples;

[0021] The first laser is used to emit activation light;

[0022] A second laser is used to emit an excitation beam, which is coupled to the activation beam through a filter.

[0023] Objective lens, used to reflect the excitation and activation light onto the biological sample;

[0024] The detector is used to receive the first imaging light, the second imaging light, and the fluorescence light.

[0025] The signal analysis unit is used to generate a reference image or a baseline image, and to generate drift data based on the reference image and the baseline image.

[0026] The control unit is used to move the worktable according to the drift data;

[0027] The synthesis unit is used to synthesize multiple fluorescence images into a super-resolution image based on drift data.

[0028] The present invention further includes a first adjustment component for adjusting the activation light or the excitation light, a second adjustment component for adjusting the first imaging light and the second imaging light, and a third adjustment component for adjusting the fluorescence light.

[0029] In this invention, the visible light generator has multiple programmable LED light sources.

[0030] The super-resolution imaging method and apparatus for biological samples according to this invention have the following beneficial effects: a visible light generator periodically emits visible light, and the drift data of the biological sample is predicted based on a reference image generated from the visible light. During the synthesis of the super-resolution image, the fluorescence image is corrected using the drift data. Furthermore, to address the issue of increased fluorescence imaging period due to increased calibration states, this invention adjusts the intensity of the excitation light by adjusting the transmittance of the filter, controls the luminescence state of the fluorescent probe, slows down the photobleaching rate of the fluorescent probe, and prolongs the effective duration of fluorescence imaging. Attached Figure Description

[0031] Figure 1 This is a flowchart of the super-resolution imaging method for biological samples according to the present invention;

[0032] Figure 2 This is a schematic diagram of the excitation process of fluorescent probe transitions.

[0033] Figure 3 A schematic diagram of fluorescence intensity curves for different excitation light intensities;

[0034] Figure 4 This is a schematic diagram of the preset fluorescence imaging time in this invention;

[0035] Figure 5 This is a schematic diagram of the fluorescence imaging time with calibration status according to the present invention;

[0036] Figure 6 This is a schematic diagram of the reference image of the present invention;

[0037] Figure 7 This is a schematic diagram of a reference image for the present invention;

[0038] Figure 8 This is a schematic diagram of the reference image after displacement adjustment according to the present invention;

[0039] Figure 9 This is a block diagram of the super-resolution imaging device for biological samples according to the present invention;

[0040] Figure 10 This is a structural diagram of the first adjustment component of the present invention;

[0041] Figure 11 This is a structural diagram of the second adjustment component of the present invention;

[0042] Figure 12 This is a structural diagram of the third adjustment component of the present invention. Detailed Implementation

[0043] To better understand the purpose, technical solution, and advantages of this application, the application is described and illustrated below in conjunction with the accompanying drawings and embodiments. Example 1

[0044] Super-resolution imaging requires the continuous capture of fluorescence images of biological samples, a process that typically takes several minutes or tens of minutes. Environmental factors such as vibration can cause drift in the laser or biological sample; the longer the time, the greater the drift. This invention's super-resolution imaging method for biological samples periodically emits a first imaging beam and a second imaging beam, predicting the drift data of the biological sample based on the generated reference image. Figure 1 As shown, the method includes the following steps.

[0045] Step 1: Inject a fluorescent probe into the biological sample, fix the biological sample on the stage, and have a visible light generator sequentially emit a first imaging beam and a second imaging beam into the biological sample. The fluorescent probe is used to label specific biological structures and is then excited by a laser of a specific wavelength to produce fluorescence. The fluorescent probe can be a fluorescent protein molecule or a photoconverting fluorescent dye. The stage refers to the sample stage of the fluorescence microscope. To achieve high-precision translation, the stage can be a piezoelectric direct-drive nanostage. The visible light generator can emit first and second imaging beams with complementary shapes. In this embodiment, the peak emission wavelength of the visible light generator is approximately 660 nm, and the power can be 1000 mW.

[0046] Step 2: After the first and second imaging rays pass through the biological sample, they generate first and second contrast data in the detector, respectively. A reference image is then synthesized based on the first and second contrast data. The contrast of the reference image affects the subsequent calculation of the displacement of the reference image. Biological samples are typically colorless and transparent, with minimal intensity change after light transmission, making it difficult to obtain high-contrast images. Inside the biological sample, the thickness and refractive index of various organelles differ significantly, resulting in noticeable differences in the phase change of light after transmission. This embodiment uses phase imaging, which detects the phase change of the imaging rays, to replace bright-field imaging, improving the accuracy of subsequent calculations and thus enhancing the precision of the drift data estimation.

[0047] In this embodiment, the first and second imaging rays form an asymmetric illumination field to detect the phase information of the biological sample. The first and second contrast data captured by the detector constitute differential phase-contrast imaging, which can obtain a high-contrast image of the sample. According to the differential algorithm, the reference image I0 = 2(I1-I2) / (I1+I2), where I1 is the first contrast data and I2 is the second contrast data. In practical applications, I1 and I2 are usually expressed as a pixel matrix.

[0048] Step 3: Entering imaging mode, the first laser emits an activation beam, and the second laser emits an excitation beam. The activation and excitation beams couple and are emitted through the objective lens onto the biological sample. The fluorescence light generated by the fluorescent probe is reflected by the objective lens to the detector, which then generates a fluorescence image. The wavelength of the activation beam matches the transition wavelength of the fluorescent probe, for example, 405 nm. The activation beam primarily excites the fluorescent molecules from the ground state to the excited state. When the fluorescent molecule absorbs the energy of the activation beam, it transitions to the excited state. The wavelength of the excitation beam, for example, is 640 nm, which matches the absorption wavelength of fluorescent probes such as Cy5 and Alexa Fluor 647 to achieve optimal excitation. After absorbing the excitation beam, the fluorescent molecule returns from the excited state to the ground state by emitting photons, simultaneously generating fluorescence. The fluorescence light is reflected to the detector, which captures the fluorescence image.

[0049] Step 4: Entering the detection state, the visible light generator emits the first and second imaging rays again, and synthesizes a reference image based on the first and second contrast data generated in the detector. The algorithm for generating the reference image is the same as that for the baseline image, and will not be described in detail here.

[0050] Step 5: Calculate the displacement of the reference image when the cross-correlation value between the reference image and the baseline image reaches its maximum. Determine the drift data of the biological sample based on this displacement. Starting with lateral displacement = 0 and longitudinal displacement = 0, substitute different displacement values ​​and calculate the cross-correlation value. The displacement (u, v) at which the cross-correlation value reaches its maximum is the drift data of the reference image. This displacement is used as an estimate of the drift data of the biological sample.

[0051] Step 6: Store the drift data using the sequence number of the fluorescence image as an index. If the drift data is greater than the reference value, proceed to Step 7; otherwise, proceed to Step 8. For any measured drift data, associate it with the sequence number of the fluorescence image as the basis for correction in subsequent super-resolution image synthesis. The reference value is, for example, 120 nm.

[0052] Step 7: Enter calibration mode. The control unit moves the stage according to the drift data, and the excitation light is coupled to the activation light through a filter. In this embodiment, the stage is driven by piezoelectric ceramics and guided by a flexible hinge. The deformation of the piezoelectric ceramics controls the stage to achieve nanoscale movement. In practical applications, the measurement of drift data will inevitably have a certain error. To avoid excessive stage movement causing the biological sample to oscillate back and forth near the target position, the stage movement value is opposite to the drift data. Therefore, the stage movement value can be set to (-0.95u, -0.95v). A calibration mode is added. To avoid insufficient fluorescence intensity due to increased imaging time, this invention adds a filter between the activation light and the objective lens. This reduces the intensity of the current excitation light, but extends the effective duration of fluorescence imaging.

[0053] Step 8: If the number of frames of the obtained fluorescence image is less than a preset value N, return to step 3; otherwise, synthesize multiple fluorescence images into a super-resolution image based on the drift data. N is, for example, 1000 frames. This embodiment first denoises the fluorescence image, then adjusts the position of each pixel in the fluorescence image according to the drift data, aligning the biological samples in the fluorescence images at different times. Then, the resolution of the fluorescence image is improved using a synthesis algorithm based on sparse representation or a deep learning algorithm to synthesize the super-resolution image. The synthesized super-resolution image can be further processed through sharpening, denoising, and other steps. This embodiment adjusts the position of each pixel in the fluorescence image based on the drift data, which can align the biological samples within a small range. Furthermore, to reduce computational load, this embodiment can set an adjustment lower limit; for example, when the horizontal and vertical coordinates of the drift data (u, v) are both less than 5 nm, no adjustment is made. Example 2

[0054] like Figure 2 As shown, after the fluorescent probe is activated, it enters the excited state, emits photons from the excited state, and then returns to the ground state. During this process, some of the fluorescent probe returns to the ground state via the excited triplet state. The fluorescent probes in the excited state and excited triplet state undergo photobleaching at a certain rate. To avoid excessive photobleaching during imaging, users typically need to complete fluorescence imaging within 10 minutes. Figure 3 Normalized fluorescence intensity curves of Rhodamine B as the fluorescent probe at excitation intensities of 40 mW, 20 mW, 10 mW, and 5 mW are shown. The excitation intensity is positively correlated with the photobleaching rate. Given the dependence of the photobleaching rate on the excitation intensity, reducing the excitation intensity, while decreasing the activity of the current fluorescent probe to some extent, can slow down the photobleaching rate.

[0055] Reference Figure 4 and Figure 5 Adding a calibration state increases the fluorescence imaging time. In this embodiment, the preferred recording duration of the calibration state is Δt, and the transmittance k of the filter is adjusted based on the calibration state duration Δt. The photobleaching curve of the fluorescent probe can be fitted to a function P. fluo = P0e rt P0 is the initial fluorescence intensity, P fluo The normalized fluorescence intensity, t is time, and r is defined as the photobleaching rate of the fluorescent probe. Therefore, based on the function of the photobleaching curve, the transmittance is adjusted to... t0 represents the time required to acquire N frames of fluorescence images, composed of the durations of multiple imaging and detection states. This embodiment does not limit the method for adjusting the transmittance of the filter; for example, an electro-filter or a multilayer filter can be used. When the alternating current supplied to the electro-filter decreases, the disorder of the liquid crystal molecules in the electro-filter increases, and the transmittance decreases. Furthermore, when multiple calibration states occur during the imaging process, the duration Δt is accumulated. Example 3

[0056] This embodiment further discloses a preferred method for calculating the cross-correlation value. (See reference...) Figure 6 The baseline image and Figure 7 The reference image has a pixel value of A(x,y) at pixel (x,y), and the reference image has a pixel value of B(x,y). The biological sample drifts on the worktable, causing a mismatch in the positions of the two pixels. In this embodiment, the reference image is moved to maximize the cross-correlation value, ensuring a high degree of overlap between the biological sample in the reference image and the baseline image. Figure 8 As shown. At this point, the displacement of the reference image represents the drift data of the biological sample.

[0057] Specifically, after the reference image is moved by (u, v), the point with the same pixel value as pixel B(x, y) is designated as B(xu, yv). The cross-correlation value between pixel B(xu, yv) and pixel A(x, y) in the baseline image is calculated. Starting from u=0, v=0, different displacement values ​​are substituted to iteratively calculate the cross-correlation value. The displacement (u, v) at which the cross-correlation value reaches its maximum is used to determine the drift data (u, v) of the biological sample.

[0058] This embodiment uses a pixel matching algorithm to calculate the cross-correlation value between the reference image and the reference image. The cross-correlation value between the reference image and the baseline image. A(x,y) is the pixel value at coordinates (x,y) on the baseline image, and B(xu,yv) is the pixel value at coordinates (x,y) on the reference image after a displacement of (u, v). That is, after the reference image is moved (u,v), the coordinates (x,y) on the original image are expressed as (xu,yv) on the new image. μ A μ is the pixel mean of the reference image. B σ is the pixel mean of the reference image. A σ is the pixel variance of the reference image. B Let m and n be the pixel variance of the reference image, respectively, and m and n be the number of pixels at the x and y coordinates of the reference image. It should be noted that pixel values ​​in the image can be expressed as the light intensity at the corresponding coordinates. Example 4

[0059] like Figure 9 As shown, this embodiment discloses an imaging apparatus for implementing the super-resolution imaging method for biological samples, including a stage, a visible light generator, a first laser, a second laser, an objective lens, a detector, a signal analysis unit, a control unit, and a synthesis unit. This super-resolution imaging apparatus for biological samples includes an imaging state, a detection state, and a calibration state. The first and second lasers operate in the imaging and calibration states, respectively. The visible light generator operates in the detection state, and the control unit operates in the calibration state. In the detection state, the visible light generator emits visible light, and the drift data of the biological sample is predicted based on the reference image generated by the visible light. The drift data is used as a correction reference for the fluorescence image. The present invention determines whether to enter the calibration state based on the drift data, and moves the stage in the calibration state.

[0060] The stage of this invention is used to place biological samples. A visible light generator is used to emit a first imaging ray and a second imaging ray onto the biological sample. The visible light generator has multiple programmable LED light sources. The programmable LED light sources have two different operating states, and the first and second imaging rays emitted in the two operating states form two complementary shapes. A first laser is used to emit an activation ray. A second laser is used to emit an excitation ray, which is coupled to the activation ray through a filter. To control the operating time, the first and second lasers may also be equipped with components such as shutters. An objective lens is used to reflect the excitation and activation rays onto the biological sample. A detector is used to receive the first imaging ray, the second imaging ray, and the fluorescence ray. The detector is, for example, an sCMOS camera. A control unit is used to move the stage according to drift data and control the operating state of the filter. In a further embodiment, the control unit is also used to adjust the emission time of the visible light generator. A signal analysis unit is used to generate a reference image or a baseline image, and to generate drift data based on the reference image and the baseline image. A synthesis unit is used to synthesize multiple fluorescence images into a super-resolution image based on the drift data. In this embodiment, the signal analysis unit and the synthesis unit are used to perform light analysis and processing, and image synthesis, respectively. The signal analysis unit and the synthesis unit can be independent computers or different virtual working units of a computer. Physically, they can share or independently use necessary components such as data processing modules and storage modules.

[0061] The imaging device of this embodiment further includes a first adjustment component for adjusting the activation light or the excitation light, a second adjustment component for adjusting the first imaging light and the second imaging light, and a third adjustment component for adjusting the fluorescence light. Figure 10 As shown, the first adjustment assembly comprises at least a collimating lens 11, a first modulation lens 12, a first color filter 13, and a second modulation lens 14. The collimating lens 11 is used to adjust the direction of the laser beam (composed of the activation and excitation beams), and the first color filter 13 is used to filter other wavelengths of laser light, improving the purity of the excitation and activation beams. The first modulation lens 12 and the second modulation lens 14 adjust the diameter of the laser beam, and the parallel beam from the second modulation lens 14 is focused onto the biological sample via the objective lens. Figure 11 As shown, the second adjustment component consists of at least an aspherical condenser lens 21 and a microscope condenser lens 22. The first and second imaging rays are collimated by the aspherical condenser lens 21 and then focused by the microscope condenser lens 22 to illuminate the biological sample. Figure 12As shown, the third adjustment component consists of at least a long-pass filter 31, a cylindrical lens 32, and a beam expander lens 33. The fluorescent light output from the beam expander lens 33 enters the detector. In this embodiment, the excitation and activation light and the fluorescent light use the same optical channel. To reduce light interference, a dichroic mirror can also be arranged below the objective lens. Because there is a certain wavelength difference between laser and fluorescence, the dichroic mirror separates the laser and fluorescence, achieving specific imaging.

[0062] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A super-resolution imaging method for biological samples, characterized in that, Includes the following steps: Step 1: Inject fluorescent probes into biological samples, fix the biological samples on the worktable, and the visible light generator emits the first imaging light and the second imaging light to the biological samples in sequence. Step 2: After the first imaging ray and the second imaging ray pass through the biological sample, they generate first contrast data and second contrast data in the detector, respectively. Then, a reference image is synthesized based on the first contrast data and the second contrast data. Step 3: Entering the imaging state, the first laser emits an activation light and the second laser emits an excitation light. After the activation light and the excitation light are coupled, they are emitted to the biological sample through the objective lens. The fluorescent light generated by the fluorescent probe is reflected to the detector through the objective lens, and the detector generates a fluorescent image. Step 4: Enter the detection state. The visible light generator emits the first imaging light and the second imaging light again. A reference image is synthesized based on the first comparison data and the second comparison data generated in the detector. Step 5: Calculate the displacement of the reference image when the cross-correlation value between the reference image and the baseline image reaches its maximum, and determine the drift data of the biological sample based on this displacement. Step 6: Store the drift data using the sequence number of the fluorescence image as an index. If the drift data is greater than the baseline value, proceed to step 7; otherwise, proceed to step 8. Step 7: Enter calibration state. The control unit moves the stage according to the drift data. The excitation light is coupled to the activation light through a filter. The transmittance k of the filter is adjusted according to the duration Δt of the calibration state. t0 is the total time for acquiring N frames of fluorescence images; Step 8: If the number of frames of the obtained fluorescence image is less than the preset value N, return to step 3; otherwise, synthesize multiple fluorescence images into a super-resolution image based on the drift data.

2. The super-resolution imaging method for biological samples according to claim 1, characterized in that, In step 2, the reference image I0 = 2(I1-I2) / (I1+I2), where I1 is the first comparison data and I2 is the second comparison data.

3. The super-resolution imaging method for biological samples according to claim 1, characterized in that, In step 5, the cross-correlation value between the reference image and the baseline image is... Where A(x,y) is the pixel value at coordinates (x,y) on the reference image, B(xu,yv) is the pixel value at coordinates (x,y) on the reference image after the displacement (u, v), and μ A μ is the pixel mean of the reference image. B σ is the pixel mean of the reference image. A σ is the pixel variance of the reference image. B Let m be the pixel variance of the reference image, and m and n be the number of pixels at the horizontal and vertical coordinates of the reference image, respectively.

4. The super-resolution imaging method for biological samples according to claim 1, characterized in that, The wavelength of the activation light is matched with the transition wavelength of the fluorescent probe, and the wavelength of the excitation light is matched with the absorption wavelength of the fluorescent probe.

5. An imaging apparatus for implementing the super-resolution imaging method for biological samples as described in claim 1, characterized in that, include: A workbench for placing biological samples; A visible light generator is used to emit first and second imaging rays onto biological samples; The first laser is used to emit activation light; A second laser is used to emit an excitation beam, which is coupled to the activation beam through a filter. Objective lens, used to reflect the excitation and activation light onto the biological sample; The detector is used to receive the first imaging light, the second imaging light, and the fluorescence light. The signal analysis unit is used to generate a reference image or a baseline image, and to generate drift data based on the reference image and the baseline image. The control unit is used to move the worktable according to the drift data; The synthesis unit is used to synthesize multiple fluorescence images into a super-resolution image based on drift data.

6. The imaging device according to claim 5, characterized in that, It also includes a first adjustment component for adjusting the activation light or the excitation light, a second adjustment component for adjusting the first imaging light and the second imaging light, and a third adjustment component for adjusting the fluorescence light.

7. The imaging device according to claim 5, characterized in that, The visible light generator has multiple programmable LED light sources.

Images