System and method for differentiating and separating sperm cells

Abstract

A method and system for classifying mammalian sperm cells is provided. The classification may, for example, be within the framework of semen sexing. A suspension with motile sperm cells suspended in a medium is passed in front of a video capturing utility under conditions that permit said utility to capture images of single sperm cells. The medium is devoid of any cell-staining or biomarker-identifying dyes and the cells in the medium were not pre-stained with such dyes. One or more video frames of individual sperm cells in the medium are captured and fed into a computing utility that is configured for analyzing the video stream using one or more pre-trained Al models to detect individual sperm cells and one or more sperm cell features of the individual sperm cells and to classify the detected sperm cells as belonging to a defined category based on the sperm cell features identified in said analyzing.

Description

[0001] SYSTEM AND METHOD FOR DIFFERENTIATING AND SEPARATING SPERM CELLS

[0002] TECHNOLOGICAL FIELD

[0003] The present invention is in the field of sperm cell analysis and separation. Particularly, for obtaining enriched populations of sperm cells with desired characteristics.

[0004] BACKGROUND ART

[0005] References considered to be relevant as background to the presently disclosed subject mater are listed below:

[0006] - US 5,514,537

[0007] - EP 2761275

[0008] - US 3,687,806

[0009] - KR 101916959

[0010] - US 2003 / 0162238

[0011] - US 2014 / 0182005

[0012] - US 4,191,749

[0013] - US 5,021,244

[0014] - US 6,153,373

[0015] - US 6,489,092

[0016] - US 4,083,957

[0017] - US 2003 / 0087860

[0018] Acknowledgement of the above references herein is not to be inferred as meaning that these are in any way relevant to the patentability of the presently disclosed subject mater. BACKGROUND

[0019] Differentiating between types of sperm cells in a semen sample according to desired characteristics is needed in various practices such as livestock insemination, in vitro fertilization, genetic consultation, and the like. The desired characteristics include for example, fertilization potential, motility, morphology, and genetic content of the sperm cells. One of the sperm cells’ traits that is of major importance is the sex chromosome content of the sperm cells, namely, the determination whether a sperm cell carries an X chromosome, or a Y chromosome, or any aberration thereof.

[0020] Determination of the sex chromosome content of sperm cells is used for selecting the sex of the offspring, for example in the livestock industry, to maximize husbandry efficiency. Sex selection in livestock animals is mainly performed prior to fertilization, by increasing the content of X chromosome-bearing or Y chromosome-bearing sperm cells in semen samples intended for artificial insemination. The process of differentiating sperm cells according to their sex chromosome content is known as semen sexing.

[0021] Currently several technologies are used for semen sexing, for example methods that involve the separation of sperm cells according to their density and / or size (US 5,514,537., labeling sperm cells by specific marker molecules (EP 2761275 Bl, US 3,687,806 A, KR 101916959 Bl, US 2003 / 0162238 A, US 2014 / 0182005 Al, US 4,191,749, US 5,021,244, US 6,153,373, US 6,489,092), separation of sperm cells according to their electrical properties (US 4,083,957 A), and gene editing (US 2003 / 0087860 Al). The most common method for semen sexing is currently the measurement of fluorescence signal after DNA-staining. In this method, X chromosomebearing sperm cells are identified by their higher fluorescent signal compared to Y chromosome-bearing sperm cells, when stained with Hoechst DNA-binding stain.

[0022] GENERAE DESCRIPTION

[0023] It was realized, in accordance with this disclosure, that current methods of semen sexing have a few shortcomings that include a damage to sperm cells and reduction in their fertilization success rate mainly due to DNA staining and excitation by laser beams and the need for trained operators in the performance of such methods. The current disclosure provides a method and system that make use of sperm visualization and a pretrained Al-based model for the purpose of detecting individual sperm cells and one or more sperm cell features of the individual sperm cells. Based on these features sperm cells may be classified, for example, among others, into such that contain an X or a Y chromosome, into living and dead cells, or into cells with high and low fertility potential. The disclosure comprises classifying genetic or biological traits that may be unknown or may not be detectable by standard visual inspection techniques.

[0024] The term "pre-trained as used in relation to the Al -based model or algorithm is intended to denote that the algorithm has been subjected to some pre -training. However, it should be understood, that such an algorithm, while pre-trained, may also undergo continued training when used in practice; for example, with a training set specific for a specific ejaculate -providing mail mammal (e.g. a bull), a set specific for a defined herd, etc. In other words the algorithm may undergo continued refinement when in use. Also, the refinement may be by external data.

[0025] Thus, provided by this disclosure is a method and system for classifying mammalian sperm cells in a suspension of such cells. In the following the term “sperm cells suspension ” may be used herein to denote a medium that comprises moving sperm cells suspended therein (namely sperm cells that are not immobilized and freely swim in the medium) that were not exposed to or pre-stained with dyes. The sperm cells suspension is typically a semen sample diluted in an appropriate medium. The sperm cells suspension may also be referred to in short as "suspension ".

[0026] One of the purposes of such classification, according to embodiments of this disclosure, is to separate sperm cells that meet certain defined criteria from other cells, particularly for the purpose of semen sexing. Unlike prior art methods and systems, the classification is performed without (i) immobilizing the cells, (ii) use of laser beams to illuminate the sperm cells, or without (iii) exposing the cells to a stain or dye that stains cells or identifies certain biomarkers, prior or during the performance of the classification. In other words, the classification is performed without any alteration to the sperm cells natural properties or behavior.

[0027] This disclosure concerns two independent aspects: a method for such classification, and a system for such classification. These aspects may be referred to herein as the “method aspect” and the “system aspect”, respectively.

[0028] Provided by the method aspect is a method for classification of mammalian sperm cells that comprises passing the sperm cells suspension in front of a video capturing utility, capturing one or more video frames of individual sperm cells in the suspension, and feeding the captured video frames into a computing utility for analysis and classification. Said passing is carried out under conditions that permit the video capturing utility to capture images of single sperm cells. As noted above, the sperm cells suspension is devoid of any cell-staining or biomarker-identifying dyes, the sperm cells or at least a portion thereof are motile, and furthermore, the cells in the suspension were not exposed to or pre-stained with dyes. The computing utility is configured for analyzing the fed video frames using one or more pre-trained Al models to detect individual sperm cells and one or more sperm cell features of the individual sperm cells. Based thereon the detected sperm cells can be classified as belonging to a defined category based on the sperm cell features identified in said analyzing.

[0029] The term “video stream” may be used herein to denote a feed of data representative of one or more video frames that are fed into the computing utility.

[0030] Provided by the system aspect is a system for the classification of mammalian sperm cells that comprises (1) a flow channel configured for transferring the sperm cells suspension therethrough; (2) a video capturing utility configured for capturing images of single sperm cells flowing through the flow channel and generating a video stream; and (3) a computing utility running one or more pre-trained Al models and configured for receiving and analyzing the video stream to (i) detect individual sperm cells and one or more sperm cells features of the individual sperm cells and (ii) classify the detected sperm cells as belonging to a defined category based on said features.

[0031] Certain embodiments of the above two aspects will now be described. An embodiment that is described in a manner implying that it is directed to one aspect applies, mutatis mutandis also to the other aspect. For example, where the embodiment implies as relating to the method, for example with reference to the identification of a certain sperm cell-related feature, it also applies to an embodiment of the system in which the computer utility is configured for such identification. Embodiments described without any such implications should be understood as applying to both aspects.

[0032] The features that are being detected and analyzed as well as the classification may, by some embodiments, comprise one or more of the following:

[0033] (a) detection and analysis of visual mobility features of one or more parts of the sperm cells;

[0034] (b) detection and analysis of structure-related features of at least one part of the sperm cells; (c) sperm sexing, namely classifying the sperm cells as X chromosomecontaining or Y chromosome-containing sperm cell;

[0035] (d) classifying a sperm cell as being a dead cell or a living cell;

[0036] (e) classifying a sperm cell as being a normal cell or an abnormal cell;

[0037] (f) classifying sperm cells according to their fertilization potential; or

[0038] (g) differentiating sperm cells from non-sperm cells.

[0039] Embodiments (c) - (g) may, for example, be carried out using the detection and analysis of embodiments (a) - (b). By an embodiment the computing utility is configured to analyze the data using one or more pre-trained Al models. The pre-trained Al models may, by some embodiments, be an ensemble of two or more pre-trained Al models. For example, the one or more pre-trained Al models comprise a first pre-trained Al model and a second pre-trained Al model, the first pre-trained Al model being configured for analyzing the video stream and providing an input to a second pre-trained Al model. In some embodiments, at least one of the Al models is based on one or more of the following: computer vision-based Al, machine learning (e.g. deep learning), and computer vision-based Al combined with machine learning.

[0040] By some embodiments, the classification of the sperm cells serves the purpose of physically separating sperm cells with defined characteristics from other cells. The separation may comprise one or more of the following:

[0041] (i) separating sperm cells from non-sperm cells;

[0042] (ii) separating X chromosome-containing from Y chromosome-containing sperm cells, e.g., for the purpose of obtaining a sperm cell-containing suspension with sperm cell population enriched with X chromosomecontaining or Y chromosome-containing sperm cells;

[0043] (iii) separating dead cells from living cells;

[0044] (iv) separating abnormal cells from normal cells; and

[0045] (v) separating sperm cells with a fertility potential above a defined threshold from other sperm cells.

[0046] For said separating, the system may comprise a cell separating utility controllable by the computing utility and configured for separating sperm cells with defined characteristics from other cells. The cell separating utility may, for example (and without limitation) be a cell sorter or a microfluidic unit. The separation may be carried out at different temperatures, for example between about 4°C to about 39°C. A specific, but non-limiting embodiment of the separation, is for obtaining a sperm cell-containing suspension with a sperm cell population enriched with X chromosome-containing or Y chromosome-containing sperm cells.

[0047] For the capturing of video frames of individual sperm cells, the system may comprise at least one light source that is configured for illuminating the sperm cells suspension passing in the flow channel. The at least one light source may be selected from a group consisting of an artificial visible light source, a natural visible light source, an ultraviolet (UV) light source, a far infra-red (IR) light source, a near IR light source; it may be polarized or non-polarized light; it may be reflected light, transmitted light, a combination of reflected light and transmitted light; it may be bright field illumination, dark field illumination, phase contrast illumination; etc. As can be appreciated, this disclosure is not limited by the nature of light that is used to visualize the sperm cells (and other matter) in the sperm cells suspension. The at least one light source may also be a broad band source or may be at defined wavelengths.

[0048] The at least one light source may also be configured to illuminate the sperm cells suspension passing in the flow channel by a combination of lights having different characteristics: combination of different discrete wavelengths, combination of different illumination angles, combination of different light focusing techniques, etc. Combination of light of different characteristics may be by illuminating light from different sources at the same time or alternating between lights of different characteristics, for example, a pulse of light of one characteristic, then a pulse of another, and so forth. By proper timing of such pulses or using different optical and / or digital filtering techniques, each captured cell may be captured by light of different characteristics, which may aid in the analysis that is carried out in accordance with this disclosure.

[0049] It is also possible, by some embodiments, to use ambient light without any illumination.

[0050] BRIEF DESCRIPTION OF THE DRAWINGS

[0051] To better understand the subject matter that is disclosed herein and to exemplify how it may be carried out in practice, embodiments will now be described, as non-limiting example only, with reference to the accompanying schematic drawings, in which:

[0052] Fig. 1 is a schematic illustration of a system for differentiating between various types of sperm cells in a sperm cells suspension. Fig. 2 is a schematic illustration of a system similar to that of Fig. 1, coupled to a cell sorter.

[0053] Fig. 3 is a schematic illustration of an artificial neural network for determining a type of individual sperm cells according to visual signals.

[0054] DETAILED DESCRIPTION OF EMBODIMENTS

[0055] In the following description some exemplary embodiments will be described, it is to be understood that the subject matter is not limited in its application, details and arrangement of the elements as set forth in the following description or depicted in the drawings; these being rather examples of the full scope of this disclosure as described above and defined in the appended claims.

[0056] The present disclosure provides a technology for classifying and differentiating between various types of sperm cells in a sperm cells suspension, typically in real-time while the sperm cells are imaged as they pass through a flow channel in front of a video images capturing device. Individual sperm cell images are then analyzed using a pretrained artificial intelligence (Al) model. As already noted above, there are two aspects of this disclosure: one aspect concerns a method for classifying mammalian sperm cell; the other aspect concerns a system for such classification. The method may, typically, comprise a step for separating the sperm cells into two or more distinct populations of cells, each population enriched with cells of specific trait, e.g. population of cells enriched with X or Y chromosome-containing cells within the framework of sperm sexing. The system may, comparably, include a cell separation utility.

[0057] In accordance with the present disclosure, the sperm cells are not manipulated (other than being suspended and diluted in a medium), namely they are not exposed to any dyes prior to or during the classification procedure.

[0058] The sperm cells suspension is, typically, a diluted semen sample. The semen sample is, typically, diluted with a medium to yield a sperm cells suspension such that when passing it in front of the video capturing utility, most of the cells will not overlap and be spaced apart; for example, will pass in front of the video capturing utility one after the other. Such dilution, in some embodiments may be in the range of 1:5, 1: 10, 1:20, 1:50, 1: 100, 1:200, 1:500, 1: 1,000, 1:2,000, 1:5,000, 1: 10,000 or in some embodiment even a higher dilution. The dilution may depend on the sperm cells concentration in the original ejaculate, the system features, such as the diameter of the flow channel, and other considerations. The diluting medium may be any medium suitable for that purpose, including such known or used in the art. Specific, and non-limiting, examples are common semen extenders, custom-made mediums, saline, PBS, or other buffers. The sperm cells suspension comprises mobile sperm cells that were not exposed to a dye that stains such cells and does not contain such a dye.

[0059] The present disclosure encompasses the analysis of any type of sperm cells suspension, from any organism and in any condition. It may, for example be a fresh sperm cells suspension, namely one that is analyzed immediately after collection, a frozen sperm cells suspension that was thawed, and the like.

[0060] The sperm cells suspension is passed through a flow channel under conditions that permit the video capturing utility to capture images of single sperm cells. Typically, although not exclusively, this includes passing the flowing sperm cells suspension through a narrow flow channel under conditions such that the cells would pass in front of the video-capturing utility spaced apart to permit it to capture images of one cell at a time. Such conditions may also include certain concentrations of the sperm cells or, conversely, the extent of dilution of the original semen sample.

[0061] In accordance with the present disclosure the video capturing utility captures one or more video frames of each of the individual sperm cells in the medium. The video frames may be successive video frames or selected, non-successive video frames.

[0062] The video stream is analyzed to detect individual sperm cells and sperm cell visual features. These features may comprise one or more visual motility features and / or one or more visual structure-related features e.g. features that are identifiable in a single image. The visual motility features that may be taken into consideration by the Al algorithm may include, but are not limited to, speed of movement, lateral head displacement, linearity of motion path, straight-line velocity, curvilinear velocity, average path velocity, amplitude of lateral head displacement, beat frequency, spinning rate, motion dynamics, dynamic positioning, direction of movement, bending dynamics, swimming dynamics, response to flow of medium in which the sperm cell swims, or any combination thereof. The visual structure-related features that may be taken into consideration by the Al algorithm include, but are not limited to, color, transparency, phase contrast, size, shape, positioning, position relative to other objects in the at least one image, orientation, bending, circularity, eccentricity, solidity, convexity, curvature, skeleton, center of mass, center of gravity, fill factor, Feret diameter, compactness, inertia tensor, fractal dimensions, density, symmetry, texture, organelle key points, length of tail, bending of tail, relative depth of the sperm cell or of parts of the sperm cell, Z position, spatial relationship between the sperm cell and other objects, spatial relationship between different parts of the sperm cell, aspect ratio of different parts of the sperm cell, or any combination thereof.

[0063] Use is made in accordance with the present disclosure with computer vision (CV), artificial intelligence (Al), machine learning (ML), and deep learning (DL), for identifying and distinguishing between sperm cells by processing visual features and motion features of imaged sperm cells, without staining or labeling in any way.

[0064] The captured video frames are fed into a computing utility (e.g., a processor) configured for analyzing the video stream using one or more pre-trained Al models. In an embodiment, the pre-trained Al models are an ensemble of two or more pre-trained Al models.

[0065] In an embodiment, the one or more pre-trained Al models comprise a first pretrained Al model and a second pre-trained Al model, the first pre-trained Al model is configured for analyzing the video stream that comprises one or more imaged frames, and providing an input to a second pre-trained Al model that is configured to classify the at least one sperm cell according to the input provided by the first pre-trained Al model.

[0066] Any of the pre-trained Al models may be a supervised Al model or an unsupervised Al model.

[0067] One embodiment of the present disclosure is a method and system for semen sexing, namely classifying the sperm cells as X chromosome-containing or Y chromosome-containing sperm cells and separating them into two populations, as aforesaid. It should be noted that such classification and separation is not absolute, but rather, it is a determination that the individual sperm cell includes the specific trait, in this case the type of sex chromosome, at a high probability. Thus, a sperm cell population enriched with X chromosome-containing sperm cells, may contain a smaller proportion of Y chromosome-containing sperm cells. The method and system of the present disclosure can be used in various practices such as livestock insemination, in vitro fertilization, genetic consultation, and the like.

[0068] Other embodiments of the present disclosure comprise differentiating, and optionally separating, dead sperm cells from living sperm cells, abnormal sperm cells from normal sperm cells, and sperm cells from non-sperm cells. As used herein the term “abnormal” sperm cell refers to a sperm cell which presents visual features outside the normal spectrum, i.e., which shows a discrepancy e.g., in a chromosomal content of a sperm cell, for example, but not limited to: aneuploidy, polyploidy, DNA fragmentation and the like; in the structure of the sperm cell, for example, but not limited to: macrocephaly, microcephaly, tapered head, amorphous head, pinhead sperm, thickened neck, thin neck, bent tail, short tail, multiple tails, coiled tail, absent tail (acephaly), long tail (flagellated head), agglutination and the like; or in the motility of a sperm cell for example, but not limited to: asthenozoospermia, hyperactivated motility, hypo activated motility, immotile sperm cell, circular motility, non-progressive motility, and the like.

[0069] The method and system by some embodiments of the present disclosure enables a determination of the fertility potential based on the sperm cells’ visual features.

[0070] The method and system of embodiments of the present disclosure can be used in offspring sex selection in assisted reproduction procedures, like artificial insemination, in vitro fertilization (IVF) and the like. Such organisms include mammals, such as, but not limited to, cows, buffalo, pigs, sheep, horses, and humans. More particularly, the semen sexing technology of the present subject matter can be used in the livestock industry, for example for artificial insemination of farm animals, and it can also be used for sexing of pets or wild animals.

[0071] Additionally, because of the safety of the semen sexing technology of the present disclosure, in terms of no DNA staining, labeling, or manipulation of the sperm cells, this technology can also be used in sexing of human semen samples, for example during assisted reproduction treatments, selection of the sex of the offspring for example in cases when one of the parents, or both, is a carrier of an X-linked disorder, and the like.

[0072] A unique characteristic of the method and system of the present disclosure is that the DNA of the sperm cells is not labeled or stained prior to or during the classification process. Moreover, the sperm cells are not manipulated by any other means, in contrast with currently existing technologies, thereby reducing potential damage to the sperm cells.

[0073] To summarize, the present disclosure provides an easy-to-use, automatic, laser- free, and label-free semen analysis and optional separation while preserving the integrity and fertility potential of the sperm cells. Referring now to Fig. 1, schematically illustrating an exemplary embodiment of a system 1, intended for the analysis of sperm cells, for example whether they are viable, and whether they contain an X or Y chromosome.

[0074] System 1 comprises an analysis sub-system 2 comprising at least one light source 12 configured to emit light and illuminate a sperm cells suspension 500 (represented by the sperm cell cartoon) passing through a flow channel 502. A video capturing utility 13 that comprises at least one camera 14 with an associated optical assembly 15, is configured to acquire images of the illuminated sperm cells suspension. A computing utility 100 operates an algorithm that executes at least one CV Al, is in data communication with video capturing utility 13 via a data link represented by arrow 16 and is configured to receive data feed representative of the captured video images and analyze the data using the CV Al algorithms, and generates an output 17 of classification of an individual sperm cell in the sperm cells suspension 500 that can be relayed to an output device 18 that may, for example, be a computer display.

[0075] The at least one light source 12 may be any suitable light that can illuminate in a manner permitting the visualization of an image captured by the video capturing utility 13. According to one embodiment, the at least one light source is natural light, for example sunlight. According to another embodiment, the light source is an artificial light in the visible, ultraviolet (UV) or infrared (IR) including (near IR light or a far IR light) range. In some embodiments the light emitted by light source 12 may be of a broadband illumination. In other embodiments the illumination may be a narrow bandwidth illumination of a distinct wavelength, a combination of two or more lights of a distinct wavelength, etc. As can be appreciated, the video-capturing utility 13 should be configured for capturing lights that match the nature of the light emitted by the light source 12. The at least one light source 12 is schematically illustrated as a light source that is positioned opposite the video capturing utility 13, whereby the sperm cells suspension 500 is illuminated by transmitted light (namely light passing through the sperm cells suspension coming from a light source that is substantially opposite the videocapturing utility). It should be noted that the light source, by some embodiments, may be reflected light, a combination of reflected and transmitted light, bright field illumination or dark field illumination, the light may be polarized, the illumination may be a phase contrast illumination or any combination of the aforesaid. The illumination by some embodiments may be continuous or may be pulsed. It may also be configured by yet other embodiments for a combination of pulsed and continuous illumination; for example, continuous at one wavelength and pulsed at another.

[0076] The captured video images captured by the video capturing utility 13 may be successive video frames or selected, non-successive video frames. Video capturing utility 13 may be an ensemble of two or more cameras, for example each recording images at a different wavelength, at a different focal point within channel 502, at a different magnification, etc. The optical assembly 15 may be geared and optimized to the nature of the light illumination and may also be configured to provide a certain image magnification or focus, as the case may be.

[0077] According to embodiments of this disclosure, the at least one camera 14 may comprise a charged-coupled device (CCD) camera, e.g., a 3 CCD camera whose imaging system uses three separate CCDs, a complementary metal oxide semiconductor (CMOS) camera, a combination of a CCD camera and a CMOS. The at least one camera 14, by some embodiments, may comprise a line scan camera, an area scan camera, a three- dimensional (3D) scan camera, or any combination of such cameras. According to some embodiments, the optical assembly 15 may comprise a set of filters, for example interchangeable filters. By some embodiments the filters may be digital filters.

[0078] According to some embodiments of this disclosure, the computer utility 100 is configured to use either all, or part of, the visual features of at least one of the images captured by the at least one camera 14, for the purpose of classifying a sperm cell in suspension 500. The analysis may comprise determining an X / Y chromosome content of individual sperm cells. The visual features of the sperm cells in suspension 500 that are analyzed may comprise (without limitation) one or more of the following: color, transparency, phase contrast, size, shape, positioning, position relative to other objects in the at least one image, orientation, bending, circularity, eccentricity, solidity, convexity, curvature, skeleton, center of mass, center of gravity, fill factor, Feret diameter, compactness, inertia tensor, fractal dimensions, density, symmetry, texture, organelle key points, length of tail, bending of tail, relative depth of the sperm cell or of parts of the sperm cell, Z position, spatial relationship between the sperm cell 500 and other objects, spatial relationship between different parts of the sperm cell 500, aspect ratio of different parts of the sperm cell 500, or any combination thereof.

[0079] Computing utility 100 may also be configured to use visual motion (also referred to herein as “motility”) features, that may be determined by the analysis of successive frames, to classify the sperm cells. The motility features include (without limitation): overall motion pattern, motion patterns of one part of the versus others, or a combination thereof. The visual motion feature, may include, but not limited to, speed of movement, motion dynamics, dynamic positioning, direction of movement, bending dynamics, swimming dynamics, response to flow of medium in which the sperm cell swims, or any combination thereof.

[0080] The data received in computing utility 100 from the at least one camera 14 may be configured to preprocess the data that may comprise, but not limited to, at least one of the following: image filtering, image enhancement, image transformation, image segmentation, image registration, image restoration, image compression, image fusion and the like.

[0081] Computing utility 100 executes the sperm cell classification algorithms that comprise pre-trained Al algorithms. The algorithms may be based on classical computer vision, machine learning and deep learning principles and are pre-trained with large data sets of sperm cells. Training the algorithm to differentiate between sperm cells may involve exposing the algorithm to images of defined cells or cell populations, optionally under the same conditions in which the classification of the sperm cells is intended to be performed. Such conditions may comprise, among others, one or more of the following: no staining; the cells not being immobilized and swimming freely in the medium; the sperm sell suspension being passed in front of the video-capturing utility under the same conditions and through a similar flow channel as in the performance of the classification; illumination that is the same as that intended for the performance of the classification; the sperm cell suspension having the same dilution as that intended during the classification; etc. There may be a variety of training techniques, the following are some examples:

[0082] (1) A ground truth approach: cells may be separated after being imaged by the image capturing utility and each cell is then examined for one or more specific biological properties or traits. Thus, each sperm cell may be tested, by histological / molecular / immunological / labeling methods and this information can then be fed to the algorithm, as part of its training set. This may, for example, involve obtaining a ground truth on the sex chromosome content. Also, such ground truth can also be the sex of the embryo that is formed when using the specific cell for IVF. In the case of fertility potential, the ground truth can be the fertilization outcome of the specific sperm cell as determined, for example, by a variety of techniques known per se. (2) A statistical approach: the algorithm can be trained by imaging a population of sperm cells separated by other techniques and know to contain a high proportion (e.g. above 80% or above 90%) of sperm cells that meet a defined criterion, e.g. population of sperm cells containing a majority of cells carrying a defined sex chromosome or sperm cells having a high-fertility potential or both. In distinction from the ground truth approach, in the statistical approach, there is a built-in uncertainty that is fed into the algorithm. The advantage of this approach is that the training set is inherently much larger than the ground truth approach.

[0083] (3) Combination of (1) and (2).

[0084] (4) The algorithm may be continuously improved or fine-tuned, for example, by continuously testing a sample of sperm cells classified as being all of a defined class and obtaining ground truth data and feeding it back to the algorithm. As noted above, the algorithm may be fine-tuned for a specific herd or individual males in a herd that exhibit properties that may be different to an extent from the average population on which the Al model of the algorithm was based and this approach may permit such fine-tuning.

[0085] The Al model at the heart of the classifying algorithm may also be continuously fine-tuned from data feed from extraneous sources that include data collected and finetuned in other Al-based sperm cells classification systems concomitantly operating in the same or remote facilities.

[0086] According to some embodiments, the computing utility 100 is configured to perform any type of segmentation method and provide any type of segmentation output accordingly, for example, but not limited to: pixel-wise segmentation method and output, super pixel segmentation method and output, regions of X / Y chromosomes segmentation method and output, background segmentation method and output, and any other segmentation method and output.

[0087] According to an exemplary embodiment, the algorithms of the computing utility 100 are configured to determine whether an individual sperm cell in suspension 500 contains an X-chromosome, or a Y-chromosome. According to yet a further exemplary embodiment, the algorithms of the analyzer 100 are configured to identify chromosomal abnormalities of individual sperm cells 500.

[0088] According to one embodiment, the analysis by computing utility 100 is based on two separate CV Al models. A first of these two may comprise a first CV Al model configured to analyze at least one image and provide an input to the second CV Al model that is configured to classify the at least one sperm cell according to the input provided by the first computer-vision Al model.

[0089] The computing utility may, by some embodiments, be configured to determine any type of irregularity in a chromosomal content of a sperm cell, for example, but not limited to aneuploidy, polyploidy, DNA fragmentation and the like. It may also, by other embodiments, be configured to determine any type of structural irregularity of a sperm cell, for example, but not limited to: macrocephaly, microcephaly, tapered head, amorphous head, pinhead sperm, thickened neck, thin neck, bent tail, short tail, multiple tails, coiled tail, absent tail (acephaly), long tail (flagellated head), agglutination and the like. Additionally, by some further embodiments, it may be configured to determine any type of motility irregularity of a sperm cell, for example, but not limited to: asthenozoospermia, hyperactivated motility, hypo activated motility, immotile sperm cell, circular motility, non-progressive motility, and the like.

[0090] Output device 18 may be a display configured to display results of the analysis of a single sperm cell or of a population of cells to a user, the display having one of a myriad of possible display configurations. The output device 18 may also be a cell sorting utility for sorting cells into separate populations that meet defined criteria, such as for the purpose of semen sexing, as described in reference to Figs. 2 and 3.

[0091] The following are non-limiting examples of types of cell separations that are encompassed by the present disclosure: i. separating sperm cells from non-sperm cells; ii. separating X chromosome-containing from Y chromosome-containing sperm cells; iii. separating dead cells from living cells; iv. separating abnormal cells from normal cells; and separating sperm cells with a fertility potential above a defined threshold from other sperm cells.

[0092] The separation can be performed at a temperature range of between about 4°C to about 39°C. The separation can be performed in any dedicated cell separation device, for example in a cell sorter or in a microfluidic unit.

[0093] Referring now to Fig. 2, schematically illustrating an exemplary embodiment of a system 1A, for separating between sperm cell populations with different characteristics, for example separating X-containing from Y-containing sperm cells within the framework of semen sexing. In Fig. 2, like elements to those of Fig, 1 were given like reference numerals and the reader is referred to the above description for an understanding of their function. In the system 1A output device 18 of Fig, 1 is constituted by a cell sorting utility 182. Cell sorting utility 182 may operate by one of a variety of cell sorting techniques, for example such techniques known per se.

[0094] The exemplary cell sorting utility 182 shown in Fig. 2 is a flow cytometry cell sorter configured to sort the sperm cells into a plurality of groups according to the classification received from the analyzer 100 through output link 17, for example, according to their sex chromosome content. The analyzed sperm cells suspension 500A passing through flow channel 502 A, may be analyzed, for example as being either X chromosome - containing sperm cells or Y chromosome-containing sperm cells and then sorted, as represent by arrows 1823 and 1825 into two (or at times more) respective collecting vessels, such as vessels 1827 and 1829.

[0095] Some exemplary cell sorters 182 are described below:

[0096] According to some embodiments of this disclosure, the cell sorter 182 is a cytometry cell sorter configured to electrically charge cells, or droplets of liquid containing at least one cell, according to their classification, and deflect the cells, or the droplets of liquid containing at least one cell, to a pre-determined collection vessel by using positively and negatively charged deflection plates 1826 and 1828, respectively.

[0097] According to other embodiments, cell sorter 182 is of a kind that deflects the sperm cells or droplets comprising sperm cells into specific vessels by the use of a laser beam; or of a kind that deflects droplets that contain sperm cells by the application of a controlled flow of air or a fluid; or of a kind that uses a controlled valving arrangement; or of a kind that deflects droplets that contain sperm cells by the use of a magnetic field; etc. It should be noted that this disclosure is not limited by the kind of cell sorting that is carried out.

[0098] An exemplary application of system 1A is preparation of semen samples for artificial insemination. There are occasions when there is a desire to increase the probability to get offspring of a certain gender - male, or female, for example in the livestock industry. Thus, separation of the sperm cells into suspension enriched with X chromosome-containing or Y chromosome-containing sperm cells increases the likelihood that the artificial insemination will give rise to a high proportion of the desired livestock gender. Other exemplary applications include (i) separating sperm cells from non-sperm cells, (ii) separating living sperm cells from non-living sperm cells, (iii) separating aberrant sperm cells from normal sperm cells, and (iv) separating sperm cells with a fertility potential above a defined threshold from other sperm cells, all intended to increase the potency of the sperm cells suspension and the likelihood that the artificial insemination will be successful and give rise to offsprings. As can be appreciated, the (i) through (iv) exemplary applications may be combined with the semen sexing application discussed above to obtain potent sperm cells suspension enriched with cells of the desired sex chromosome.

[0099] Another example of an application of system 1A is the elimination of sperm cells that contain chromosome abnormalities, namely autosome abnormalities, or sex chromosome abnormalities, or both autosome and sex chromosome abnormalities. According to some embodiments, the system 1 can identify sperm cells in the sperm cells suspension that contain an abnormal number of autosomal chromosomes, an abnormal number of sex chromosomes, namely aneuploid sperm cells 500, and the like.

[0100] Computing utility 100 is configured to execute one or more pre-trained CV Al algorithms.

[0101] According to some embodiments, the computing utility 100 is linked to a cloud computing server executing one or more such pre-trained CV Al algorithms which may (i) function to validate the analysis carried by computing utility 100, (ii) periodically update the one or more CV Al algorithms being executed on computing utility 100 with newer versions of such algorithms, (iii) receive input from the computing utility 100 to improve the algorithms on the serve, and (iv) any combination of (i) to (iii).

[0102] Referring now to Fig. 3, schematically illustrating an exemplary embodiment, in the form of a diagram presentation of an artificial neural network for determining a type of individual sperm cell. According to one embodiment, the computer utility 100 of the embodiments illustrated in Figs. 1 and 2 is configured to analyze acquired images of individual sperm cells in a sperm cells suspension 500 and 500A, respectively, and determine a type of each individual sperm cell. The uniqueness of the present disclosure is the use of an Al-based computational algorithm without the use of any dyes or markers that stain the sperm cells or bind to specific elements in sperm cells, whether during analysis or prior thereto. Many machine learning techniques may be employed to train the algorithm to identify sperm cells of interest, for example to identify X chromosomecontaining or Y chromosome-containing sperm cells.

[0103] As can be seen in Fig. 3 the exemplary artificial neural network comprises a plurality of nodes organized in layers. These comprise an input layer 42, with a plurality of input nodes 422, one or more hidden layers 44, two - 442 and 444 in this specific example, each with a plurality of respective hidden nodes 4422 and 4442, and an output layer 46 with a number of output nodes 462. The nodes are interconnected with weighted connections 48, and calculations are performed between each node and the nodes in the next layer.

[0104] Generally, the flow of calculation is from the input layer 42, through the hidden layers 44 toward the output layer 46. However, as known per se, during the analysis, the flow of calculation can also be in the opposite direction.

[0105] The input nodes 422 receive input data, for example, different tagged images of individual sperm cells. For example, one input node 422 receiving data relating to the size of a sperm cell, another receiving data on velocity of swimming of the cell, one receiving data relating to size of the head of the sperm cell, one receiving data on the tail length of the sperm cell. Calculations on these data are performed in the nodes of hidden layer 44, and results of the calculations are forwarded to the nodes of the output layer 46. For example, one of output nodes 462 receiving an indication when the sperm cell contains an X-chromosome, and another output node 462 receiving an indication when the sperm cell contains a Y-chromosome.

Claims

CLAIMS:

1. A method for classification of mammalian sperm cells comprising passing a suspension comprising motile sperm cells suspended in a medium in front of a video capturing utility under conditions that permit said utility to capture images of single sperm cells, wherein the medium is devoid of any cell-staining or biomarker- identifying dyes and the cells in the medium were not pre-stained with such dyes, capturing one or more video frames of individual sperm cells in the medium, feeding the captured video frames into a computing utility configured for analyzing the video stream using one or more pre-trained Al models to detect individual sperm cells and one or more sperm cell features of the individual sperm cells, and classifying the detected sperm cells as belonging to a defined category based on the sperm cell features identified in said analyzing.

2. The method of claim 1, wherein the features comprise one or more visual motion features of at least one part of the sperm cells.

3. The method of claim 1 or 2, wherein the analysis comprises analysis of visual structure-related features of at least one part of the sperm cells.

4. The method of any one of claims 1 to 3, wherein said classifying comprises classifying the sperm cells as X chromosome-containing or Y chromosome-containing sperm cell.

5. The method of any one of claims 1 to 4, wherein said classifying comprises classifying the sperm cell as being dead cell or living cell.

6. The method of any one of claims 1 to 4, wherein said classifying comprises classifying the sperm cell as being normal cell or abnormal cell.

7. The method of any one of claims 1 to 6, wherein said classifying comprises classifying sperm cells according to their fertilization potential.

8. The method of any one of claims 1 to 7, comprising differentiating sperm cells from non-sperm cells.

9. The method of any one of claims 1 to 8, wherein said analyzing comprises feeding data representative of the video capturing into a computing utility configured to analyze the data using one or more pre-trained Al models.

10. The method of claim 9, wherein the pre-trained Al models are an ensemble of two or more pre-trained Al models.

11. The method of claim 10, wherein the one or more pre-trained Al models comprise a first pre-trained Al model and a second pre-trained Al model, the first pre-trained Al model is configured for analyzing the video stream and providing an input to a second pre-trained Al model.

12. The method of any one of claims 9 to 11, wherein at least one of the Al models is based on one or more of the following: computer vision-based Al, deep learning, and computer vision-based Al combined with machine learning.

13. The method of any one of claims 1 to 12, comprising physically separating sperm cells with defined characteristics from other cells.

14. The method of claim 13, wherein said separating comprises one or more of the following:(i) separating sperm cells from non-sperm cells;(ii) separating X chromosome-containing from Y chromosome-containing sperm cells;(iii) separating dead cells from living cells;(iv) separating abnormal cells from normal cells; and(v) separating sperm cells with a fertility potential above a defined threshold from other sperm cells.

15. The method of any one of claims 13 or 14, for obtaining a sperm cell-containing suspension with sperm cell population enriched with X chromosome-containing or Y chromosome-containing sperm cells.

16. The method of any one of claims 13 to 15, wherein said separating is performed in a cell sorter or in a microfluidic unit.

17. The method of any one of claims 13 to 16, wherein said separating is performed at a temperature range of between about 4°C to about 39°C.

18. A system for classification of mammalian sperm cells comprising: a flow channel configured for transferring a medium devoid of any cell-staining or biomarker-identifying dyes and comprising motile sperm cells suspended therein not pre-stained with such dyes, therethrough; a video capturing utility configured for capturing images of single sperm cells flowing through the flow channel; and a computing utility running one or more pre-trained Al models and configured for receiving the captured video frames from the video capturing utility and for analyzing thevideo stream to (i) detect individual sperm cells and one or more sperm cells features of the individual sperm cells and (ii) classify the detected sperm cells as belonging to a defined category based on said features.

19. The system of claim 18, comprising at least one light source configured for illuminating the medium passing in the flow channel.

20. The system of claim 18 or 19, wherein the computing utility is configured for analysis of visual motion features of at least one part of the sperm cells.

21. The system of claim 18 or 19, wherein the computing utility is configured for analysis of visual structure -related signals of at least one part of the sperm cells.

22. The system of any one of claims 18 to 21, wherein said defined category comprises an indication of whether an analyzed sperm cell is an X chromosomecontaining or a Y chromosome-containing sperm cell.

23. The system of any one of claims 18 to 22, wherein said defined category comprises an indication of whether an analyzed sperm cell is a dead cell or living cell.

24. The system of any one of claims 18 to 22, wherein said defined category comprises an indication of whether an analyzed sperm cell is a normal cell or abnormal cell.

25. The system of any one of claims 18 to 24, wherein said category comprises a prediction of a fertility potential.

26. The system of any one of claims 18 to 25, wherein said category comprises an indication whether a detected cell is a sperm cell or a non-sperm cell.

27. The system of any one of claims 18 to 26, wherein the pre-trained Al model is an ensemble of two or more pre-trained Al models.

28. The system of claim 27, wherein the one or more pre-trained Al models comprise a first pre-trained Al model and a second pre-trained Al model, the first pre-trained Al model is configured for analyzing the video stream and providing an input to a second pre-trained Al model.

29. The system of any one of claims 18 to 28, wherein at least one of the Al models is based on one or more of the following: computer vision-based Al, deep learning, and computer vision-based Al combined with deep learning.

30. The system of any one of claims 18 to 29, comprising a cell separating utility controllable by the computing utility and configured for separating sperm cells with defined characteristics from other cells.

31. The system of claim 30, wherein said separating utility is a cell sorter or a microfluidic unit.

32. The system of any one of claims 30 or 31, wherein said separating is performed at a temperature range of between about 4°C to about 39°C.

33. The system of any one of claims 28 to 32, for obtaining a sperm cell-containing suspension with sperm cell population enriched with X chromosome-containing or Y chromosome-containing sperm cells.

34. The system of any one of claims 28 to 32, for separating sperm cells for yielding one or more of the following:(i) sperm cells separated from non-sperm cells;(ii) X chromosome-containing sperm cells or Y chromosome-containing sperm cells separated from, respective, Y chromosome-containing sperm cells or X chromosome-containing sperm cells;(iii) living cells separated from dead cells;(iv) abnormal cells separated from normal cells; and(v) sperm cells with a fertility potential above a defined threshold separated from other sperm cells.

35. The system of any one of claims 18 to 34, wherein the light source is selected from a group consisting of an artificial visible light source, a natural visible light source, an ultraviolet (UV) light source, a far infra-red (IR) light source, and a near IR light source.

36. The system of any one of claims 18 to 35, wherein the light source is configured to provide light in a defined wavelength.

37. The system of any one of claims 18 to 36, wherein the light source is configured to provide light of at least two distinct wavelengths.

38. The system of any one of claims 18 to 37, wherein the light source is configured to illuminate the flow channel by reflected light, by transmitted light, or by a combination of reflected light and transmitted light.

39. The system of any one of claims 18 to 38, wherein the system is configured to illuminate the flow channel by one or more of Bright Field illumination, Dark Field illumination, polarized illumination, and Phase Contrast illumination.

40. The system of any one of claims 18 to 39, wherein the light source is configured to illuminate the medium by one or both of continuous illumination and pulsed illumination.

Images