Patent Application
20250069419
The present disclosure relates to a method and a device for training a machine learning model to construct a digital image depicting an artificially stained sample. The present disclosure further relates to a method and a microscope system for constructing a digital image depicting an artificially stained sample.
In the field of digital microscopy, a typical task is to find and identify objects within a sample. For instance, within hematology and cytology, specific cell types may be found and identified in order to establish a diagnose for the patient from which the sample is taken.
Samples are typically stained which increases the contrast within the sample, which allows objects within the sample to be detected. For example, white blood cells are more or less transparent, and are therefore difficult to identify using a microscope without staining the sample. However, chemical agents used for staining may be toxic, and therefore requires laboratory personnel to follow safety procedures (e.g., using protective equipment and/or fume hoods) to avoid exposure to such chemicals. The staining process is also a complex process and may therefore increase the time needed for analysis. Due to its complexity, it can often be difficult to ensure a consistent result of the staining process. For example, depending on the quality of the used chemical agents, the resulting color of the sample may vary as well as the degree of staining. In addition, due to the chemicals needed, the safety procedures, and the extra time needed, staining samples prior to analysis may increase the economic costs associated with the analysis.
It is also typically beneficial to quickly screen a sample for specific types of cells. During such screenings, there is a compromise between speed and precision. High precision of the screening typically requires a large magnification of the sample, which allows cells to be imaged and analyzed. Hence, only a small portion of the sample is imaged at a time, and a large number of individual positions of the samples must be therefore be imaged in order to screen the entire sample which leads to a time-consuming screening process. Thus, in order to reduce the time needed for screening, the number of imaged positions could be reduced. However, given that the entire sample is to be screened, this requires that the magnification is reduced, which, on the other hand, reduces the precision in the screening, leading to a screening process that may not properly find and identify cell types.
Hence, there is a need for improvements within the art.
It is an object to, at least partly, mitigate, alleviate, or eliminate one or more of the above-identified deficiencies in the art and disadvantages singly or in any combination and solve at least the above-mentioned problem.
According to a first aspect a method for training a machine learning model to construct a digital image depicting an artificially stained sample is provided. The method of the first aspect comprising: receiving a training set of digital images of an unstained sample, wherein the training set of digital images is acquired by illuminating the unstained sample from a plurality of directions and capturing a digital image for each of the plurality of directions; receiving a ground truth comprising a digital image of a stained sample, wherein the stained sample is formed by applying a staining agent to the unstained sample; and training the machine learning model to construct a digital image depicting an artificially stained sample using the received training set of digital images of the unstained sample and the received ground truth.
Within the context of this disclosure, the wording “digital image depicting an artificially stained sample” should be construed as a computer-generated digital image of the unstained sample. This digital image may be similar, or even identical, to a digital image of the stained sample. Therefore, the depicted artificially stained sample may have a contrast similar to a depicted stained sample (which have actually been stained by a staining agent). Put differently, the depicted artificially stained sample may replicate a sample actually being stained by the staining agent.
Within the context of this disclosure, the wording “unstained sample” should be construed as a sample without applied staining agent. The unstained sample may have a relatively lower contrast compared to a stained sample. The contrast of the unstained sample may be low such that it may be difficult, or even impossible, to discern features of the unstained sample using conventional microscopy in a digital image of the unstained sample.
Within the context of this disclosure, the wording “ground truth” should be construed as information that is known to be real and/or true. Hence, in this context, since the machine learning model is trained to construct a digital image depicting an artificially stained sample, the ground truth may represent a stained sample. The ground truth may in this context therefore comprise a digital image of a stained sample. The ground truth may comprise a digital color image of a stained sample. Put differently, the digital image of the ground truth may depict a stained sample in color.
Within the context of this disclosure, the wording “staining agent” should be construed as one or more chemical agents that may be used to increase a contrast of a sample (e.g., by changing the color). The staining agent may be chosen depending on the type of sample and/or what type of objects within the sample is of interest. For example, within hematology, samples (e.g., blood smears) are typically stained by using May Grünewald-Giemsa (MGG), Wright-Giemsa (WG), or Wright. As a further example, within pathology, samples are typically stained by using Hematoxylin-cosin (H&E). Different staining agents may increase the contrast for different objects and/or substances within a sample. For example, some staining agents may be adapted increase the contrast of proteins, etc. Hence, which staining agent to be used may be chosen depending on the sample and/or on the type of objects of interest in the sample.
Hence, the machine learning model is trained to correlate the training set of digital images of the unstained sample to the ground truth (e.g., the digital image of the stained sample). The machine learning model may be trained iteratively and/or recursively until a difference between an output of the machine learning model (i.e., the digital image depicting the artificially stained sample) and the ground truth (e.g., the digital image of the stained sample) is smaller than a predetermined threshold. A smaller difference between the output of the machine learning model and the ground truth may indicate a higher accuracy of the constructed digital image depicting the artificially stained sample provided by the machine learning model. Put differently, a smaller difference between the output of the machine learning model and the ground truth may indicate that the constructed digital image depicting an artificially stained sample may to a higher degree replicate a digital image of the stained sample. Hence, preferably, the difference between the output of the machine learning model and the ground truth may be minimized. The machine learning model may be trained to construct digital images of artificially stained samples for a plurality of different sample types. In such case, the machine learning model may, for each sample type, be trained using a training set of digital images of a sample of the sample type, and a corresponding ground truth associated with the respective sample type.
By illuminating the sample from a plurality of different directions and capturing a digital image for each of the plurality of directions, information regarding finer details of the unstained sample may be captured than what normally is resolvable by a conventional microscope (i.e., using a conventional microscope illumination) used to image the unstained sample. This can be understood as different portions of Fourier space (i.e., the spatial frequency domain) associated with the unstained sample are imaged for different illumination directions. This technique may be known in the art as Fourier Ptychography. Further, by illuminating the unstained sample from a plurality of different directions and capturing a digital image for each of the plurality of directions, information regarding a refractive index associated with the unstained sample may be captured. This can be understood as an effect of refraction of light being dependent on an angle of incident for light illuminating the unstained sample and the refractive index of the unstained sample. Information regarding the refractive index of the unstained sample may, in turn, allow phase information (typically referred to as quantitative phase within the art) associated with the unstained sample to be determined. Since the plurality of digital images comprises information associated with one or more of fine details of the unstained sample, a refractive index associated with the unstained sample, and phase information associated with the unstained sample, this information may be used in the training of the machine learning model, which may, in turn, allow for a machine learning model being trained to more accurately construct the digital image depicting the artificially stained sample than what would be allowed in case the plurality of digital images were captured from only one direction or by using conventional microscopy. Using conventional microscopy (e.g., by illuminating the unstained sample from a majority of the plurality of directions up to a numerical aperture of the microscope objective used to image the unstained sample), it may be difficult, or even impossible, to capture information associated with the refractive index associated with the unstained sample and/or phase information associated with the unstained sample. Illuminating the unstained sample from a plurality of different directions may further allow for capturing information relating to details of the unstained sample which are finer than what normally is allowed by a microscope objective used to image the unstained sample. Thus, a microscope objective having a relatively lower magnification may be used while still being able to capture information related to fine details of the unstained sample. Using a relatively lower magnification microscope objective may, in turn, allow for larger portions of the unstained sample to be imaged at each imaging position. Hence, the entire unstained sample may be scanned by imaging at relatively fewer positions which, in turn, may allow for a faster scanning of the unstained sample.
Hence, the present disclosure allows for training a machine learning model to construct a digital image replicating a stained sample using a plurality of digital images of an unstained sample. Hence, by using the trained machine learning model, a digital image depicting a stained sample (i.e., the artificially stained sample) may be constructed without having to apply a staining agent to the unstained sample.
The method of the first aspect may further comprise: acquiring the training set of digital images of the unstained sample by: illuminating the unstained sample from a plurality of directions, and capturing, for each direction of the plurality of directions, a digital image of the unstained sample.
The training set of digital images may be acquired by illuminating the unstained sample with white light from the plurality of directions and capturing a digital image for each of the plurality of directions.
Within the context of this disclosure, the wording “white light” should be construed as light having a relatively broad spectrum of wavelengths. White light may, e.g., resemble sunlight. The spectrum of white light may comprise a majority of the visible light spectrum, as opposed to light having a narrow spectrum (e.g., monochromatic light).
An associated advantage is that white light (i.e. broad-spectrum light) may allow for more information about the sample to be captured in the training set of digital images, which otherwise would be missed in parts of the spectrum not covered when using narrowband light sources. In turn, that information may be used by the trained machine learning model when constructing a digital image depicting an artificially stained sample which to an even higher degree may resemble a digital image of a stained sample. Monochromatic light is typically required in imaging applications utilizing Fourier ptychography, and several different colors of monochromatic light may therefore be needed in prior art systems to capture multispectral information about the sample. Thus, using white light may reduce a total number of digital images needed to capture multispectral information about the sample compared to prior art systems utilizing Fourier ptychography. Further, using different colors of monochromatic light may only allow information about the sample for those specific wavelength ranges to be captured, and since white light may have a broader spectrum, using white light may allow for capturing information about the sample for a broader wavelength range compared to when using monochromatic light.
The training set of digital images may be acquired using a microscope objective and an image sensor, and wherein at least one direction of the plurality of directions may correspond to an angle larger than a numerical aperture of the microscope objective.
The numerical aperture of the microscope objective may be a dimensionless number associated with a range of angles over which the microscope objective accepts light. Hence, a direction larger than the numerical aperture may be understood as a direction corresponding to an angle larger than the range of angles over which the microscope objective accepts light.
By illuminating the sample from an angle larger than the numerical aperture of a microscope objective, the digital image captured for that angle of illumination may comprise information about higher spatial frequencies, and thereby finer details of the unstained sample than the microscope objective normally allows. This may, in turn, allow for the microscope objective to capture phase information associated with the unstained sample and information relating to details not normally being resolvable by the microscope objective, which may be used in the training of the machine learning model. Put differently, by illuminating the unstained sample from an angle larger than the numerical aperture of the microscope objective may allow for an improved training the machine learning model to construct a digital image depicting the artificially stained sample.
The method of the first aspect may further comprise: applying a staining agent to the unstained sample, thereby forming a stained sample; and forming the ground truth comprising the digital image of the stained sample by: acquiring a digital image of the stained sample. Put differently, the stained sample and the unstained sample may be the same sample with a difference being the applied staining agent.
An associated advantage is that features (e.g., objects) in the stained sample may be paired with corresponding features in the unstained sample, whereby an improved training of the machine learning model to construct the digital image depicting the artificially stained sample may be allowed.
The act of acquiring the digital image of the stained sample may comprise: illuminating the stained sample simultaneously from a subset of the plurality of directions; and capturing the digital image of the stained sample while the stained sample is illuminated simultaneously from the subset of the plurality of directions.
Within the context of this disclosure, the wording “subset of the plurality of directions” should be construed as more than one direction of the plurality of directions. Which of the directions, and the number of directions, of the subset may be chosen such that illumination of the stained sample is similar to conventional microscopy illumination (e.g., brightfield illumination). Hence, the digital image of the stained sample may be similar to a digital image captured by a conventional microscope system. The subset of directions may, e.g., be a majority of, or all, directions of the plurality of directions.
An associated advantage is that the same microscope system, and in particular the same illumination system, may be used to capture the training set of digital images of the unstained sample and the digital image of the stained sample. This may, in turn, allow for a more efficient handling of the unstained sample during collection of the digital images used for training. For example, the unstained sample may, after the training set of digital images has been captured, remain stationary and be stained in the microscope system. Thus, an improved pairing between features in the stained sample with corresponding features in the unstained sample may be allowed. Using the same illumination system may further reduce economic costs associated with collecting the digital images needed for training the machine learning model.
The method of the first aspect may further comprise: receiving a reconstruction set of digital images of the stained sample, wherein the reconstruction set may be acquired by illuminating the stained sample from a plurality of directions and capturing a digital image for each of the plurality of directions; and reconstructing, using a computational imaging technique and the received reconstruction set of digital images, the digital image of the stained sample.
Within the context of this disclosure, the wording “computational imaging technique” should be construed as a computational process which can form a digital image by combining information of several digital images captured during various conditions (e.g., different illumination conditions). Examples of computational imaging technique that may be suitable in this context are Fourier ptychography and machine learning. The computational imaging technique may be a reconstruction machine learning model trained to reconstruct a digital image of the stained sample. The reconstructed digital image of the stained sample may have a resolution similar to resolutions of the digital images of the reconstruction set. The reconstructed digital image of the stained sample may have a resolution relatively higher than the resolutions of one or more of the digital images of the reconstruction set. Reconstruction of a digital image having a relatively higher resolution than the resolutions of the digital images of the reconstruction set may be allowed since the reconstruction set is acquired by illuminating the stained sample from a plurality of directions and capturing a digital image for each of the plurality of directions.
The method of the first aspect may further comprise: applying the staining agent to the unstained sample, thereby forming the stained sample; and acquiring the reconstruction set of digital images of the stained sample by: illuminating the stained sample from a plurality of directions, and capturing, for each direction of the plurality of directions, a digital image of the stained sample.
The reconstruction set of digital images may be acquired using a microscope objective and an image sensor, and wherein at least one direction of the plurality of directions may correspond to an angle larger than a numerical aperture of the microscope objective.
By illuminating the stained sample from an angle larger than the numerical aperture of a microscope objective, the digital image captured for that angle of illumination may comprise information about higher spatial frequencies, and thereby finer details of the stained sample than the microscope objective normally allows. This may, in turn, allow for the microscope objective to capture phase information associated with the stained sample and information relating to details not normally being resolvable by the microscope objective, which may be used in by the computational imaging technique when reconstructing the digital image of the stained sample. Put differently, by illuminating the stained sample from an angle larger than the numerical aperture of the microscope objective may allow for an improved reconstruction of the digital image of the stained sample. In particular, it may allow for reconstructing a digital image of the stained sample having a relatively higher resolution than the digital images of the reconstruction set.
The ground truth may comprise a digital image having a relatively higher resolution than a digital image of the training set of digital images.
An associated advantage is that the machine learning model may be trained to construct a digital image of the artificially stained sample having a relatively higher resolution than one or more of the digital images of the training set. Put differently, the trained machine learning model may be used to construct a digital image depicting an artificially stained sample having a relatively higher resolution than one or more of digital images input to the trained machine learning model. Hence, the digital images (e.g., digital images of an unstained sample) used as input for the trained machine learning model may be captured using a microscope objective having a lower numerical aperture, while still allowing the constructed digital image depicting an artificially stained sample to have a resolution similar to a digital image captured using a microscope objective having a relatively higher numerical aperture.
According to a second aspect, a method for constructing a digital image depicting an artificially stained sample is provided. The method of the second aspect comprising: receiving an input set of digital images of an unstained sample, wherein the input set of digital images is acquired by illuminating the unstained sample from a plurality of directions and capturing a digital image for each of the plurality of directions; constructing an digital image depicting an artificially stained sample by: inputting the input set of digital images into a machine learning model being trained according to the method of the first aspect, and receiving, from the machine learning model, an output comprising the digital image depicting an artificially stained sample.
By inputting the input set of digital images into a machine learning model trained according to the method of the first aspect, the process of imaging the unstained sample may be more efficient, since a digital image depicting an artificially stained sample is output from the trained machine learning model using digital images of the unstained sample. Put differently, the trained machine learning model may output a digital image depicting an artificial stained sample replicating a digital image of a stained sample. Hence, there is no need to apply a staining agent to the unstained sample prior to imaging. This may, in turn, allow for a more rapid and/or more cost-effective imaging of the sample. It may further remove the need for manually handling potentially toxic chemicals needed during staining (e.g., the staining agent or other associated chemicals), which may allow for an imaging process which is safer to humans (e.g., lab personnel).
The input set of digital images of the unstained sample may be acquired using a microscope objective and an image sensor, and wherein at least one direction of the plurality of directions may correspond to an angle larger than a numerical aperture of the microscope objective.
The input set of digital images may be acquired by illuminating the unstained sample with white light from the plurality of directions and capturing a digital image for each of the plurality of directions.
The above-mentioned features of the first aspect, when applicable, apply to this second aspect as well. In order to avoid undue repetition, reference is made to the above.
According to a third aspect a device for training a machine learning model to construct a digital image depicting an artificially stained sample is provided. The device comprising circuitry configured to execute: a first receiving function configured to receive a training set of digital images, wherein the training set of digital images is acquired by illuminating an unstained sample from a plurality of directions and capturing a digital image of the unstained sample for each of the plurality of directions; a second receiving function configured to receive a ground truth comprising a digital image of a stained sample, wherein the stained sample is formed by applying a staining agent to the unstained sample; and a training function configured to train a machine learning model to construct a digital image depicting an artificially stained sample according to the method of the first aspect using the received training set of digital images of the unstained sample and the received ground truth.
The training set may be acquired by illuminating the unstained sample with white light from the plurality of directions and capturing a digital image for each of the plurality of directions.
The above-mentioned features of the first aspect and/or the second aspect, when applicable, apply to this third aspect as well. In order to avoid undue repetition, reference is made to the above.
According to a fourth aspect, a microscope system is provided. The microscope system comprising: an illumination system configured to illuminate an unstained sample from a plurality of directions; an image sensor; at least one microscope objective arranged to image the unstained sample onto the image sensor; and circuitry configured to execute: an acquisition function configured to acquire an input set of digital images by being configured to: control the illumination system to illuminate the unstained sample from each of the plurality of directions, and control the image sensor to capture a digital image for each of the plurality of directions, and an image construction function configured to: input the input set of digital images into a machine learning model being trained according to the method of the first aspect, and receive, from the machine learning model, an output comprising a digital image depicting an artificially stained sample.
The illumination system may comprise a plurality of light sources, and wherein each light source of the plurality of light sources may be configured to emit white light. Each light source of the plurality of light sources may be configured to illuminate the unstained sample from one direction of the plurality of directions.
An associated advantage is that information about the sample pertaining to a wider range of the electromagnetic spectrum may be captured compared to an illumination system comprising narrowband light sources (e.g., monochromatic light sources).
A further associated advantage is that multispectral information about the sample may be captured with fewer light sources compared to an illumination system comprising narrowband light sources (e.g., monochromatic light sources). This, since light sources configured to emit white light may emit light of a wider wavelength range compared to narrowband light sources (e.g., monochromatic light sources) typically used in Fourier ptychography applications. Thus, compared to narrowband light sources, fewer light sources configured to emit white light may be needed to cover a particular range of the electromagnetic spectrum.
The illumination system may comprise a plurality of light sources arranged on a curved surface being concave along at least one direction along the surface, and wherein each of the plurality of light sources may be configured to illuminate the unstained sample from one of the plurality of directions.
Arranging the plurality of light sources on a curved surface may be advantageous in that the distance from each light source to a current imaging position (i.e., a position or portion of the unstained sample currently being imaged) of the microscope system may be similar. Since this distance is similar, an intensity of light emitted from each light source may be similar at the current imaging position. This may be understood as an effect of the inverse square law. Thus, the unstained sample may be illuminated by light having similar intensities for each direction in the plurality of directions, which may, in turn, allow for a more homogenous illumination of the unstained sample independent of illumination direction. It may be advantageous to configure the illumination system such that the distance from each light source to the current imaging position is large enough such that each light source may be treated as a point source. This may allow the light to be quasi-coherent at the current imaging position. Hence, the distance from each light source to the current imaging position may be chosen such that an intensity of light from each light source at the current imaging position is high enough to produce the input set of digital images.
The curved surface may be formed of facets. Put differently, the curved surface may be constructed by a plurality of flat surfaces.
An associated advantage is that the illumination system may be easier to manufacture, thereby reducing associated economic costs.
A further associated advantage is that the illumination system may be modular. It may thereby be easier to replace one or more light sources (e.g., in case they break and/or are defective).
A numerical aperture of the at least one microscope objective may be 0.4 or lower. Put differently, the at least one microscope objective may have a magnification of 20 times or lower.
An associated advantage is that a larger portion of the unstained sample may be imaged at a time compared to a microscope objective having a higher numerical aperture. This may, in turn, allow for a number of individual imaging positions needed to image a majority of the unstained sample to be reduced. Thus, a time needed to image a majority of the unstained sample may thereby be reduced. This may, in particular, be advantageous in case the machine learning model is trained to construct a digital image having a relatively larger resolution than the digital images input to the machine learning model (i.e., the digital images of the input set). Hence, the unstained sample may be imaged more quickly, while the digital image depicting the artificially stained sample may have a resolution relatively higher than what the at least one microscope objective normally allows.
The above-mentioned features of the first aspect, the second aspect, and/or the third aspect, when applicable, apply to this fourth aspect as well. In order to avoid undue repetition, reference is made to the above.
According to a fifth aspect a non-transitory computer-readable storage medium is provided. The non-transitory computer-readable storage medium comprising program code portions which, when executed on a device having processing capabilities, performs the method according to first aspect or the method according to the second aspect.
The above-mentioned features of the first aspect, the second aspect, the third aspect, and/or the fourth aspect, when applicable, apply to this fifth aspect as well. In order to avoid undue repetition, reference is made to the above.
A further scope of applicability of the present disclosure will become apparent from the detailed description given below. However, it should be understood that the detailed description and specific examples, while indicating preferred variants of the present disclosure, are given by way of illustration only, since various changes and modifications within the scope of the present disclosure will become apparent to those skilled in the art from this detailed description.
Hence, it is to be understood that the present disclosure is not limited to the particular steps of the methods described or component parts of the systems described as such method and system may vary. It is also to be understood that the terminology used herein is for purpose of describing particular embodiments only and is not intended to be limiting. It must be noted that, as used in the specification and the appended claim, the articles “a”, “an”, “the”, and “said” are intended to mean that there are one or more of the elements unless the context clearly dictates otherwise. Thus, for example, reference to “a unit” or “the unit” may include several devices, and the like. Furthermore, the words “comprising”, “including”, “containing” and similar wordings do not exclude other elements or steps.
The above and other aspects of the present disclosure will now be described in more detail, with reference to appended drawings showing variants of the present disclosure. The figures should not be considered limiting the disclosure to the specific variant; instead, they are used for explaining and understanding the present disclosure. As illustrated in the figures, the sizes of layers and regions are exaggerated for illustrative purposes and, thus, are provided to illustrate the general structures of variants of the present disclosure. Like reference numerals refer to like elements throughout.
The present disclosure will now be described more fully hereinafter with reference to the accompanying drawings, in which currently preferred variants of the disclosure are shown. The present disclosure may, however, be implemented in many different forms and should not be construed as limited to the variants set forth herein; rather, these variants are provided for thoroughness and completeness, and fully convey the scope of the present disclosure to the skilled person.
A device 10 and a method 30 for training a machine learning model to construct a digital image depicting an artificially stained sample will now be described with reference to
The memory 110 may be a non-transitory computer-readable storage medium. The memory 110 may be a random-access memory. The memory 110 may be a non-volatile memory. As is illustrated in the example of
The circuitry 100 is configured to execute a first receiving function 1100, a second receiving function 1102, and a training function 1104. As is illustrated in
The first receiving function 1100 is configured to receive a training set of digital images. The training set of digital images is acquired by illuminating an unstained sample from a plurality of directions and capturing a digital image of the unstained sample for each of the plurality of directions. Put differently, the training set of digital images of the unstained sample may be acquired by illuminating the unstained sample from a plurality of directions, and capturing, for each direction of the plurality of directions, a digital image of the unstained sample. The training set of digital images may be acquired by illuminating the unstained sample with white light from the plurality of directions and capturing a digital image for each of the plurality of directions. Put differently, the training set of digital images of the unstained sample may be acquired by illuminating the unstained sample with white light from a plurality of directions, and capturing, for each direction of the plurality of directions, a digital image of the unstained sample. The unstained sample may be a sample without applied staining agent. The unstained sample may have a relatively lower contrast compared to a stained sample. The contrast of the unstained sample may be low such that it may be difficult, or even impossible, to image features of the unstained sample using conventional microscopy. For example, white blood cells are typically almost transparent and may therefore be difficult to image without staining the sample. The first receiving function 1100 may be configured to receive the training set of digital images via the transceiver 130. For example, the training set of digital images may be captured using an external microscope system and then transmitted to the transceiver 130 of the device 10. As a further example, the device 10 may form part of a microscope system and the training set of digital images may be received from an image sensor of the microscope system.
The memory 110 may be configured to store the training set of digital images, and the first receiving function 1100 may be configured to receive the training set of digital images from the memory 110. The training set of digital images may be acquired using a microscope objective and an image sensor. The unstained sample may be sequentially illuminated from at least one direction of the plurality of directions. Each digital image of the training set of digital images may be captured when the unstained sample is illuminated from at least one direction of the plurality of directions. Thus, the unstained sample may be illuminated from a single direction of the plurality of direction or from several directions of the plurality of directions at the same time.
At least one direction of the plurality of directions may correspond to an angle larger than a numerical aperture of the microscope objective. The numerical aperture of the microscope objective may be a dimensionless number associated with a range of angles over which the microscope objective accepts light. Hence, a direction larger than the numerical aperture may be understood as a direction corresponding to an angle larger than the range of angles over which the microscope objective is configured to accept light when it is used for brightfield microscopy. Since the unstained sample may be illuminated from a plurality of different directions and a digital image may be captured for each of the plurality of directions, information regarding finer details of the unstained sample may be captured than what normally may be resolvable by the microscope objective used to image the unstained sample. This can be understood as different portions of Fourier space (i.e., the spatial frequency domain) associated with the unstained sample are imaged for different illumination directions. This technique may be known in the art as Fourier ptychography. Generally in Fourier ptychography, high spatial frequencies in Fourier space associated with a sample (e.g., the unstained sample) are sampled when that sample is illuminated from a direction corresponding to a large angle of incidence. Hence, in case the unstained sample is illuminated from a direction corresponding to an angle larger than the numerical aperture of the microscope objective, even higher spatial frequencies of the Fourier space may be sampled. This is possible since the light is scattered by the unstained sample, and a portion of the light scattered by the unstained sample may be collected by the microscope objective. This illumination technique may further allow information regarding a refractive index associated with the unstained sample to be captured by the at least one microscope objective. This can be understood as an effect of refraction of light being dependent on an angle of incident for light illuminating the unstained sample and the refractive index of the unstained sample. Information regarding the refractive index of the unstained sample may, in turn, allow phase information (typically referred to as quantitative phase within the art) associated with the unstained sample to be determined. Since the plurality of digital images comprises information associated with one or more of fine details of the unstained sample, a refractive index associated with the unstained sample, and phase information associated with the unstained sample, this information may be used in the training of the machine learning model, which may, in turn, allow for a machine learning model being trained to more accurately construct the digital image depicting the artificially stained sample than what would be allowed in case the plurality of digital images were captured from only one direction. Put differently, it may allow for a machine learning model being trained to construct the digital image depicting an artificially stained sample more accurately than what would be allowed in case the plurality of digital images were captured from only one direction or using a conventional microscope (e.g., using brightfield illumination).
It is to be understood that the information relating to one or more of finer details of the unstained sample, refractive index associated with the unstained sample, and phase information associated with the unstained sample may be captured by illuminating the unstained sample from more than one of the plurality of different directions at a time, for example from a subset of the plurality of directions. The subset of the plurality of directions may comprise directions corresponding to different portions of the Fourier space of the unstained sample. The different portions of the Fourier space of the unstained sample may be partially overlapping or non-overlapping. Thus, a microscope objective having a relatively lower magnification may be used while still being able to capture information related to fine details of the unstained sample. For example, by illuminating the unstained sample from the plurality of directions, a microscope objective having a numerical aperture of 0.4 may capture information relating to fine details to the same extent that a microscope objective being used in conventional microscopy (e.g., using brightfield illumination) and having a numerical aperture of 1.25. Put differently, by illuminating the unstained sample from the plurality of directions, a microscope objective having a magnification of 20 times may capture information relating to fine details to the same extent that a microscope objective being used in conventional microscopy (e.g., using brightfield illumination) and having a magnification of 100 times. It is to be understood that the above magnifications and numerical apertures are examples only, and that various embodiments of the present disclosure may be implemented for other magnifications and/or numerical apertures as well. A suitable microscope system comprising a microscope objective and an image sensor will be described in connection with
The second receiving function 1102 is configured to receive a ground truth comprising a digital image of a stained sample. The ground truth may be information that is known to be real and/or true. In this context, since the machine learning model is trained to construct a digital image depicting an artificially stained sample, the ground truth may represent a stained sample. For example, the ground truth may comprise a digital image of a stained sample. The ground truth may be received via the transceiver 130. For example, the ground truth may be formed on a different device and/or stored on a different device and transmitted to the device 10 via the transceiver 130. The stained sample is formed by applying a staining agent to the unstained sample. The stained sample may be formed such that relative positions of features (e.g., objects) within the stained sample may remain substantially undisturbed compared to the unstained sample. Put differently, features of the unstained sample and corresponding features of the stained sample may be at a same relative position (i.e., relative to the unstained sample or the stained sample). This may, in turn, allow features in the stained sample to be correlated (or paired) with corresponding features in the unstained sample. Alternatively, or additionally, the machine learning model may be trained using CycleGAN. As is known within the art, CycleGAN is a technique capable of training image-to-image translation models without paired examples (i.e., without having paired examples of features in the stained sample and features in the unstained sample). This may be advantageous, since it may allow the unstained sample to be stained in an easier manner, since the machine learning model may be trained without features in the stained sample being correlated (or paired) with corresponding features in the unstained sample.
The staining agent may be one or more chemical agents configured to increase contrast of the unstained sample (e.g., by changing the color of the unstained sample) when applied to the unstained sample. The staining agent may be chosen depending on the type of sample and/or what type of objects within the sample is of interest. For example, within hematology, samples (e.g., blood smears) are typically stained by using May Grünewald-Giemsa (MGG), Wright-Giemsa (WG), or Wright. As a further example, within pathology, samples are typically stained by using Hematoxylin-cosin (H&E). Different staining agents may increase the contrast for different objects and/or substances within a sample. For example, some staining agents may be adapted increase the contrast of proteins, etc. Hence, which staining agent to be used may be chosen depending on the sample and/or on the type of objects of interest in the sample. The ground truth comprising the digital image of the stained sample may be formed by acquiring a digital image of the stained sample. The digital image of the stained sample may be acquired by illuminating the stained sample simultaneously from a subset of the plurality of directions. The subset of the plurality of directions may be more than one direction of the plurality of directions. Which directions, and the number of directions, of the subset may be chosen such that illumination of the stained sample is similar to conventional microscopy illumination (e.g., brightfield illumination). The subset of directions may, e.g., be a majority of, or all, directions of the plurality of directions. The digital image of the stained sample may be captured while the stained sample is illuminated simultaneously from the subset of the plurality of directions. Hence, the digital image of the stained sample may be similar to a digital image captured by a conventional microscope system using, for instance, brightfield illumination. The training set of digital images may be acquired prior to the digital image of the ground truth. This is advantageous since the training set may comprise digital images of the unstained sample, and the ground truth may comprise a digital image of the same sample with applied staining agent.
The digital image of the stained sample may be acquired in other manners than described above. For example, the digital image of the stained sample may be reconstructed using a computational imaging technique. Here, “computational imaging technique” should be understood as a computational process which can form a digital image by combining information of several digital images captured during various conditions (e.g., different illumination conditions). Examples of computational imaging technique that may be suitable in this context are Fourier ptychography and machine learning. The computational imaging technique may be a reconstruction machine learning model trained to reconstruct a digital image of the stained sample. The reconstructed digital image of the stained sample may have a resolution similar to resolutions of the digital images of the reconstruction set. The reconstructed digital image of the stained sample may have a resolution relatively higher than the resolutions of one or more of the digital images of the reconstruction set. Reconstruction of a digital image having a relatively higher resolution than the resolutions of the digital images of the reconstruction set may be allowed since the reconstruction set is acquired by illuminating the stained sample from a plurality of directions and capturing a digital image for each of the plurality of directions. Hence, the third receiving function 1108 may be configured to receive a reconstruction set of digital images of the stained sample. The reconstruction set may be received via the transceiver 130. For example, the reconstruction set may be formed on a different device and/or stored on a different device and transmitted to the device 10 via the transceiver 130. The reconstruction set may be acquired by illuminating the stained sample from a plurality of directions and capturing a digital image for each of the plurality of directions. Put differently, the reconstruction set of digital images of the stained sample may be acquired by illuminating the stained sample from a plurality of directions, and capturing, for each direction of the plurality of directions, a digital image of the stained sample. The reconstruction set of digital images may be acquired using a microscope objective and an image sensor, and at least one direction of the plurality of directions may correspond to an angle larger than a numerical aperture of the microscope objective. Thus, the reconstruction set of digital images of the stained sample may be acquired in a similar manner as the training set of digital images of the unstained sample, and the above description in connection to the training set therefore applies, mutatis mutandis, to the reconstruction set of digital images of the stained sample. The reconstruction function 1106 may be configured to reconstruct, using the computational imaging technique and the received reconstruction set of digital images, the digital image of the stained sample. By illuminating the stained sample from an angle larger than the numerical aperture of a microscope objective, the digital image captured for that angle of illumination may comprise information about higher spatial frequencies, and thereby finer details of the stained sample than the microscope objective normally allows. The illumination technique may further allow for the microscope objective to capture phase information associated with the stained sample and information relating to details not normally being resolvable by the microscope objective, which may be used in by the computational imaging technique when reconstructing the digital image of the stained sample. Put differently, by illuminating the stained sample from an angle larger than the numerical aperture of the microscope objective may allow for an improved reconstruction of the digital image of the stained sample. In particular, it may allow for reconstructing a digital image of the stained sample having a relatively higher resolution than the digital images of the reconstruction set. Hence, the digital image of the stained sample may be reconstructed from the reconstruction set of digital images of the stained sample using the computational imaging technique and the received reconstruction set of digital images of the stained sample.
The ground truth may comprise a digital image having a relatively higher resolution than a digital image of the training set of digital images. The digital image of the stained sample may be a digital image having a relatively higher resolution than each digital image of the training set of digital images. As discussed above, the digital image of the stained sample may be reconstructed using the computational imaging technique, and the reconstructed digital image may have a resolution relatively higher than the resolution of a digital image of the reconstruction set of digital images. Thus, the computational imaging technique may be configured to reconstruct the digital image of the stained sample such that the resolution of the reconstructed digital image is higher than one or more digital images of the training set of digital images of the unstained sample. Alternatively, or additionally, the digital image of the stained sample may be captured using a microscope objective having a relatively higher magnification than the microscope objective used to capture one or more digital images of the training set of digital images of the unstained sample. Put differently, since the microscope objective used to capture the digital image of the stained sample may have a relatively larger magnification, that digital image may be depict a smaller area of the stained sample compared to an image captured using the microscope objective used to capture the training set of digital images. For this reason, the digital image of the stained sample may be formed from a plurality of individual digital images of the stained sample. For example, by combining (e.g., by stitching) the plurality of individual digital images. The digital image of the stained sample may thereby depict an area of the stained sample comparable to an area of the unstained sample imaged by the microscope objective used to capture the training set of digital images. Alternatively, each digital image of the training set may be cropped such that an area of the unstained sample depicted in each cropped digital image of the training set may be similar and/or comparable to an area of the stained sample depicted in the digital image of the stained sample.
The training function 1104 is configured to train the machine learning model to construct a digital image depicting an artificially stained sample using the received training set of digital images of the unstained sample and the received ground truth. The machine learning model may be a convolutional neural network. The machine learning model may be a convolutional neural network suitable for digital image construction. The training function 1104 is configured to receive the ground truth comprising the digital image of the stained sample. The training function 1104 may receive the ground truth from the second receiving function 1102. The training function 1104 is further configured to acquire the training set of digital images of the unstained sample. As explained previously, the training set of digital images of the unstained sample is acquired by illuminating the unstained sample (possibly with white light) from a plurality of directions and capturing a digital image for each of the plurality of directions. The training function 1104 is further configured to train the machine learning model to construct a digital image depicting an artificially stained sample using the training set of digital images of the unstained sample and the received ground truth. Hence, the present disclosure allows for training the machine learning model to construct digital images replicating a stained sample using a plurality of digital images of an unstained sample. Hence, by using the trained machine learning model, a digital image depicting a stained sample (i.e., the artificially stained sample) may be constructed without having to apply a staining agent to the unstained sample. In case the ground truth comprises a digital image of the stained sample having a relatively higher resolution than one or more digital images of the training set of digital images of the unstained sample, the machine learning model may be trained to construct a digital image of the artificially stained sample having a relatively higher resolution than one or more of the digital images of the digital images input to the machine learning model (e.g., the training set). The training function 1104 may be configured to train the machine learning model iteratively and/or recursively until a difference between an output of the machine learning model (i.e., the digital image depicting an artificially stained sample) and the ground truth (e.g., the digital image depicting the stained sample) is smaller than a predetermined threshold. Hence, the training function 1104 may train the machine learning model to correlate the training set of digital images of the unstained sample to the ground truth (e.g., the digital image depicting the stained sample). A smaller difference between the output of the machine learning model and the ground truth may indicate a higher accuracy of the digital image depicting the artificially stained sample provided by the machine learning model. Hence, preferably, the difference between the output of the machine learning model and the ground truth may be minimized. A skilled person realizes that minimizing functions (e.g., the loss function) may be associated with tolerances. For example, the loss function may be regarded as being minimized even though the minimized loss function has a value which is not a local and/or global minimum.
The machine learning model may be trained using a plurality of training sets and a plurality of corresponding ground truths. Put differently, the machine learning model may be trained using a plurality of different samples. This may allow for an improved training of the machine learning model to construct the digital image depicting an artificially stained sample. The machine learning model may be trained to construct digital images of artificially stained samples for a plurality of different sample types. In such case, the machine learning model may, for each sample type, be trained using a training set of digital images of an unstained sample of the sample type, and a corresponding ground truth associated with a stained sample of the respective sample type (e.g., a digital image of a stained sample of the sample type). This may allow the trained machine learning model to construct digital images depicting artificially stained samples of different sample types.
A microscope system 20 and a method 40 for constructing a digital image depicting an artificially stained sample will now be described with reference to
Even though the image sensor 270 is illustrated on its own in
The illumination system 260 is configured to illuminate the unstained sample 292 from a plurality of directions. As is illustrated in
The at least one microscope objective 280 is arranged to image the unstained sample 292 onto the image sensor 270. The at least one microscope objective 280 may comprise a first microscope objective. A numerical aperture of the first microscope objective may be 0.4 or lower. Put differently, the first microscope objective may have a magnification of 20 times or lower. Hence, a larger portion of the unstained sample 292 may be imaged at a time compared to a microscope objective having a relatively higher numerical aperture. This may, in turn, allow for a number of individual imaging positions needed to image a majority of the unstained sample 292 to be reduced. Thus, a time needed to image a majority of the unstained sample 292 may thereby be reduced. This may, in particular, be advantageous in case the machine learning model is trained to construct a digital image having a relatively larger resolution than the digital images input to the machine learning model (i.e., the digital images of the input set). Hence, the unstained sample 292 may be imaged more quickly, while the digital image depicting the artificially stained sample may have a resolution relatively higher than what the at least one microscope objective normally allows. The first microscope objective may be used when capturing the digital images of the unstained sample 292. The digital images captured using the first microscope objective may be used when training the machine learning model (i.e., the training set of digital images of the unstained sample) and/or when constructing the digital image depicting an artificially stained sample using the trained machine learning model. It is further to be understood that the first microscope objective may be used to form the ground truth. Put differently, the first microscope objective may be used to capture a digital image of a stained sample to be used as ground truth when the machine learning model is trained.
The at least one microscope 280 may comprise a second microscope objective having a magnification of 100 times and/or a numerical aperture of 1.25. The second microscope objective may be used to acquire a digital image having a relatively higher resolution than a digital image acquired using the first microscope objective. Hence, the second microscope objective may be used to acquire digital images (e.g., the digital image of the stained sample) for use when training the machine learning model. Put differently, digital images captured using the second microscope objective may be used to form the ground truth used when training the machine learning model.
The numerical aperture and magnification of the first microscope objective and/or the second microscope objective are examples only and may be chosen depending on, e.g., a type of the unstained sample 292. For example, the numerical aperture of the first microscope objective may have a magnification of 10 times and/or a numerical aperture of 0.25. It is to be understood that the microscope system 20 may comprise further optics which may be used together with the at least one microscope objective 280 to image the unstained sample 292 onto the image sensor 270. For example, the microscope system may, as illustrated in the example of
The circuitry 200 is configured to execute an acquisition function 2100 and an image construction function 2102.
The acquisition function 2100 is configured to acquire an input set of digital images by being configured to control the illumination system 260 to illuminate the unstained sample 292 from each of the plurality of directions. The input set of digital images of the unstained sample 292 may thereby be acquired using the at least one microscope objective 280 and the image sensor 270. At least one direction of the plurality of directions 262 may correspond to an angle larger than a numerical aperture 282 of the microscope objective 280. The acquisition function 2100 may be configured to acquire the input set of digital images by being configured to control the illumination system 260 to illuminate the unstained sample 292 with white light from each of the plurality of directions.
The acquisition function 2100 is further configured to control the image sensor 270 to capture a digital image for each of the plurality of directions. Put differently, the acquisition function 2100 may be configured to receive the input set of digital images of an unstained sample 292. The input set of digital images is acquired by illuminating the unstained sample 292 from a plurality of directions 262 and capturing a digital image for each of the plurality of directions 262. The input set of digital images may be acquired by illuminating the unstained sample 292 with white light from the plurality of directions 262 and capturing a digital image for each of the plurality of directions 262. As discussed previously, at least one direction of the plurality of directions may correspond to an angle larger than a numerical aperture 282 of the microscope objective 280. For example, direction 2620 in
The image construction function 2102 is configured to input the input set of digital images into a machine learning model being trained as described in connection with
A skilled person would be aware of machine learning, and in particular as to how a machine learning model may be trained and/or how a trained machine learning model may be used. However, in brief, the machine learning model may be a type of supervised machine learning model, for example a network such as U-net or Pix2pix. The machine learning model may be a transformer-based network such as SwinIR. The machine learning model may be a convolutional neural network. The machine learning model may be trained to predict a desired output using example input training data and a ground truth, i.e. the “correct” or “true” output. Put differently, the ground truth may be used as a label for the input training data. The input training data may comprise data pertaining to different outcomes, and each input training data may thereby be associated with a ground truth associated with that particular input training data. Hence, each input training data may be labelled with an associated ground truth (i.e., “correct” or “true” output). The machine learning model may comprise a plurality of layers of neurons, and each neuron may represent a mathematical operation which is applied to the input training data. Typically, the machine learning model comprises an input layer, one or more hidden layers, and an output layer. The first layer may be referred to as the input layer. The output of each layer (except the output layer) in the machine learning model may be fed to a subsequent layer, which in turn produces a new output. The new output may be fed to a further subsequent layer. The output of the machine learning model may be an output of the output layer. The process may be repeated for all layers in the machine learning model. Typically, each layer further comprises an activation function. The activation function may further define the output of a neuron of the layer. For example, the activation function may ensure that the output from a layer is not too large or too small (e.g., tending towards positive or negative infinity). Further, the activation function may introduce non-linearity into the machine learning model. During the training process, weights and/or biases associated with the neurons of the layers may be adjusted until the machine learning model produces predictions for the input training data that reflect the ground truth. Each neuron may be configured to multiply the input to the neuron with a weight associated with that neuron. Each neuron may be further configured to add a bias associated with that neuron to the input. Put differently, an output from a neuron may be a sum of the bias associated with the neuron and a product of the weight associated with the neuron and the input. The weights and biases may be adjusted in a recursive process and/or an iterative process. This may be known as backpropagation within the art. A convolutional neural network may be a type of neural network comprising one or more layers that represents a convolution operation. In this context, the input training data comprises digital images. A digital image may be represented as matrix (or as an array), and each element in the matrix (or array) may represent a corresponding pixel of the digital image. The value of an element may thereby represent a pixel value of the corresponding pixel in the digital image. Hence, the input and output to the machine learning model may be numerical (e.g., a matrix or an array) representing digital images. In this context, the input is a set of digital images (i.e., the training set or the input set). Thus, the input to the machine learning model may be a plurality of matrices, or a three-dimensional matrix. It is to be understood that the machine learning model may take further input during training. An example of such input may be a type of staining agent used when forming the stained sample. Such input may then be used when constructing a digital image depicting an artificially stained sample using the trained machine learning model. Further, in this context, the output is a digital image. Thus, the output of the machine learning model may be a matrix representing the constructed digital image depicting an artificially stained sample.
The person skilled in the art realizes that the present disclosure by no means is limited to the preferred variants described above. On the contrary, many modifications and variations are possible within the scope of the appended claims.
For example, a plurality of machine learning models may be trained for different staining agents. Hence, a type of the artificial staining of the constructed digital image may be chosen by inputting the input set of digital images into the machine learning model trained using that type of staining agent. It is further to be understood that, instead of training a plurality of machine learning models for different staining agents, a single machine learning model may be trained for a plurality of different samples and types of staining agents. For instance, the type of staining agent may be used as a further input to the machine learning model during training. Hence, the trained machine learning model may thereby take the input set of digital images and the type of staining agent to be used for the artificial staining as inputs. The trained machine learning model may be used to output a digital image depicting an artificially stained sample replicating a sample being stained with the type of staining agent used as input to the trained machine learning model.
It is to be understood that, even though the construction of the digital image depicting an artificially stained sample is described in connection with the microscope system of
Additionally, variations to the disclosed variants can be understood and effected by the skilled person in practicing various embodiments of the present disclosure, from a study of the drawings, the disclosure, and the appended claims. It will be appreciated that other modifications or adaptations of the methods, devices, systems, and/or or specific structures described herein may become apparent to those skilled in the art. The embodiments specifically illustrated and/or described herein are provided merely to exemplify particular applications of the present disclosure. These descriptions and drawings should not be considered in a limiting sense, as it is understood that the present disclosure is in no way limited to only the disclosed embodiments. All such modifications, adaptations, or variations are considered to be within the spirit and scope of the present disclosure, and within the scope of the appended claims.
Embodiment 1. A method for training a machine learning model to construct a digital image depicting an artificially stained sample, the method comprising: receiving a training set of digital images of an unstained sample, wherein the training set of digital images is acquired by illuminating the unstained sample from a plurality of directions and capturing a digital image for each of the plurality of directions; receiving a ground truth comprising a digital image of a stained sample, wherein the stained sample is formed by applying a staining agent to the unstained sample; and training the machine learning model to construct a digital image depicting an artificially stained sample using the received training set of digital images of the unstained sample and the received ground truth.
Embodiment 2. The method according to embodiment 1, further comprising: acquiring the training set of digital images of the unstained sample by: illuminating the unstained sample from a plurality of directions, and capturing, for each direction of the plurality of directions, a digital image of the unstained sample.
Embodiment 3. The method according to embodiment 1 or 2, wherein the training set of digital images is acquired by illuminating the unstained sample with white light from the plurality of directions and capturing a digital image for each of the plurality of directions.
Embodiment 4. The method according to any one of embodiments 1-3, wherein the training set of digital images is acquired using a microscope objective and an image sensor, and wherein at least one direction of the plurality of directions corresponds to an angle larger than a numerical aperture of the microscope objective.
Embodiment 5. The method according to any one of embodiments 1-4, further comprising: applying a staining agent to the unstained sample, thereby forming a stained sample; and forming the ground truth comprising the digital image of the stained sample by: acquiring a digital image of the stained sample.
Embodiment 6. The method according to embodiment 5, wherein the act of acquiring the digital image of the stained sample comprises: illuminating the stained sample simultaneously from a subset of the plurality of directions; and capturing the digital image of the stained sample while the stained sample is illuminated simultaneously from the subset of the plurality of directions.
Embodiment 7. The method according to any one of embodiments 1-4, further comprising: receiving a reconstruction set of digital images of the stained sample, wherein the reconstruction set is acquired by illuminating the stained sample from a plurality of directions and capturing a digital image for each of the plurality of directions; and reconstructing, using a computational imaging technique and the received reconstruction set of digital images, the digital image of the stained sample.
Embodiment 8. The method according to embodiment 7, further comprising: applying the staining agent to the unstained sample, thereby forming the stained sample; and acquiring the reconstruction set of digital images of the stained sample by: illuminating the stained sample from a plurality of directions, and capturing, for each direction of the plurality of directions, a digital image of the stained sample.
Embodiment 9. The method according to embodiment 7 or 8, wherein the reconstruction set of digital images is acquired using a microscope objective and an image sensor, and wherein at least one direction of the plurality of directions corresponds to an angle larger than a numerical aperture of the microscope objective.
Embodiment 10. The method according to any one of embodiments 1-9, wherein the ground truth comprises a digital image having a relatively higher resolution than a digital image of the training set of digital images.
Embodiment 11. A method for constructing a digital image depicting an artificially stained sample, the method comprising: receiving an input set of digital images of an unstained sample, wherein the input set of digital images is acquired by illuminating the unstained sample from a plurality of directions and capturing a digital image for each of the plurality of directions; constructing a digital image depicting an artificially stained sample by: inputting the input set of digital images into a machine learning model being trained according to the method of any one of embodiments 1-10, and receiving, from the machine learning model, an output comprising the digital image depicting an artificially stained sample.
Embodiment 12. The method according to embodiment 11, wherein the input set of digital images of the unstained sample is acquired using a microscope objective and an image sensor, and wherein at least one direction of the plurality of directions corresponds to an angle larger than a numerical aperture of the microscope objective.
Embodiment 13. The method according to embodiment 11 or 12, wherein the input set of digital images is acquired by illuminating the unstained sample with white light from the plurality of directions and capturing a digital image for each of the plurality of directions.
Embodiment 14. A device for training a machine learning model to construct a digital image depicting an artificially stained sample, the device comprising circuitry configured to execute:
Embodiment 15. The device according to embodiment 14, where in the training set is acquired by illuminating the unstained sample with white light from the plurality of directions and capturing a digital image for each of the plurality of directions.
Embodiment 16. A microscope system comprising: an illumination system configured to illuminate an unstained sample from a plurality of directions; an image sensor; at least one microscope objective arranged to image the unstained sample onto the image sensor; and circuitry configured to execute: an acquisition function configured to acquire an input set of digital images by being configured to: control the illumination system to illuminate the unstained sample from each of the plurality of directions, and control the image sensor to capture a digital image for each of the plurality of directions, and an image construction function configured to:
input the input set of digital images into a machine learning model being trained according to the method of any one of embodiments 1-9, and receive, from the machine learning model, an output comprising a digital image depicting an artificially stained sample.
Embodiment 17. The microscope system according to embodiment 16, wherein the illumination system comprises a plurality of light sources, and wherein each light source of the plurality of light sources is configured to emit white light.
Embodiment 18. The microscope system according to embodiment 16 or 17, wherein the illumination system comprises a plurality of light sources arranged on a curved surface being concave along at least one direction along the surface, and wherein each of the plurality of light sources is configured to illuminate the unstained sample from one of the plurality of directions.
Embodiment 19. The microscope system according to embodiment 18, wherein the curved surface is formed of facets.
Embodiment 20. The microscope system according to any one of embodiments 16-19, wherein a numerical aperture of the at least one microscope objective is 0.4 or lower.
Embodiment 21. A non-transitory computer-readable storage medium comprising program code portions which, when executed on a device having processing capabilities, performs the method according to any one of embodiments 1-10 or the method according to any one of embodiments 11-13.
| Number | Date | Country | Kind |
|---|---|---|---|
| 22151198.3 | Jan 2022 | EP | regional |
This application is a U.S. national stage application under 35 U.S.C. 371 of International Application No. PCT/EP2023/050636, filed Jan. 12, 2023, which claims priority to and the benefit of European Application No. EP 22151198.3, filed Jan. 12, 2022, the contents of which are incorporated into the present application by reference in their entireties.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2023/050636 | 1/12/2023 | WO |
