DEVICE FOR MEASURING INTENSIVE AUTOFLUORESCENCE OF BIOLOGICAL SAMPLES AND METHODS FOR USING SAME - Patent application

Abstract

The present invention refers to a device for measuring the intrinsic autofluorescence of molecules present in a biological sample, said device comprising: a) a light sheet, virtual light sheet or light line delivery path module comprising a laser generator capable of generating ultrashort pulses with durations between 1 microsecond and 1 attosecond and optical elements for shaping the laser light into a light sheet, virtual light sheet or light line; b) an imaging chamber comprising an imaging dish in which the biological sample is placed, a translation stage, an illumination objective and a collection objective both immersed in an immersion fluid in a sealed immersion chamber, both objectives being positioned below the imaging dish, said immersion chamber being placed within an incubation chamber, allowing environmental control; and c) a fluorescence detection path comprising optical elements, light splitting elements such as diffractive elements or dichroic mirrors, an array detector, and a device control and signal processing unit, said fluorescence detection path having means for collecting spectra of the fluorescence of the biological sample building a multi- or hyperspectral data cube. In a second aspect, the present invention relates to a method for assessing the quality of a biological sample by measuring the intrinsic autofluorescence of molecules present in said biological sample, using the device described above.

Description

[Technical field]

[0001] The present invention relates to a device, in particular a microscope, and an imaging method using same, capable of exciting and measuring intrinsic autofluorescence from molecules present in various biological samples, such as autofluorescent molecules including metabolites, and distinguishing and quantifying the presence of these molecules by their spectral signatures in a non-invasive manner, and based thereon assessing the quality of said biological samples. The device and method are particularly useful for classifying the quality of pre-implantation human embryos. [Background technology]

[0002] Autofluorescence is the spontaneous emission of light in the ultraviolet (UV), visible and near infrared (IR) spectral range from a biological substrate when it is excited with light (photons) of a suitable wavelength, well known as its resonant frequency. This is possible thanks to the single photon absorption effect (SPA). SPA is a linear absorption process in which a single photon alone has enough energy to excite an electron in an atom or molecule from one energy state, usually the ground state, to an excited state. After excitation, the electron loses energy and returns to a stable state, emitting a photon of light in the process that can be collected and detected by a detector. Some autofluorescent molecules present in organic materials are NADH (free and bound), retinol, folic acid, FAD, among others.

[0003] However, this SPA effect has some limitations in microscopy terms. Lasers excite molecules throughout the sample with light both in and out of the focal plane (e.g., in a confocal microscope). This results in greater defocusing of the excitation, which leads to faster photobleaching, a gradual decrease in the fluorescent ability of most fluorochromes, phototoxicity (where the imaged organism / cell is damaged), and signal superposition from nearby molecules that can blur the image and make it difficult to interpret. Most used autofluorescent molecules or fluorochromes have excitation spectra ranging from 300 to 600 nm. Light from the visible spectrum scatters heavily in biological tissues, significantly limiting the depth it can penetrate with enough power to excite fluorochromes; for example, single-photon confocal microscopes can only penetrate about 200 μm into biological tissues. The wavelength of light is proportional to its energy level, the shorter the wavelength, the more energy it has. This is a linear relationship, and a photon with a wavelength of 400 nm has twice the energy of a photon with a wavelength of 800 nm. For biological imaging it is desirable to avoid short wavelengths.

[0004] The two-photon absorption effect (TPA) would overcome the aforementioned problems, i.e. deep penetration, sharp focusing, and reduced optical damage. TPA is a nonlinear effect in which two photons of the same or different frequency can excite an electron in a molecule from one energy state (usually the ground state) to a higher energy state in a single quantum event. After excitation, the electron reduces its energy and returns to a stable state, emitting a photon of light in the process. For this to occur, the two photons must travel within a femtosecond (10 -15 The light has to hit the molecule within a few seconds of hitting it. This requires a focused laser that can generate very fast pulses of light with high power.

[0005] A key advantage of two-photon microscopy is the ability to restrict excitation to a thin focal volume in thick samples. The object focal point is the only space with sufficient photon density to ensure that two photons are delivered to the fluorophore at the same time. The requirement for nearly simultaneous absorption of the two photons means that excitation is achieved only near the focal plane where the laser light is most concentrated. This has two distinct advantages. First, there is little tissue damage to areas above and below the focal plane that are not imaged. In addition, there is no out-of-focus light that can interfere and blur the image, making it difficult to interpret.

[0006] As commented before, light from the visible spectrum scatters heavily in living tissue, significantly limiting the depth to which it can penetrate with sufficient power to excite fluorescent dyes. Most commonly used fluorescent dyes have excitation spectra ranging from 300 to 600 nm. Thus, the wavelengths used to excite the same dye with two photons are usually in the range of 800 to 1,200 nm, within the infrared spectrum. The infrared light used for two-photon microscopy scatters considerably less than visible light in organic tissue. This means that infrared laser light has enough power to excite fluorescent dyes up to a depth of about 1 mm in living tissue.

[0007] On the other hand, fundamental for biological imaging is to develop efficient microscopes that can minimize light exposure to the sample, increase imaging speed, and reduce phototoxicity while at the same time allowing imaging with high spatial and temporal resolution. A strategy to improve some of these trade-offs is a set of techniques collectively referred to as light sheet microscopy, also called selective plane illumination microscopy (SPIM). As a general overview, SPIM decouples the illumination and detection pathways, allowing for more efficient illumination / detection strategies. This decoupling expands the range of microscopy strategies that can be set up to acquire images with optimal architectures that fit the requirements of many different types of samples.

[0008] The light sheet is typically formed by a cylindrical lens that forms a sheet of light at and around the focal point. Excitation is restricted to the focal plane, thus limiting additional phototoxic damage derived from out-of-focus light entering the detector and unnecessary illumination. An alternative method to achieve a light sheet involves using a laser light formed with a low numerical aperture objective lens and rapidly scanning the beam laterally. In both cases, a planar area of ​​the sample is illuminated and simultaneously projected into a camera that collects photons of the entire plane in one shot. This fast imaging strategy combined with suppressed light illumination significantly reduces phototoxicity during SPIM imaging.

[0009] Light sheets (both static and scanned) have a hyperbolic profile in the x,z plane, which does not produce an exact planar sheet, but rather gives an approximation of the planar illumination profile. Several strategies can be employed to shape the beam to perform light sheet fluorescence. These include, but are not limited to, (a) planar light sheets, (b) digitally scanned light sheets (DSLMs), which simulate a light sheet by line scanning a typical diffraction-limited Gaussian-distributed collimated laser beam across the sample in the lateral and axial directions, (c) two-photon excitation DSLMs (2PE DSLMs), which improve tissue penetration, (d) 2PE Bessel beam light sheets, which produce thin illumination and a large field of view (FOV), and (e) Airy beams in either continuous or 2P mode, which further produce thin layers with small side lobes useful for achieving isotropic resolution in 3D. More complex structured illumination methods include lattice light sheets and LITE, lateral interference tilted excitation. Each of these methods and additional methods provide planar illumination with some advantages and disadvantages based on tradeoffs between basic imaging parameters, FOV, axial resolution, and deep penetration.

[0010] As diverse as the types of planar beams are the various SPIM modalities. Several architectures can be assembled to optimize the illumination / imaging strategy of the sample to be imaged. The most common setup consists of illumination and detection objectives in an orthogonal configuration, but additional objectives, cameras, or spatial configurations can be designed to increase the illumination capabilities of the system. In L SPIM, a light sheet is formed by using a cylindrical lens in conjunction with a large Gaussian beam. A pre-shaped (flattened) beam is focused onto the back focal plane of the lens and sent through the lens to form the sheet. T SPIM involves having two sheets from opposite sides that can be superimposed. The second sheet enhances the imaging quality, especially in thick tissues, by compensating for light scattering from the first sheet. Both sheets can be perfectly superimposed. Multi-view SPIM combines two excitation and two imaging paths, with the cameras synchronized with each other. To obtain optimal resolution, the sample is imaged twice, the first at 0° and the second time typically rotated by 45°. The resolution when an object is imaged in the XY direction can be much higher compared to the resolution when the object is imaged in the Z direction. Inverted SPIM (iSPIM) and dual-view inverted SPIM (diSPIM) involve placing the objective lens in a V-shape above the sample. The sample rests on a platform (a petri dish or similar). The objective lens is moved up and down to acquire a z-stack, or the sample is in a position relative to the objective lens. In essence, iSPIM is an L SPIM, but taken at a different angle and the sample is fixed to the stage. In diSPIM, excitation and emission occur through both lenses. Excitation alternates between the two sides, as does imaging.

[0011] Single-objective SPIM includes highly inclined and laminated optical sheet microscopy (HILO) and oblique plane microscopy (OPM). OPM provides a small, fixed light sheet through which a sample can be scanned using a typical piezo z-stage for volumetric imaging. However, this requires that at some point along the optical train, the angled image field (due to oblique illumination) be corrected. This is achieved by projecting the image for detection by a second objective, placed at an angle that corrects for the oblique geometry.

[0012] Spectral imaging is imaging that uses multiple bands on the electromagnetic spectrum. There are two main subcategories of spectral imaging called multispectral and hyperspectral. Multispectral captures images in a small number of spectral bands, typically between 3 and 15. Hyperspectral imaging combines two methodologies: spectroscopy (the science of acquiring and describing the spectral properties of materials) and imaging (the process of forming a visual representation of anything by scanning with a detector and an electromagnetic beam). The goal of hyperspectral imaging is to obtain a spectrum for each pixel in an image of a scene, with the goal of finding objects, identifying materials, and detecting processes. The technique splits an image into wavelength bands that extend beyond the visible range.

[0013] For each pixel, the spectral image provides information about its intensity and spectrum. It is a 3D data set and can be seen as a "cube" of information. Each point of the cube represents one numerical value and the spectral image is described as I(x,y,λ), where x,y are the pixel coordinates and λ is the spectrum used in each particular band (also called channel). It can be seen as a superposition of images Ii(x,y), each collected using a different wavelength λ, or as the spectrum I(λ) at each pixel (x,y) extracted from the superposition of all images Ii(x,y). This can be extended in the vertical direction (z) from the x,y plane to have a 3D reconstructed image, i.e. a 4D spectral image I(x,y,z,λ).

[0014] There are four basic techniques for acquiring a three-dimensional (x,y,λ) data set of a spectral cube. The choice of technique depends on the particular application, as each technique has its own advantages and disadvantages. Three of these techniques are methods of scan-based imaging systems. In this case, the system operates by acquiring the spectrum of single or simultaneous pixels from an x,y plane on the detector, and when the spectrum of these pixels has been measured, another connected area is measured, and this is done successively until all spectra from all desired fields of view (total imaged area) have been recorded. The mechanism is then moved perpendicular to the first x,y plane (i.e. in the z direction), and the process is repeated successively until all desired z volumes have been measured. This is done by passing light through a prism or galvanometric scanning mirror (fast) that is in the optical path to the detector, or by translating the specimen with a mechanical stage (slow).

[0015] On the other hand, embryo profiling for pregnancy rate prediction generally focuses on short-term biomarkers including visual profile and RNA and protein expression in the surroundings of the embryo, preferably to avoid the occurrence of damage.Embryonic profiling for health prediction focuses on genome and often involves cell sampling from the embryo for preimplantation genetic diagnosis in cases where there is a risk of genetic disease.

[0016] Current clinical procedures for recognizing viable embryos are subjective, and these include time-lapse incubators, preimplantation genetic screening, and / or multiple embryo transfer. Time-lapse incubators have built-in bright-field microscopes, which allow monitoring morphological changes in the embryos, but do not evaluate implantation potential, and preimplantation genetic testing (PGT) for aneuploidy discards embryos that display genetic abnormalities or specific gene mutations, but neither of these two procedures demonstrates a significant increase in implantation / pregnancy rates, and in addition, transferring multiple embryos may result in multiple pregnancies, which are accompanied by many health complications and high costs.

[0017] In view of the above, there remains a need to provide a device, in particular a microscope, and a method of using the same, that is capable of classifying the health of a biological sample by its spectral signature, in particular an embryo, by exciting and measuring intrinsic autofluorescence arising from molecules present in said biological sample in a non-invasive manner. Summary of the Invention [Problem to be solved by the invention]

[0018] The inventors have developed a device and a method for using it that allows to measure in a non-invasive manner the intrinsic autofluorescence originating from molecules, such as metabolites, present in a biological sample. The device achieves minimization of potential photodamage by using two-photon excitation of the autofluorescence in a "light line", light sheet or virtual light sheet configuration with orthogonal objective optics. This, among other features of the device, allows to classify in real time the quality of said biological sample, in particular the quality of live embryos.

[0019] The device of the present invention has a mounting configuration that is very similar to the classical way of mounting samples that is routine in clinics and biological laboratories, and therefore it is easier for the device to be used by end users, while it avoids the mounting of samples into capillaries or cuvettes or cylinder settings that are required in typical light sheet microscopes. [Means for solving the problem]

[0020] It is therefore an object of the present invention to provide a device for measuring the intrinsic autofluorescence of molecules present in a biological sample, said device comprising: a) a light sheet, virtual light sheet, or light line transmission path module including a laser generator capable of generating ultrashort pulses with durations between 1 microsecond and 1 attosecond, and optical elements for shaping the laser light into a light sheet, virtual light sheet, or light line; b) an imaging chamber including an imaging dish in which the biological sample is placed, a translation stage, an illumination objective and a collection objective both immersed in an immersion fluid in a sealed immersion chamber, both objectives positioned below the imaging dish, and the immersion chamber is placed within an incubation chamber, allowing for environmental control; c) a fluorescence detection path comprising an optical element, a light splitting element such as a diffractive element or a dichroic mirror, an array detector, and a device control and signal processing unit, said fluorescence detection path having means for collecting spectra of the fluorescence of said biological sample constructing a multi- or hyperspectral datacube.

[0021] To achieve two-photon excitation of the sample, preferably an ultrashort pulse laser generator is used. Preferably, the laser generator generates ultrashort pulses with durations between 1 microsecond and 1 attosecond, more preferably between 5 femtosecond and 900 femtosecond, more preferably between 50 femtosecond and 600 femtosecond, even more preferably between 100 femtosecond and 300 femtosecond. Preferably, the pulse repetition frequency is between 1 Hz and 1 GHz, and the average power is between 0.1 and 100 watts. Preferably, the laser generator generates a laser with a wavelength between 300 nm and 1,200 nm, more preferably between 500 nm and 1,100 nm, even more preferably between 740 nm and 820 nm. Any laser generator with similar parameters that achieves the same TPA effect could be used in the present invention.

[0022] The laser beam passes through a series of optical elements on the transmission path that allow to control its shape and directionality. Preferably, said optical elements on the transmission path are lenses, light shaping elements, slits, scanning mirrors or combinations thereof.

[0023] In a first embodiment, the light shaping elements in the transmission path can generate a propagation-invariant laser beam propagating along the X direction. Effectively, this illuminates a single line, called the light line, from the sample (along the X direction). Using a scanning mirror and a sample translation stage, this light line is moved with respect to the sample to scan the entire volume of the sample. Detection is performed by a focusing objective positioned in a direction perpendicular to the light propagation (Z direction), and the signal is collected one X line at a time. This is called light line multi photon hyperspectral (HS) imaging.

[0024] In a second embodiment, the light shaping elements in the transmission path can generate a propagation-invariant laser beam propagating along the X direction, similar to the first embodiment. Effectively, this illuminates a single line (along the X direction) from the sample. However, in contrast to the above embodiment, in this case the light line is rapidly scanned in the vertical direction (Y direction) by a resonant scanner, effectively forming a virtual light sheet in the XY plane. The "fast" phrase is relative to the imaging speed of the camera. This is called virtual light sheet multiphoton HS imaging.

[0025] In a further embodiment, the light shaping elements in the transmission path can directly generate a light sheet in the XY plane, for example by using a cylindrical lens, or a spatial light modulator, which is called light sheet multiphoton HS imaging.

[0026] In the device of the invention, the generated XY light sheet scans the sample in the vertical direction (Z direction) or the light line scans the sample in the vertical direction (Y and Z direction), thus achieving scanning over the entire volume of the sample. In one embodiment, the relative scanning of the sample is performed keeping the light sheet or light line stationary while the sample is mounted on a translation stage. In another embodiment, the sample is kept stationary and the light sheet or light line is scanned using a scanning mirror. Both embodiments can be used in combination. The device of the invention includes both options.

[0027] In addition, by using scanning mirrors and / or translation stages, the scanning of the light sheet or line over the sample volume can be done in two different ways: in one embodiment, this is done plane by plane; in another embodiment, this is done plane by plane, but at a rate much faster than the camera capture rate, so that the entire sample is quasi-simultaneously illuminated.

[0028] Preferably, the device also includes a wide-field illumination source, e.g. a white diode, attached to the flexible hand, which is placed close to, e.g. above, the sample and provides a movable ambient illumination in the imaging chamber according to the user's requirements. On the one hand, this facilitates the navigation of the sample. On the other hand, it makes it possible to obtain pseudo-bright-field or pseudo-dark-field images of the sample, which can be used as references, similar to those typically used in laboratories or clinics to study and classify samples.

[0029] Preferably, the bottom of the imaging dish of the imaging chamber is made from a material having a refractive index as close as possible to that of water (n=1.3325). More preferably, said material is selected from polytetrafluoroethylene (PTFE) or fluorinated ethylene propylene (FEP).

[0030] Preferably, the illumination objective and the collection objective are positioned below the imaging dish at approximately 45 degrees with respect to the bottom of the imaging dish and approximately 90 degrees with respect to each other.

[0031] Preferably, the incubation chamber is equipped with a control means such that several environmental parameters are controlled. Thus, several environmental parameters are controlled in the incubation chamber, such as temperature, humidity, pH, CO2 and O2 concentrations. The control means controls the temperature in the incubation chamber so that it is the optimal temperature for the relevant sample under observation.

[0032] Preferably, said optical elements in the detection path are laser light filters, lenses, light shaping elements, slits, mirrors and / or light splitting elements such as de-scanning mirrors, diffractive elements or dichroic mirrors, or combinations thereof.

[0033] In the first embodiment described above, in the detection path, the fluorescence generated by the light line illumination in the X direction is directed to a descanning mirror. The descanning mirror ensures that the fluorescence light line is always projected at the same position while the sample volume is scanned by the laser light line. The slit is effectively the entrance slit of a diffractive element such as a prism, a diffraction grating, or a similar element. The latter is used to disperse and spatially separate the different wavelengths contained in the fluorescence spectrum. The fluorescence spectrum of the light line is thus dispersed along a direction perpendicular to the light line. The fluorescence is then focused and shaped to fit the detection size of an array detector such as a CCD camera, a CMOS camera, or a similar detector. In effect, this forms an image on the array detector, one dimension corresponding to the spatial information contained in the received light line from the sample (X direction) and the second dimension encoding the spectrum of the fluorescence at the corresponding pixel from the light line. As the sample volume is scanned with the light line, a multi- or hyperspectral data cube is constructed, which for each point of the sample contains the sample coordinates X, Y, Z and the fluorescence spectrum. The number of color channels into which the spectrum is binned is determined by the diffractive optics used and the number of pixels into which the spectrum is projected on the detector array. With this configuration, spectra can be acquired with a resolution of a few nanometers to tens of nanometers. The system can be later calibrated with a calibration standard to determine the correspondence between pixel number and fluorescent color. To acquire a pseudo-brightfield or pseudo-darkfield image of the sample, the laser source is turned off, a widefield light source is appropriately positioned near the sample, and the light collected by the collection objective is directed, bypassing the descanning mirror, slit, and diffractive element, and focused onto the camera.

[0034] In the second embodiment described above, the fluorescence originating from the XY light sheet or virtual light sheet receiving plane is directed to a dichroic mirror. The latter splits the light in two, selected to transmit light above a certain wavelength λ1 and reflect the remaining light. The process of splitting the fluorescence in two is repeated with two further dichroic mirrors, with threshold wavelengths λ2 and λ3, respectively. The choice of threshold wavelengths depends on the spectrum that needs to be detected. After this process, the original multicolor fluorescence image of the light sheet receiving XY plane of the sample is split into four images corresponding to the four spatially separated color channels. The detection area of ​​the array detector is then split into four equal parts, and the four fluorescence image channels are imaged simultaneously through a set of redirecting mirrors and focusing lenses. In effect, the array detector records an image, which includes four images of the sample, each image corresponding to the fluorescence spectrum of the sample in a specific color band. By scanning the volume of the sample with the XY light sheet, a multispectral data cube is obtained. The latter includes the X, Y, Z coordinates of each pixel and its spectrum. The spectrum contains only four data points, one from each of the respective color bands. Although this may give somewhat limited fluorescence spectral information, it is sufficient to distinguish healthy from non-healthy biological samples via a subsequently employed data analysis approach via dimensionality reduction such as phasor analysis. In addition, artificial intelligence may also be used to classify the acquired spectral profile. To acquire a pseudo-brightfield or pseudo-darkfield image of the sample, the laser source is turned off, a wide-field light source is appropriately positioned near the sample, and the light collected by the collecting objective is directed bypassing the dichroic mirror and focused onto the camera.

[0035] In the above embodiment, the use of three dichroic mirrors is described, resulting in a four-color channel image. However, it is also possible to include an additional set of dichroic mirrors (with different threshold wavelengths) in the same position and automatically exchange the two sets of dichroic mirrors. This would allow to obtain two successive images of four different color channels each. Thus, in effect, an eight-channel image can be obtained. This can be further repeated if necessary to obtain more channels.

[0036] Preferably, said array detector in the detection path is a charge-coupled device (CCD) camera or an electron multiplying CCD (EMCCD) camera or a complementary metal oxide semiconductor (CMOS) camera or a scientific CMOS (SCMOS) camera.

[0037] In this way, we provide a device comprising a multiphoton laser based emission light sheet to achieve two-photon excitation of the sample, placed in an imaging dish inside an incubation chamber allowing full environmental control, and a fluorescence detection path to collect the fluorescence spectrum emitted by the sample in the manner described above, to enable classification of biological samples with intrinsic autofluorescent molecules using a snapshot technique that reduces photodamage of the sample, providing fast processing speed, reduced data storage capacity and even immediately interpretable raw images.

[0038] In a second aspect, the present invention relates to a method for assessing the quality of a biological sample by measuring the intrinsic autofluorescence of molecules present in the biological sample, using the device described above.

[0039] The method comprises: a) obtaining a raw multispectral or hyperspectral image from the fluorescence signals emitted by a biological sample using the device of the present invention; b) image processing the spectral image acquired in step (a); and c) classifying the processed images obtained in step (b).

[0040] Optionally, in step (a) a standard simulated brightfield or simulated darkfield image is acquired.

[0041] The image processing in step (b) of the method of the present invention includes basic image processing such as background removal, noise removal, signal normalization, image segmentation, image processing by dimensionality reduction such as constructing a phasor or histogram representation, signal unmixing, where each HS image is divided into several images, each corresponding to the location and intensity of the signal originating from a single autofluorescent molecule, as well as other steps known to the skilled person, among others.

[0042] Optionally, the raw HS image and the simulated brightfield or pseudo darkfield image acquired in step (a) are also classified.

[0043] To construct the phasor representation in step (b), the spectrum of each raw pixel is normalized and then Fourier transformed, and the real (amplitude G) and imaginary (phase S) parts of the nth (typically 1st / 2nd, n=1,2) harmonics are used as coordinates in a scatter (phase) diagram.

[0044] Each pixel in the phasor plot corresponds to a different spectrum from the original multi- or hyperspectral image. Thus, similar spectra occurring at different locations in the raw multi- or hyperspectral image are placed in the same pixel (or adjacent pixels) on the phasor plot. In effect, the original raw multi-dimensional spectral image is binned into a histogram and thus its dimensionality is reduced to a 2D phasor plot. The phasor plot contains characteristic features, determined by the composition and amount of autofluorescent compounds in the specimen. These features, for example the center of mass of the phasor for a particular specimen, can be used as an indicator for the quality of the specimen.

[0045] To build a histogram of the multi- or hyperspectral information of the sample, all spectra from all pixels are summed to obtain one spectrum with multiple features characteristic of the autofluorescence of a particular sample. These features can be further used as indicators for the quality of the specimen.

[0046] The spectrum acquired in step (b) is a composite spectrum with contributions from potentially multiple autofluorescent molecules, and it is therefore often beneficial to unmix the raw spectral data against the contributions of these individual autofluorescent molecules. Spectral unmixing can be performed in several ways, but can be done by identifying reference spectral positions of pure autofluorescent molecules present in the sample. These can be acquired in a separate experiment by imaging pure solutions of said molecules, or can be acquired directly from regions in the original image that contain only signal from said single molecules, if such regions exist.

[0047] Effectively, during the unmixing procedure, for each pixel of the original image, the relative contribution of the spectra of the different pure autofluorescent molecules that contributed to the formation of that composite spectrum is calculated. The unmixed spectral data can be mapped back to their original location on the HS image, thus obtaining characteristic maps of the embryo, one for each different autofluorescent molecule, as shown, for example, in Figures 2 and 4. These maps can provide complementary quantitative information on the concentration and spatial distribution of the different autofluorescent molecules. Thus, they can be used as complementary information to assess the quality of the biological specimen.

[0048] Once raw HS images, or pseudo brightfield / pseudo darkfield images, have been acquired and histogram phasor plots and unmixed images constructed for each imaged specimen, an independent assay is used to assess the quality of the specimens and correlate them with the raw HS images, or pseudo brightfield / pseudo darkfield images, histograms, phasor plots, and unmixed images. After such libraries of healthy and unhealthy specimens and correlated images have been constructed, the next step is to assess the quality of previously unknown biological specimens. This step may also be enhanced by applying machine learning algorithms.

[0049] The methods of the present invention allow for classification of biological samples, including but not limited to cells, embryos, body fluids, or any other biological material, based on their autofluorescence profile, which distinguishes between viable and non-viable embryos for implantation, healthy and non-healthy oocytes, tumors and normal cells, infected and healthy tissue, or any other quality that affects the molecular composition of the biological sample.

[0050] Preferably, the biological sample may be selected from the list comprising cells, biopsy tissues, organoids, oocytes, embryos, biological fluids such as blood, plasma, saliva, semen, urine, or any other biological sample, provided that autofluorescent molecules are present in said biological sample. More preferably, said biological sample is an embryo, more preferably a human embryo.

[0051] Preferably, the autofluorescent molecules present in the biological sample to be measured may be selected from a list including free and protein-bound reduced nicotinamide adenine dinucleotide (NADH), flavin adenine dinucleotide (FAD), retinol, retinoic acid, folic acid, lipid pigments such as lipofuscin and protoporphyrin IX (PPIX), proteins such as elastin, or collagen.

[0052] The present invention will now be described with reference to figures which are not intended to limit the invention. [Brief description of the drawings]

[0053] [Figure 1] FIG. 1 is a schematic diagram showing a first embodiment of a device of the present invention. [Diagram 2] FIG. 2 is a graph of data obtained from an embryo using the device of FIG. 1. [Diagram 3] FIG. 2 is a schematic diagram showing a second embodiment of the device of the present invention. [Figure 4] FIG. 4 is a graph of data obtained from embryos using the device of FIG. 3. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0054] FIG. 1 depicts a schematic diagram of a first embodiment of the device of the invention. Three different zones or modules are illustrated (A, B, and C). Zone A is a light sheet transmission path that includes an ultrashort pulse laser generator 1. The laser beam generated by this generator is in the near infrared (NIR) range. The laser beam 2 passes through a series of optical elements that allow to control its shape and directionality. The purpose of the lenses 3, 3' is to form a telescope system that allows to expand or reduce the diameter of the laser beam. The lens is followed by a light shaping element 4, which can be a cylindrical lens, a mask, a spatial light modulator, a slit 5, or a combination of these, whose function is to achieve the shaping of the laser beam into the desired light line. The light shaping element generates a propagation-invariant laser beam that propagates along the X direction. In effect, this illuminates a single line (along the X direction) from the sample. Using a scanning mirror 6 and a sample translation stage, this light line is moved with respect to the sample to scan the entire volume of the sample.

[0055] Zone or module B is the imaging chamber. The biological sample 7, which may be an embryo or cell, a biopsy tissue, an organoid, blood, plasma, or any other biological tissue, tissue or fluid, is positioned in the imaging dish 8 and immersed in the biological medium 8'. The bottom of the imaging dish 8 is made of a material with a refractive index as close as possible to that of water (n=1.3325), for example polytetrafluoroethylene (PTFE), fluorinated ethylene propylene (FEP), or other similar substances. Matching the refractive indices of the biological medium 8', the objective immersion liquid 11, and the imaging dish bottom ensures minimal optical distortion and imaging artifacts occurring at the respective interfaces. This is particularly important in the selected imaging geometry where the objective lenses 9, 9' are positioned below the imaging dish, at about 45 degrees from the bottom of the imaging dish. The imaging dish 8 is mounted on a translation stage 10, allowing both coarse positioning of the sample in the field of view and scanning of the sample across a static light line. The light line is transmitted through an illumination objective 9. The ultrashort NIR pulses generate extensive intrinsic autofluorescence from the sample. This fluorescence signal is mainly in the visible part of the spectrum, many of which originate from various molecules (e.g., metabolites) that are indicative of the health or metabolic state of the biological sample. The fluorescence is collected by a collection objective 9' and sent further to the "fluorescence detection path" (zone or module C). Both objectives 9, 9' are inserted into a sealed immersion chamber 11' and immersed in an immersion liquid 11, usually water, with a refractive index matching that of the biological medium 8'. The objectives 9, 9' are positioned in an inverted configuration below the sample 7 and imaging dish 8. The sample is illuminated by the light line at an angle of about 45 degrees relative to the imaging dish 8 and about 90 degrees with respect to the other objective, although several configurations are possible depending on the particular set of objectives used. The entire immersion chamber 11' is placed within an incubation chamber 12, which allows for environmental control, maintaining a stable physiological temperature (typically 37°C), and is equipped with control means to regulate humidity, CO2 and O2 concentrations as well as pH.The imaging chamber also houses a wide-field light source 30 mounted on a flexible arm 31 for acquiring simulated bright-field or simulated dark-field images.

[0056] Zone or module C is the detection path. The autofluorescence light collected by the collecting objective 9' is first passed through a near infrared (NIR) filter 13, which is used to cut off the residual light from the excitation laser. It then passes through another telescope system (lenses 14, 14') to adjust the size of the beam. From here, the fluorescence light (generated by the light line illumination in the X direction) is directed to a descanning mirror 15, which ensures that the fluorescence light line is always projected at the same position on a slit 16 while the sample volume is scanned by the laser light line. This slit 16 is effectively the entrance slit of a diffractive element 17. The diffractive element 17 is used to disperse and spatially separate the different wavelengths contained in the fluorescence spectrum. The fluorescence spectrum of the light line is thus dispersed along a direction perpendicular to the light line. This fluorescence is then focused through lens 18 and shaped through light shaping element 19 and slit 20 to match the detector size of array detector 21, which may be a charge-coupled device (CCD) camera, a complementary metal oxide semiconductor (CMOS) camera, or any similar detector. Effectively, this forms an image 22 on the array detector, one dimension corresponding to the spatial information contained in the received line of light from the sample (X direction) and the second dimension encoding the spectrum of fluorescence at the corresponding pixel from the line of light. As the line of light scans the sample volume, a hyperspectral data cube is constructed, which contains, for each point of the sample, the sample coordinates X, Y, Z, and the fluorescence spectrum. The number of color channels into which the spectrum is binned is determined by the diffractive optics used and the number of pixels into which the spectrum is projected on the detector array. In this configuration, spectra can be acquired with a resolution of a few nanometers to tens of nanometers. The system is calibrated with calibration standards to determine the correspondence between pixel number and fluorescent color. The data is analyzed and classified using software in device control and signal processing unit 23.

[0057] FIG. 2 is a graph of data obtained from an embryo using the device of FIG. 1. As raw data, the device obtains a raw HS image that is multidimensional, since each pixel of the image contains a pseudo-continuous autofluorescence spectrum. Alternatively, it can be seen as a data cube of a large number of images, each image corresponding to a particular color channel. The device can also obtain pseudo-brightfield or pseudo-darkfield images of the sample. The image analysis includes various steps, such as constructing a spectral histogram, representing a phasor plot, and unmixing the HS image for the contributions of the different autofluorescent dyes. The reference spectra for the pure autofluorescent dyes used for unmixing can be identified as points on the phasor plot. All this data is statistically analyzed and subjected to classification to obtain a characteristic autofluorescence profile that is linked to the quality of the sample, in this particular case the embryo. All profiles from all embryos are compared to a reference control plot from a viable embryo, making it possible to classify the quality of said embryo. Further on, this allows the classification of viable and non-viable embryos.

[0058] FIG. 3 depicts a schematic diagram of another embodiment of the device of the invention. Three different zones or modules are illustrated (A, B, and C). Zone A is a light sheet / virtual light sheet transmission path that includes an ultrashort pulse laser generator 1. The laser beam generated by this generator is in the near infrared (NIR) range. The laser beam 2 passes through a series of optical elements that allow to control its shape and directionality. The purpose of the lenses 3, 3' is to form a telescope system that can expand or reduce the diameter of the laser beam. The lens is followed by a light shaping element 4, which can be a cylindrical lens, a mask, a spatial light modulator, a slit 5, or a combination of these, whose function is to achieve the shaping of the laser beam into the desired shape, i.e., a light line to constitute (i) a light sheet or (ii) a virtual light sheet. In the former case, the light shaping element generates the light sheet directly in the XY plane, for example via a cylindrical lens. In the latter case, the light shaping element generates a propagation-invariant laser beam propagating along the X direction, effectively illuminating a single line (along the X direction) from the sample. This light line is then rapidly scanned in the Y direction to form a virtual light sheet in the XY plane, where rapidly means rapidly compared to the imaging speed of the camera. In either case, using a scanning mirror 6 and a sample translation stage 10, this light sheet / virtual light sheet is moved with respect to the sample to scan the entire volume of the sample.

[0059] Zone or module B is the imaging chamber. A biological tissue sample 7, which may be an embryo or cell, a biopsy tissue, an organoid, blood, plasma, or any other biological tissue, tissue or fluid, is positioned in an imaging dish 8 and immersed in a biological medium 8'. The bottom of the imaging dish 8 is made of a material with a refractive index as close as possible to that of water (n=1.3325), such as polytetrafluoroethylene (PTFE), fluorinated ethylene propylene (FEP), or other similar substances. Matching the refractive indices of the biological medium 8', the objective immersion liquid 11, and the imaging dish bottom ensures minimal optical distortion and imaging artifacts occurring at the respective interfaces. This is particularly important in the selected imaging geometry where the objective lenses 9, 9' are positioned below the imaging dish, at about 45 degrees from the bottom of the imaging dish. The imaging dish 8 is mounted on a translation stage 10, allowing both coarse positioning of the sample in the field of view and scanning of the sample across a static light sheet. The light sheet or virtual light sheet is transmitted through the illumination objective 9. The ultrashort NIR pulses generate extensive intrinsic autofluorescence from the sample. This fluorescence signal is mainly in the visible part of the spectrum, many of which originate from various molecules (e.g., metabolites) that are indicative of the health or metabolic state of the biological sample. The fluorescence is collected by the collection objective 9' and sent further to the "fluorescence detection path" (zone or module C). Both objectives 9, 9' are inserted into a sealed immersion chamber 11' and immersed in an immersion liquid 11, usually water, with a refractive index matching that of the biological medium 8'. The objectives 9, 9' are positioned in an inverted configuration below the sample 7 and imaging dish 8. The sample is illuminated by the light sheet at an angle of about 45 degrees relative to the imaging dish 8 and about 90 degrees with respect to the other objective, although several configurations are possible depending on the particular set of objectives used. The entire immersion chamber 11' is placed within an incubation chamber 12, which allows for environmental control, maintaining a stable physiological temperature (typically 37°C), and is equipped with control means to regulate humidity, CO2 and O2 concentrations as well as pH.The imaging chamber also houses a wide-field light source 30 mounted on a flexible arm 31 for acquiring simulated bright-field or simulated dark-field images.

[0060] Zone or module C is the detection path. The autofluorescence collected by the collecting objective 9' is first passed through a NIR filter 13, which is used to cut off the residual light from the excitation laser. It then passes through another telescope system (lenses 14, 14') to adjust the size of the beam. From here, the fluorescence is directed to a mirror 23. The fluorescence (from the XY light sheet receiving plane) is directed to a dichroic mirror 24, which splits the light in two, selected to transmit light above a certain wavelength λ1 and reflect the remaining light. The process of splitting the fluorescence in two is further repeated with dichroic mirrors 25 and 26, with respective threshold wavelengths λ2 and λ3. The choice of threshold wavelength depends on the spectrum that needs to be detected. After this process, the original multicolor fluorescence image of the light sheet receiving XY plane of the sample is split into four images corresponding to the four spatially separated color channels. The detection area of ​​the array detector is then effectively split into four parts and the four fluorescence image channels are imaged simultaneously through a set of redirecting mirrors and focusing lenses 27, 27', 27'', 27'''. Effectively, the array detector 28 records an image 29, which includes four images of the sample, each image corresponding to the fluorescence spectrum of the sample in a particular color band. By scanning the volume of the sample with an XY light sheet, a multispectral data cube is obtained. The latter includes the X, Y, Z coordinates of each pixel and its spectrum. The spectrum includes only four data points, acquired from each of the respective color bands. This information is sufficient to distinguish between healthy and non-healthy biological samples through a subsequently employed data analysis approach via dimensionality reduction, such as phasor analysis. Artificial intelligence may also be used to classify the acquired spectral profiles.

[0061] FIG. 4 is a graph of data obtained from an embryo sample using the device of FIG. 3. As raw data, the device obtains a raw HS image, which is multidimensional, since in this case each pixel of the image contains a four-point autofluorescence spectrum. Alternatively, it can be seen as a data cube of four images, each image corresponding to a particular color channel. The device can also obtain pseudo-brightfield or pseudo-darkfield images of the sample. The image analysis includes various steps, such as constructing a spectral histogram, representing a phasor plot, and unmixing the HS image for the contributions of the different autofluorescent dyes. The reference spectra for the pure autofluorescent dyes used for unmixing can be identified as points on the phasor plot. All this data is statistically analyzed and subjected to classification to obtain a characteristic autofluorescence profile that is linked to the quality of the sample, in this particular case the embryo. All profiles from all embryos are compared to a reference control plot from a viable embryo, making it possible to classify the quality of said embryo. Further on, this allows the classification of viable and non-viable embryos.

Claims

1. 1. A device for measuring the intrinsic autofluorescence of molecules present in a biological sample, comprising: (a) a transmission path module including a laser generator that generates ultrashort pulses of laser light having a duration between 1 microsecond and 1 attosecond, and an optical element that shapes the laser light into a light sheet, a virtual light sheet, or a light line; (b) an imaging chamber including an imaging dish in which the biological sample is placed, a translation stage, an illumination objective and a collection objective both immersed in an immersion fluid within a sealed immersion chamber, the illumination objective and the collection objective positioned below the imaging dish, and the immersion chamber located within an incubation chamber to provide environmental control; (c) a device comprising a fluorescence detection path including optical elements, light-splitting elements such as diffractive elements or dichroic mirrors, an array detector, and a device control and signal processing unit, the fluorescence detection path having means for collecting spectra of fluorescence of said biological sample constructing a multi- or hyperspectral datacube.

2. 10. The device of claim 1, wherein the laser generator generates laser light having a wavelength in the range of 300 nm to 1,200 nm and / or having a pulse repetition frequency in the range of 1 Hz to 1 GHz with an average power between 0.1 Watts and 100 Watts.

3. The device of claim 1 or 2, wherein the optical element of the transmission path module is a lens, a light shaping element, a slit, a scanning mirror, or a combination thereof.

4. 3. The device according to claim 1 or 2, wherein the bottom of the imaging dish of the imaging chamber is made of a material having a refractive index as close as possible to the refractive index of water (n=1.3325), preferably selected from polytetrafluoroethylene (PTFE) or fluorinated ethylene propylene (FEP).

5. The device of claim 1 or 2, wherein the illumination objective and the collection objective are positioned below the imaging dish at about 45 degrees to the bottom of the imaging dish and at about 90 degrees to each other.

6. The incubation chamber controls temperature, humidity, pH, CO ２ and O ２ 3. A device according to claim 1 or 2, comprising control means capable of measuring and controlling some environmental parameters, such as concentration.

7. the fluorescence detection path comprises an optical element including a lens, a mirror, a light splitting element such as a diffraction element or a dichroic mirror, a slit, a descanning mirror, or a combination thereof; the fluorescence detection pathway comprises an array detector, which is a charge-coupled device (CCD) camera or an electron-multiplying CCD (EMCCD) camera or a complementary metal-oxide semiconductor (CMOS) camera or a scientific CMOS (SCMOS) camera; the fluorescence detection path includes a descanning mirror onto which fluorescence generated by the light line illumination in the X direction is directed; 3. The device of claim 1, wherein the fluorescence detection path comprises a configuration in which a first dimension of an image formed in an array detector of the fluorescence detection path is spatial information contained in a line of received light (X direction) from the biological sample, and a second dimension is a spectrum of fluorescence at a corresponding pixel from the line of light for each point on the biological sample.

8. the fluorescence detection path includes at least three dichroic mirrors that split the multicolored fluorescent light beam into at least four spatially separated color channels, and the three dichroic mirrors split the multicolored fluorescent light beam into the four spatially separated color channels; the fluorescence detection path comprises a detection area of ​​the array detector divided into four equal parts, and four fluorescence image channels are imaged simultaneously through a set of redirecting mirrors and focusing lenses; the fluorescence detection path comprises an image formed in an array detector of the fluorescence detection path comprising four images of the biological sample, each image corresponding to a fluorescence spectrum of the biological sample in a particular color band; 3. The device of claim 1, wherein the fluorescence detection path comprises at least one additional set of three dichroic mirrors in the same position and automatically swapping between two sets of dichroic mirrors.

9. 3. A method for assessing the quality of a biological sample by measuring the intrinsic autofluorescence of molecules present in said sample, using a device according to claim 1 or 2, comprising: (a) acquiring raw multispectral or hyperspectral images from fluorescence signals emitted by the biological sample using a device according to claim 1 or 2; (b) processing the image acquired in step (a); (c) classifying the image processed in step (b).

10. 10. The method of claim 9, wherein the biological sample is selected from the list including cells, biopsied tissues, organoids, oocytes, embryos, biological fluids such as blood, plasma, saliva, semen, urine, or any other biological sample, provided that autofluorescent molecules are present in the biological sample.

11. 10. The method of claim 9, wherein the biological sample is an embryo, preferably a human embryo, and the molecules present in the biological sample to be measured are selected from the group comprising free and protein-bound reduced nicotinamide adenine dinucleotide (NADH), flavin adenine dinucleotide (FAD), retinol, retinoic acid, folic acid, lipid pigments such as lipofuscin and protoporphyrin IX (PPIX), proteins such as elastin or collagen.

12. 10. The method of claim 9, wherein when a line of light is used in the transmission path module, a first dimension of an image formed at the array detector of the fluorescence detection path is spatial information contained in a line of received light (X direction) from the biological sample, and a second dimension encodes a spectrum of fluorescence at a corresponding pixel from the line of light for each point on the biological sample.

13. 13. The method of claim 12, wherein a hyperspectral datacube is constructed as the line of light volumetrically scans the biological sample, the hyperspectral datacube comprising, for each point of the biological sample, coordinates X, Y, Z and a fluorescence spectrum of the biological sample.

14. the fluorescence detection path includes at least three dichroic mirrors that split the polychromatic fluorescence collected beam into at least four spatially separated color channels, and the three dichroic mirrors split the polychromatic fluorescence collected beam into at least four spatially separated color channels; the fluorescence detection path comprises an array detector divided into four equal regions, and four fluorescence color images are simultaneously captured through a set of redirecting mirrors and focusing lenses; 10. The method of claim 9, wherein the fluorescence detection path comprises an image formed including four images of the biological sample in an array detector of the fluorescence detection path, each image corresponding to the fluorescence spectrum of the biological sample in a particular color band.

15. volumetrically scanning the biological sample with an XY light sheet results in a multispectral data cube containing the X, Y, Z coordinates of each pixel and its fluorescence spectrum; image processing of the image includes obtaining a spectral histogram, a phasor plot, and / or spectrally unmixed data; 10. The method of claim 9, wherein the device uses machine learning algorithms to assess the quality of the biological sample and classify the biological sample.

Images