Method and apparatus for categorising fish oocytes

Abstract

A method of analysing a sample of a plurality of fish oocytes (100) is disclosed The method comprises, for each oocyte in the sample: capturing a first set of images of the oocyte from a plurality of rotational angels when rotated about a first axis of rotation; capturing a second set of images of the oocyte from a plurality of rotational angels when rotated about a second axis of rotation, which second axis of rotation is orthogonal to the first axis of rotation; identifying at least one characteristic (104) of lipid droplet coalescence in the captured first and second sets of images; and based on the identified at least one characteristic of lipid droplet coalescence, classifying the oocyte as belonging to one of a plurality of predefined categories, and based on the distribution of the sampled oocytes in the plurality of categories, attributing a quality parameter to the sample of oocytes, the quality parameter being any one of: a predicted fertilisation rate of the sampled oocytes; a predicted eyed-embryo success rate of the sampled oocytes; and a predicted hatching success rate of the sampled oocytes. A computer-implemented method of attributing a quality measure to a sample of fish oocytes is also disclosed, as is a related computer program and a related oocyte analysis apparatus. A computer-implemented method for training a neural network for use in a system for analysing a sample of fish oocytes is also disclosed.

Description

[0001] METHOD AND APPARATUS FOR CATEGORISING FISH OOCYTES

[0002] Field of the invention

[0003] The present disclosure relates to a method for analysing fish oocytes, in particular with the purpose of predicting at least one of: a fertilisation success rate; an eyed-embryo success rate; and a hatching success rate of a batch of oocytes drawn from a fish female of the family Salmonidae, e.g. the species Salmo salar.

[0004] The disclosure also relates to a computer-implemented method of attributing a quality measure to a sample of fish oocytes, a computer program having instructions which when executed by a computing device or system causes the computing device to perform such a method, and an oocyte analysis apparatus comprising a processor and a memory being configured to perform said method.

[0005] The disclosure further relates to a computer-implemented method for training a neural network for use in a system for analysing a sample of fish oocytes.

[0006] Background

[0007] An oocyte is the female gamete involved in reproduction. In this disclosure the term “oocyte” is used for both un-ovulated oocytes and ovulated ova.

[0008] In salmonid aquaculture fish farming, oocytes are traditionally collected from fish females directly from the ovaries after slaughter but can also be obtained from living broodstock, so called stripping. Spermatozoa and oocytes are mixed to allow the oocytes to be impregnated and fertilized. The fertilised oocytes are then incubated and hatched in an incubator.

[0009] It is known that the hatching success rate in a batch of fertilised oocytes may vary. Therefore, to increase the general efficiency of aquaculture fish farming, it is of interest to identify and select those batches being the most likely to yield viable hatchlings or fry.

[0010] An easy and reliable tool to predict the quality of oocytes in fish is lacking in the aquaculture industry, especially for salmonids. Oocytes with low quality are not easily traced in in eyed- egg production facilities. Non-viable oocytes may turn white but are usually only detectable after the fertilisation process.

[0011] Currently, there is no system commercially available that is capable of distinguishing oocytes based on the visual appearance except when they are over-ripe.

[0012] Lipid droplets are found in almost all cells contributing to various cellular functions. Mainly, lipid droplets provide lipid storage for energy generation, membrane synthesis and protein degradation. Teleost fish oocytes have a significant amount of lipids that are essential for embryonic and larval growth. In some oocytes the lipid droplets have the tendency to coalesce in one pole of the oocyte. Distribution of lipid droplets in oocytes has been studied in several salmonid species such as brown trout (Salmo trutta fario) and Arctic charr (Salvelinus alpinus). In these studies, different patterns of lipid droplet distribution and sizes have been demonstrated to be associated to the quality of oocytes.

[0013] In Mansour, N., Lahnsteiner, F., & Patzner, R. A. (2007), “Distribution of lipid droplets is an indicator for egg quality in brown trout, Salmo trutta farter, Aquaculture, 273(4), 744- 747. doi:https: / / doi.org / 10.1016 / j. aquaculture.2007.09.027, (Mansour et al., 2007), oocytes were classified into four categories where category 1 consisted of evenly spread lipid droplets, category 2 had some droplets starting to coalesce and category 3 had one or more coalesced areas. Lastly, in category 4 the majority of lipid droplets were coalesced.

[0014] In Mansour, N., Lahnsteiner, F., McNiven, M. A., & Richardson, G. F. (2008), “Morphological characterization of Arctic char, Salvelinus alpinus, eggs subjected to rapid post-ovulatory aging at 7 °C”, Aquaculture, 279(1), 204-208, (Mansour et al., 2008) three different patterns of lipid distributions and their possible relation to oocyte quality were investigated.

[0015] In both studies it was found that oocytes with evenly distributed lipid droplets give higher fertilisation rates than oocytes with coalesced lipid droplets.

[0016] However, it has not been possible to transfer this knowledge into a tool capable of predicting the quality of oocytes in fish, which tool is sufficiently fast and reliable to meet the requirements of the aquaculture industry.

[0017] One object of the present disclosure is to address this problem and provide an apparatus and a method to a provide a practical and cost-effective method to predict fish oocyte quality upon oocyte collection, in particular for salmonoid oocytes, e.g. oocytes of the species Salmo salar, and especially with the purpose of determining the probability of the oocyte developing to viable hatchling if fertilised.

[0018] Another object of the disclosure is to provide an apparatus and a method for assessing the quality of fish oocytes having a high through-put.

[0019] Summary of the invention

[0020] According to a first aspect, the present disclosure provides a method of analysing a sample of a plurality of fish oocytes, comprising:

[0021] - for each oocyte in the sample:

[0022] - capturing a first set of images of the oocyte from a plurality of rotational angels when rotated about a first axis of rotation;

[0023] - capturing a second set of images of the oocyte from a plurality of rotational angels when rotated about a second axis of rotation, which second axis of rotation is orthogonal to the first axis of rotation; - identifying at least one characteristic of lipid droplet coalescence in the captured first and second sets of images; and

[0024] - based on the identified at least one characteristic of lipid droplet coalescence, classifying the oocyte as belonging to one of a plurality of predefined categories, and

[0025] - based on the distribution of the sampled oocytes in the plurality of categories, attributing a quality parameter to the sample of oocytes, the quality parameter being any one of: a predicted fertilisation rate of the sampled oocytes; a predicted eyed- embryo success rate of the sampled oocytes; and a predicted hatching success rate of the sampled oocytes.

[0026] The step of classifying the oocyte as belonging to one of the plurality of predefined categories may comprise supplying the first and second sets of images to a machine learning system having been trained on a dataset of oocyte images comprising:

[0027] - at least one first class of oocytes images depicting oocytes being associated with at least one of: a first fertilisation rate; a first eyed-embryo success rate; and a first hatching success rate; and

[0028] - at least one second class of oocytes images depicting oocytes being associated with at least one of: a second fertilisation rate; a second eyed-embryo success rate; and a second hatching success rate, wherein at least one of said first fertilisation rate, said first eyed-embryo success rate, and said first hatching success rate is higher than said second fertilisation rate, said second eyed- embryo success rate, and said second hatching success rate, respectively.

[0029] The at least one first class of oocytes images may comprise images displaying oocytes having a lipid droplet coalescence region covering less than a predetermined ratio of the total area of the depicted oocyte, and the at least one second class of oocytes images may comprise images displaying oocytes having a lipid droplet coalescence region covering more than the predetermined ratio of the total area of the depicted oocyte.

[0030] The predetermined ratio may be any one of: within the range of 11% - 14%; within the range of 12% - 13%; and 12%.

[0031] The at least one first class of oocytes images may comprise images displaying oocytes having a lipid droplet coalescence region having an area size less than a predetermined area size value, and wherein said at least one second class of oocytes images comprises images displaying oocytes having a lipid droplet coalescence region having an area size larger than the predetermined area size value.

[0032] The predetermined area size value may be any one of: within the range of 1.3 mm2- 1.7 mm2; within the range of 1.4 mm2- 1.6 mm2; and 1.5 mm2. The at least one characteristic of lipid droplet coalescence may comprise the area size of a lipid droplet coalescence region in the oocyte, and the method may comprise, for each oocyte in the sample, measuring or evaluating the area size of the lipid droplet coalescence region in the captured first and second sets of images.

[0033] The plurality of categories may comprise a first category for oocytes having a lipid droplet coalescence area size being less than a predetermined area size value and a second category of oocytes having a lipid droplet coalescence area size being larger than the predetermined area size value. The predetermined area size value is any one of: within the range of 1.3 mm2- 1.7 mm2; within the range of 1.4 mm2- 1.6 mm2; and 1.5 mm2.

[0034] The at least one characteristic of lipid droplet coalescence may comprise a relative area size of a lipid droplet coalescence region in the oocyte, and the method may comprise, for each oocyte in the sample:

[0035] - measuring or evaluating the area size of the lipid droplet coalescence region and the area size of the oocyte in the captured first and second sets of images; and

[0036] - calculating the relative area size of the lipid droplet coalescence region by dividing the measured or evaluated area size of the lipid droplet coalescence region with the measured or evaluated area size of the oocyte.

[0037] The plurality of categories may comprise a first category for oocytes having a lipid droplet coalescence relative area size being less than a predetermined relative area size value and a second category for oocytes having a lipid droplet coalescence relative area size being greater than the predetermined relative area size value.

[0038] The predetermined relative area size value may be any one of: within the range of 0.11 to 0.14; within the range of 0.12 to 0.13; and 0.12.

[0039] According to a second aspect, the present disclosure provides a computer-implemented method of attributing a quality measure to a sample of fish oocytes, comprising the steps of:

[0040] - for each oocyte in the sample:

[0041] - obtaining at least one image, taken or captured by an imaging device, of an oocyte of the sample depicting lipid droplets in the oocyte; and

[0042] - inputting the obtained at least one image to a trained learning model and classifying the oocyte as belonging to one of a plurality of predefined categories based on coalescence of the lipid droplets;

[0043] - based on the distribution of the sampled oocytes in the plurality of categories, attributing the quality parameter to the sample of oocytes, the quality parameter being any one of: a predicted fertilisation rate of the sampled oocytes; a predicted eyed- embryo success rate of the sampled oocytes; and a predicted hatching success rate of the sampled oocytes. The obtained at least one image may comprise:

[0044] - a first set of images of the oocyte taken or captured by the imaging device from a plurality of rotational angels when the oocyte is rotated about a first axis of rotation; and

[0045] - a second set of images of the oocyte taken or captured by the imaging device from a plurality of rotational angels when the oocyte is rotated about a second axis of rotation, which second axis of rotation is orthogonal to the first axis of rotation.

[0046] According to a third aspect, the present disclosure provides a computer program having instructions which when executed by a computing device or system causes the computing device to perform the method according to the second aspect.

[0047] According to a fourth aspect, the present disclosure provides a fish oocyte analysis apparatus comprising a processor and a memory and being configured to perform the method according to the second aspect.

[0048] According to a fifth aspect, the present disclosure provides a computer-implemented method for training a neural network for use in a system for analysing a sample of fish oocytes, the method comprising the steps of

[0049] - inputting training data to the neural network to train the neural network through machine learning; wherein the training data comprises:

[0050] - data representing coalescence of lipid droplets in fish oocytes; and

[0051] - a scoring data on any one of; an oocyte fertilisation rate of; an oocyte eyed-embryo success rate; and an oocyte hatching success rate.

[0052] Said machine learning system may be a deep learning system comprising a neural network, e.g. implemented using the TENSORFLOW toolkit. Likewise, said trained learning model may have been trained using the TENSORFLOW toolkit.

[0053] In each of said first, second, third, fourth and fifth aspects said fish oocytes may be any one of: Salmonidae oocytes; and oocytes of the species Salmo salar.

[0054] Above-discussed preferred and / or optional features of each aspect of the disclosure may be used, alone or in appropriate combination, in the other aspects of the disclosure.

[0055] The claimed invention is specified in the independent claims of this application. Advantageous adaptations and versions of the claimed invention are specified in the independent claims.

[0056] Description of the drawings Following drawings are appended to facilitate the understanding of the disclosure and the invention as it is defined in the claims:

[0057] Figs, la-lf are images of Atlantic salmon oocytes (Salmo salary,

[0058] Fig. 2 is a diagram showing the distribution of oocytes in five different lipid droplet coalescence categories;

[0059] Fig. 3 is a diagram showing the distribution of oocytes in two sub-groups of the lipid droplet coalescence categories in Fig. 2;

[0060] Fig. 4a is a box plot showing the ratio of oocytes in respective sub-group having a fertilisation rate of >90% and <90%, respectively;

[0061] Fig. 4b is a box plot showing the ratio of oocytes in respective sub-group having an eyed- embryo success rate of >70% and <70%, respectively;

[0062] Fig. 4c is a box plot showing the ratio of oocytes in respective sub-group having a hatching success rate of >50% and <50%, respectively;

[0063] Fig. 5 illustrates an embodiment of a fish oocyte analysis apparatus;

[0064] Fig. 6 shows a detailed part-view of the apparatus according to Fig. 5 with covers removed for clarity;

[0065] Fig. 7 shows a detailed part-view of the apparatus according to Fig. 5 from an alternative angle;

[0066] Fig. 8 shows a detailed part-view of the apparatus according to Fig. 5 with a disc of a conveyor system of the apparatus removed for clarity;

[0067] Fig. 9 shows a further detailed part-view of the apparatus according to Fig. 5;

[0068] Fig. 10 is a flow chart schematically illustrating an embodiment of a method for operating a fish oocyte analysis apparatus;

[0069] In the drawings, like reference numerals have been used to indicate common parts, elements or features unless otherwise explicitly stated or implicitly understood by the context.

[0070] Detailed description of the invention

[0071] In the following, one or more specific embodiments of the invention will be described in more detail with reference to the drawings. However, it is specifically intended that the invention is not limited to the embodiments and illustrations contained herein but includes modified forms of the embodiments including portions of the embodiments and combinations of elements of different embodiments as come within the scope of the following claims. It should be appreciated that in the development of any such actual implementation, as in any engineering or design project, numerous implementation- specific decisions must be made to achieve the developer’s specific goals, such as compliance with system and / or business-related constraints, which may vary from one implementation of the invention to another. Moreover, it should be appreciated that such a development effort might be complex and time consuming, but would nevertheless be a routine undertaking of design, fabrication and manufacture for the skilled person having the benefit of this disclosure.

[0072] Figs, la-lf are images of six different oocytes 100 of the species Salmo salar (Atlantic salmon) showing lipid droplets 102 inside the oocytes, wherein:

[0073] Fig. la shows an oocyte in which the lipid droplets are evenly distributed throughout the oocyte;

[0074] Fig. lb shows an oocyte having a region 104 in which some of the lipid droplets have coalesced forming a small lipid droplet coalescence or globule having a relatively small area size;

[0075] Fig. 1c shows an oocyte having a region 104 in which some of the lipid droplets have coalesced forming a lipid droplet coalescence having a somewhat larger area size than the lipid droplet coalescence in Fig. lb;

[0076] Fig. Id and Fig. le show oocytes having a region 104 in which some of the lipid droplets have coalesced forming a lipid droplet coalescence having a larger area size than the lipid droplet coalescence in Fig. 1c; and

[0077] Fig. If shows an oocyte having a region 104 in which almost all of the lipid droplets have coalesced and formed into a large lipid globule.

[0078] The inventors of the present invention have found that by identifying lipid droplets depicted in images of fish oocytes and measuring at least one characteristic of the lipid droplets, the fertilisation success rate, the eyed-embryo success rate, and / or the hatching success rate of a batch of oocytes can be predicted.

[0079] As is known in the art, the fertilisation success rate is the ratio between the number of oocytes that are successfully fertilised and the total number of oocytes that are mixed with milt in a fertilisation process. The eyed embryo success rate of a batch of fertilised eggs is the ratio between the number of eyed embryos, i.e. embryos that reach the eyed-egg stage, and the total number of fertilised eggs. The “eyed-egg stage”, sometimes also referred to as the “eyed-embryo stage” is the stage during incubation when the eyes of developing salmonide embryos, i.e. embryos of the family Salmonidae, can be seen through the transparent egg membrane / chorion. This usually occurs about halfway through incubation. The hatching success rate is the ratio between the number of successfully hatched larvae, i.e. hatchlings, and the total number of fertilised eggs. In a study the inventors have analysed batches of oocytes collected from Salmo salar (Atlantic salmon) females.

[0080] Oocytes in ovarian fluid from 100 females were collected on a single day and stored at 4°C until the next day. Upon collection, aliquots of the oocytes were fertilised with milt from different males, as part of the production at a broodstock facility. The fertilisation rate of each female was measured and based on the fertilisation rates oocytes (unfertilised) from 16 females were collected for further analyses. In particular, oocytes from 8 females having fertilisation rates >90% (high quality) and 8 females having fertilisation rates between 85- 0% (sub-optimal quality) were collected for further analyses.

[0081] The oocyte batches from each of the 16 females were further split into a first sub-batch and a second sub -batch.

[0082] The oocytes of the first sub-batches were subjected to a fertilisation trial in which the oocytes were mixed with milt. To minimize the male factor, cryopreserved milt from a single male was used for all 16 females.

[0083] In the fertilisation trial the oocytes from each first sub-batch (i.e. from each female) were fertilised with cryopreserved milt at a sperm to egg ratio of 2x106: 1. The oocytes were mixed with milt and activated by AquaBoost® activator. After incubation for 2 minutes the eggs were rinsed with 0.9% saline water to remove excess spermatozoa and ovarian fluid. To obtain initial fertilisation rates for the selected females, approximately 100 fertilised eggs were transferred to a separate incubator system and the fertilisation rate (8-16 cell stage) was assessed for all 16 females. The remaining eggs from each first sub-batch (i.e. from each female) were further incubated until 79 days post fertilisation (dpf) and the eyed-embryo success rate and the hatching success rate of the oocytes from each female were measured. Dead eggs (white eggs) were removed every week to minimise contamination.

[0084] The fertilisation trial confirmed that the collected oocyte batches had fertilisation rates ranging between 0-100% (see Table 1 below). However, oocytes from 9 females showed fertilisation rates >90% (high quality) and 7 with fertilisation rates < 80% (sub-optimal quality). The subsequent eyed-embryo success rate and hatching success rate for some oocyte batches - i.e., Fish ID 3, 5 and 15 - were observed to be low despite having a high fertilisation rate. Also, fish ID 12 showed a substantially decrease in eyed-embryo success rate - ending up with 0% hatching success rate - even though having a fertilisation rate of 76.9%. Quality based on

[0085] Fish ID Fertilisation rate (%) Eyed-embryo success (%} Hatching success (%) fertilisation rate

[0086] 1 97.1 94.3 94.0Hlgh

[0087] 2 97.5 87.0 93.5H,gh

[0088] 3 90.5 63.5 60.1Hlgh

[0089] 4 95.3 87.0 85.7Hlgh

[0090] 5 90.3 79.7 81.5Hlgh

[0091] 6 93.5 90.6 88.7Hlgh

[0092] 7 97.4 81.8 76.8Hlgh

[0093] 8 93.1 92.2 94.0Hlgh

[0094] 9 5.8 0.0 0.0Low

[0095] 10 71.8 50.0 33.3Low

[0096] 11 40.8 19.3 11.9Low

[0097] 12 76.9 1.0 0Low

[0098] 13 67.0 69.8 64.9Low

[0099] 14 66.0 52.6 44.6Low

[0100] 15 91.6 76.6 48.8Hlgh

[0101] 16 73.8 56.8 54.2Low

[0102] Table 1. Fertilisation rate, eyed-embryo success rate and hatching success rate of the oocytes from 16 selected Salmo salar (Atlantic salmon) females.

[0103] The oocytes of the second sub-batches were subjected to an optical analysis. From the oocytes of each second sub-batch, i.e. from each female, approximately 20 oocytes were picked randomly, and images were then captured of each oocyte from two directions using a stereoscopic zoom microscope at 10 X magnification.

[0104] Each oocyte was manually inspected in order to identify any region of coalesced lipid droplets in the oocyte. A first image was then captured of the oocyte from a first direction most prominently displaying the coalesced lipid droplets (if present in the oocyte). As previously discussed, if lipid droplets coalesce, they tend to do so at one pole of the oocyte.

[0105] If no coalesced lipid droplets were found in the oocyte during the manual inspection, the first image was captured from an arbitrary direction.

[0106] Consequently, if the lipid droplets have coalesced in the oocyte, the first image would reveal a first hemisphere of the oocyte showing a region of lipid droplet coalescence and, possibly, non-coalesced lipid droplets. If the lipid droplets have not coalesced in the oocyte, the first image would reveal a first hemisphere of the oocyte showing only non-coalesced lipid droplets.

[0107] After capturing the first image, the oocyte was rotated 180° and a second image of the oocyte was captured from a direction opposite from which the first image was captured. Capturing of the second image of the oocyte was performed to exclude the presence of a lipid droplet coalescence in the second hemisphere of the oocyte (i.e. the hemisphere opposite the first hemisphere), thus confirming that the lipid droplet coalescence, if indeed present in the oocyte, was captured in the first image.

[0108] During the imaging process, the oocytes were submerged in phosphate buffered saline (PBS) to maintain the moisture in the sample.

[0109] In the study the average area size of the oocytes in the images, i.e. the size of the projected area of an oocyte, was approximately 11.5 mm2-12 mm2.

[0110] Based on the distribution of lipid droplets in the captured images, each oocyte was classified according to the following:

[0111] Category 1 : lipid droplets evenly distributed throughout the oocyte (i.e. no region of lipid droplet coalescence present);

[0112] Category 2: one or two small lipid globules starting to coalesce in a smaller region with average area size of 0.46 mm2± 0.29 mm2, i.e. corresponding to approximately 1.5% - 6.5% of the total area of the oocyte;

[0113] Category 3 : more lipid droplets starting to coalesce and spreading into a greater region with an average area size of 0.99 mm2± 0.31 mm2, i.e. corresponding to approximately 5.7% - 11.3% of the total area of the oocyte;

[0114] Category 4: large region with coalesced lipid droplets with an average area size of 2.43 mm2± 0.76 mm2, i.e. corresponding to approximately 13.9% - 27.7% of the total area of the oocyte; and

[0115] Category 5: lipid droplets almost all coalesced and forming one large lipid globule with an average area size of 8.19 mm2± 2.65 mm2, i.e. corresponding to approximately 46.2% - 94.3% of the total area of the oocyte.

[0116] The area size referred to above is the area size of the lipid droplet coalescence region as projected onto the captured two-dimensional image.

[0117] Fig. la represents an oocyte in Category 1, Fig. lb an oocyte in Category 2, Fig. 1c an oocyte in Category 3, Figs. Id and le oocytes in Category 4, and Fig. If an oocyte in Category 5. In hatcheries oocytes in Category 5 are usually referred to as over-ripe.

[0118] Fig. 2 is a diagram showing the distribution of the oocytes among the different categories. The majority of the 16 females had oocytes in at least four categories and the number of oocytes in each category varied between the females. Category 5 oocytes were observed only in five females (in fish ID 7, 10, 11, 12 and 13) along with other categories. By sub-grouping the categories into a first sub-group X and a second sub-group Y - subgroup X containing Categories 1-3 and sub-group Y containing Categories 4 and 5 - a significant correlation with fertilisation rate, eyed-embryo success rate and hatching success rate was found.

[0119] Fig. 3 is a diagram showing the distribution of oocytes in sub-group X (oocytes in Categories 1, 2 and 3) and in sub-group Y (oocytes in Categories 4 and 5) for each female. After subgrouping the oocytes accordingly, two females (fish ID 2 and 8) had oocytes only in subgroup X whereas one female (fish ID 10) had more than 95% of the oocytes in sub-group Y.

[0120] The correlation between sub-group X and high rates of each of the fertilisation rate, the eyed- embryo success rate and the hatching success rate was found to be significant. In particular, Category 2 contributed significantly to predict oocyte quality.

[0121] Figs. 4a-4c are box plots presenting differences between sub-groups X and Y using a Kolmogorov-Smirnov test, wherein Fig. 4a shows the difference between sub-group X (Categories 1, 2 and 3) and sub-group Y (Categories 4 and 5) calculated based on the fertilisation rate (Kolmogorov-Smirnov D = 0.857, p = 0.002), Fig. 4b shows the difference between sub-group X and sub-group Y calculated based on the eyed-embryo success rate (Kolmogorov-Smirnov D = 1.000, p = 0.0002), and Fig. 4c shows he difference between sub-group X and sub-group Y calculated based on the hatching success rate (Kolmogorov D = 0.909, p = 0.002).

[0122] According to the results presented in Fig. 4b, a higher eyed-embryo success rate (>70%) was obtained when there were more oocytes (>75%) from sub-group X present in the oocyte batch. The eyed-embryo success rate was significantly reduced (<70%) when more oocytes (>40%) from sub-group Y were present in the oocyte batch. In other words, the females that had the majority of oocytes in Categories 1,2 and 3 were observed to have high fertilisation rates, and lower fertilisation rates were seen in females that had more oocytes from Categories 4 and 5.

[0123] A similar pattern is shown in Figs. 4a and 4c between sub-groups and subsequent fertilisation rates and hatching success rates, which resulted in a higher fertilisation rate and hatching success rate in the presence of more oocytes from sub-group X and low number of oocytes from sub-group Y.

[0124] In addition, a Spearman’s correlation analysis between sub-groups and fertilisation outcome (fertilisation rate, eyed-embryo success rate, and hatching success rate) reveals that subgroup X shows a positive correlation for fertilisation rate (rs = 0.624), eyed-embryo success rate (rs = 0.723) and hatching success rate (rs = 0.754) while negative correlations is observed for sub-group Y with fertilisation rate (rs= - 0.624), eyed-embryo success rate (rs = -0.723) and hatching success rate (rs = -0.754).

[0125] The results show that oocyte batches that belong to sub-group X show a high fertilisation rate, a high eyed-embryo success rate and a high hatching success rate, whereas the corresponding rates for oocytes from sub-group Y are low. The results also show that whilst the correlation between the sub-groups is the strongest for eyed-embryo success rate, there is also a strong correlation between the sub-groups and the fertilisation rate and the hatching success rate, respectively.

[0126] In other words, the results indicate that the characteristics of lipid droplets in an oocyte, and in particular the area size of lipid droplet coalescence in an oocyte, can be used as an indicator for oocyte quality in Salmonidae family species, in particular for providing a quality measure for the fertilisation rate, the eyed-embryo success rate and / or the hatching success rate for a batch of oocytes. In an aquaculture setting, this can be utilised to provide a fast and efficient method of evaluating the quality of oocytes produced by a broodstock, e.g. to identify females in the broodstock producing high quality oocytes and / or to verify the quality of the oocytes collected from broodstock females.

[0127] The above findings can be utilised for providing an improved method of analysing samples of Salmonidae oocytes.

[0128] Such a method may comprise, for each oocyte in a sample:

[0129] - capturing a first set of images of the oocyte from a plurality of rotational angels when rotated about a first axis of rotation ;

[0130] - capturing a second set of images of the oocyte from a plurality of rotational angels when rotated about a second axis of rotation, which second axis of rotation is orthogonal to the first axis of rotation;

[0131] - identifying at least one characteristic of lipid droplet coalescence in the captured first and second sets of images; and

[0132] - based on the identified at least one characteristic of lipid droplet coalescence, classifying the oocyte as belonging to one of a plurality of predefined categories.

[0133] Based on the distribution of the sampled oocytes in the plurality of categories, a quality parameter or measure can then be attributed to the sample of oocytes, which quality parameter may be any one of: a predicted fertilisation rate of the sampled oocytes; a predicted eyed-embryo success rate of the sampled oocytes; and a predicted hatching success rate of the sampled oocytes.

[0134] In other words, the method involves measuring or evaluating at least one characteristic of the lipid droplets in each oocyte in the sample and classifying each oocyte according to the identified characteristic or characteristics. The sample of oocytes can then be attributed the quality parameter or measure based on how the oocytes distribute between the categories, i.e. based on how many of the oocytes are allocated to each category. In an aquaculture setting, this will allow different samples of oocytes to easily be compared with each other, e.g. to allow samples of oocytes from different females to be compared with each other. According to the method, the classification of each oocyte may be based on to what extent the lipid droplets in the respective oocyte have coalesced, and the plurality of categories may comprise categories representing different levels or extent of lipid droplet coalescence.

[0135] A machine learning system may advantageously be used in the method. In particular, the step of classifying the oocyte as belonging to one of the plurality of predefined categories may comprise supplying the first and second sets of images to a machine learning system having been trained on a dataset of oocyte images comprising:

[0136] - at least one first class of oocytes images depicting oocytes being associated with at least one of a first fertilisation rate; a first eyed-embryo success rate; and a first hatching success rate; and

[0137] - at least one second class of oocytes images depicting oocytes being associated with at least one of a second fertilisation rate; a second eyed-embryo success rate; and a second hatching success rate, wherein at least one of said first fertilisation rate, said first eyed-embryo success rate, and said first hatching success rate is higher than said second fertilisation rate, said second eyed- embryo success rate, and said second hatching success rate, respectively.

[0138] The at least one first class of oocytes may for example classes corresponding to the abovediscussed Category 1, Category 2 and Category 3, and the least one second class of oocytes may for example classes corresponding to the above-discussed Category 4 and Category 5. As previously discussed, it has been demonstrated that oocytes in Category 1, Category 2 and Category 3 have a relatively high fertilisation rate, eyed-embryo success rate and hatching success rate, respectively, whereas oocytes in Category 4 and Category 5 have a relatively low fertilisation rate, eyed-embryo success rate and hatching success rate, respectively.

[0139] Consequently, using a machine learning system having been trained to distinguish oocytes in Category 1, Category 2 and Category 3 from oocytes in Category 4 and Category 5 will allow the method predict the fertilisation rate, the eyed-embryo success rate and the hatching success rate for the analysed sample.

[0140] Utilising the findings of the above-discussed study, the at least one first class of oocytes images may comprise images displaying oocytes having a lipid droplet coalescence region with an area size less than a predetermined area size value, and the at least one second class of oocytes images may comprise images displaying oocytes having a lipid droplet coalescence region having an area size larger than the predetermined area size value.

[0141] For Salmo salar, the predetermined area size value may be any one of within the range of 1.3 mm2- 1.7 mm2; within the range of 1.4 mm2- 1.6 mm2; and 1.5 mm2, which predetermined area size defines the border between high-quality oocytes and low-quality ditto. For Salmonidae family species other than Salmo salar, other predetermined area size values may be used. Instead of using the lipid droplet coalescence region area size as a measure to distinguish classes, it may be advantageous to use the relative area size of the lipid droplet coalescence region, i.e. the area size of the lipid droplet coalescence region in relation to the total area size of the oocyte. This would more readily allow for the size of the oocytes varying, not only between different Salmonidae family species but also between individual females within the same species.

[0142] Consequently, as an alternative the at least one first class of oocytes images may comprise images displaying oocytes having a lipid droplet coalescence region covering less than a predetermined ratio of the total area of the depicted oocyte, and the at least one second class of oocytes images may comprise images displaying oocytes having a lipid droplet coalescence region covering more than the predetermined ratio of the total area of the depicted oocyte.

[0143] The predetermined ratio may be any one of: within the range of 11% - 14%; within the range of 12% - 13%; and 12%., which predetermined ratio defines the border between high-quality oocytes and low-quality ditto, as demonstrated by the above-discussed study.

[0144] Since the relative area size is independent of oocyte size, the above-identified predetermined ratio may be used for different Salmonidae family species.

[0145] Instead of utilising a trained machine learning system for classifying oocytes in the analysed sample, traditional image analysis may be used in the method.

[0146] According to this alternative, the method may comprise, for each oocyte in the sample, the step of measuring or evaluating the area size of the lipid droplet coalescence region in the captured first and second sets of images and the plurality of categories may comprise a first category for oocytes having a lipid droplet coalescence area size being less than a predetermined area size value and a second category of oocytes having a lipid droplet coalescence area size being larger than the predetermined area size value, wherein the predetermined area size value again being any one of: within the range of 1.3 mm2- 1.7 mm2; within the range of 1.4 mm2- 1.6 mm2; and 1.5 mm2.

[0147] The number of oocytes having an area size of the lipid droplet coalescence region being above and below the predetermined area size value, respectively, will represent a quality measure for the sample of oocytes, wherein the larger the number of oocytes in the sample falling within the category having a lipid droplet coalescence region area size below the predetermined area size value, the higher the rates of fertilisation, eyed-embryo success and hatching success, respectively, can be expected for the batch of oocyte the sample was drawn.

[0148] Again, however, it may be advantageous to use the relative area size of the lipid droplet coalescence region in the analysis instead of using the lipid droplet coalescence region area size. According to this alternative, the method may comprise, for each oocyte in the sample, the step of measuring or evaluating, in the captured first and second sets of images, the area size of the lipid droplet coalescence region and the area size of the oocyte, and the step of calculating the relative area size of the lipid droplet coalescence region by dividing the measured or evaluated area size of the lipid droplet coalescence region with the measured or evaluated area size of the oocyte.

[0149] The plurality of categories may then comprise a comprise a first category for oocytes having a lipid droplet coalescence relative area size being less than a predetermined relative area size value and a second category for oocytes having a lipid droplet coalescence relative area size being greater than the predetermined relative area size value, wherein the predetermined relative area size value may be any one of: within the range of 0.11 to 0.14; within the range of 0.12 to 0.13; and 0.12.

[0150] In the following, an embodiment of a fish oocyte analysis apparatus will be discussed with reference to Figs. 5-9, which apparatus is suitable for implementing the above-discussed methods.

[0151] The apparatus is configured to analyse oocytes of the species Salmo salar. However, it is understood that a corresponding apparatus, with suitable modifications, can be used for analysing oocytes from other Salmonidae family species.

[0152] Figs. 5-9 illustrate an embodiment of a fish oocyte analysis apparatus 200 which may be used to implement a method of providing a quality parameter for a batch of oocytes collected from a Salmonidae family female, e.g. any one of the above-discussed methods. Fig. 5 illustrates the apparatus 200 in an isometric view. The apparatus 200 comprises a cabinet or frame 201 and Figs. 6-9 show detailed part-views of the cabinet 201 with cover panels 202 and 204 removed for clarity.

[0153] The apparatus 200 comprises a receptacle 206 including an inlet 208 for receiving a sample of fish oocytes, e.g. collected from ovaries of fish females or stripped from living broodstock, and a liquid carrying the oocytes. The liquid carrying the oocytes may be e.g. seawater, or a water-based salt solution. In the present embodiment, the receptacle 206 has the form of a substantially rectangular open-top tray.

[0154] The apparatus 200 also comprises an imaging system 210 configured for capturing images of the oocytes (see Fig. 6) and a conveyor system 112 configured for transporting the oocytes from the receptacle 206 to the imaging system 210.

[0155] The apparatus 200 further comprises an analysis system 214 configured for analysing images captured by the imaging system 210. The analysis system 214 is preferably a computer implemented analysis system. In Fig. 5 the analysis system 214 is schematically illustrated inside the cabinet 201. However, in other embodiments of the apparatus the analysis system may be located separate from the cabinet 201, e.g. in a separate, external computer system. The analysis system 214 may comprise a central processing unit (CPU), volatile and / or nonvolatile memory and data input and output devices. In the disclosed embodiment, the apparatus comprises an output device 258 in the form of a computer screen (see Fig. 5) on which the result of the analysis may be presented. The apparatus 200 may also comprise an input device (not shown), e.g. in the form of a keyboard, which may be used to input commands or instructions to control the apparatus 200, including to the analysis system 214.

[0156] The imaging system 210 comprises a first imaging station 216 and a second imaging station 218. With reference to Fig. 7, which shows a detailed part- view of the apparatus 200 from a second viewing angle, the first imaging station 216 comprises a first camera 220 and an associated first lighting system 222. The second imaging station 218 comprises a second camera 224 and an associated second lighting system 226.

[0157] As previously stated, the apparatus 200 comprises a conveyor system 212 configured for transporting oocytes from the receptacle 206 to the imaging system 210. The conveyor system 212 comprises a rotatable disc 228 (see Fig. 7) having, along its periphery, a plurality of through-openings 230. The disc 228 is rotatably arranged about a substantially horizontal axis extending above and across the receptacle 206. The disc 228 has a diameter which is sufficient to allow the disc 228 to extend down into the receptacle 206 and allow oocytes fed to the receptacle to be captured by the openings 230 when the openings are immerged in the liquid carrying the oocytes.

[0158] The receptacle 206 may comprise one or a plurality of guide groves configured for guiding the oocytes towards and into the openings and a spillway passage allowing the liquid in which the oocytes are transported into the receptacle to be separated from the oocytes and conducted to an outlet and ejected. The disc 228 may be recessed in a side-wall of the receptacle 206.

[0159] The diameter of a ripe Salmo salar oocyte is approximately 4 mm. For oocytes of the species Salmo salar, experiments have shown that the openings 230 may have a diameter of approximately 6.5 mm and the disc may have a thickness of approximately 3 mm to allow effective capture of the oocytes. Generally, however, it is understood that the dimensions of the openings 230 should be adapted to the species of oocytes to be analysed.

[0160] With reference to Fig. 8, which shows a detailed part-view of the apparatus 200 with the disc 228 removed for clarity, the conveyor system 212 further comprises a first picking arrangement 232 which is configured to pick an oocyte from an opening 230 in the disc 228 and position the oocyte in front of the first camera 220. In the present embodiment, the first picking arrangement 232 comprises a substantially rectilinear first suction tube or straw 234 which extends substantially orthogonal to the optical axis 236 of the first camera 220 and also, in this embodiment, substantially parallel to the rotational axis of the disc 228, i.e. substantially orthogonal to the disc 228. The conveyor system 212 further comprises a second picking arrangement 238 which is configured to pick an oocyte from the first picking arrangement 232 and, in this embodiment, position the oocyte in front of the second camera 224. In the present embodiment, the second picking arrangement 238 comprises a substantially rectilinear second suction tube or straw 240 which extends substantially orthogonal to the optical axis 242 of the second camera 224 and also substantially orthogonal to the first suction tube 234, i.e. substantially parallel to the disc 228 in this embodiment.

[0161] The first picking arrangement 232 comprises a first motor and pneumatic arrangement configured to operate the first suction tube 234. In the present embodiment, the first motor and pneumatic arrangement is housed inside the apparatus 200 and, therefore, is not visible in the figures. The second picking arrangement 238 comprises a second motor and pneumatic arrangement 244 (see e.g. Fig. 7) configured to operate the second suction tube 240.

[0162] By means of the first motor and pneumatic arrangement, the first suction tube 234 is movable in its axial direction 246 and is also rotatable about this axis 246, as is illustrated in Fig. 8. Likewise, by means of the second motor and pneumatic arrangement 244, the second suction tube 240 is movable in its axial direction 248 and also rotatable about this axis 248, as is also illustrated in Fig. 8.

[0163] A chute 250 (see Fig. 8) is arranged to receive oocytes released by the second suction tube 240.

[0164] Operation of the apparatus 200 is typically preceded by an initial step of providing a sample of oocytes to be analysed. As previously discussed, the sample may for example be collected from a batch of oocytes stripped from a female broodstock and the analysis may be performed with the object analyse the quality of the batch.

[0165] The collected or stripped oocytes, suspended in the carrying liquid, are entered into the receptacle 206 through the inlet 208. If the apparatus is operated in a continuous mode, in which oocytes and carrying liquid are continuously feed to the receptacle 206 through the inlet 208, the carrying liquid in which the oocytes are transported into the receptacle 206 may be conducted to an outlet 252 (see Fig. 5) and ejected to maintain the carrying liquid in the receptacle 206 at a constant level, balancing the carrying liquid and oocytes entering through the inlet 208. If the apparatus 200 is operated in a batchwise mode, a batch of oocytes to be analysed may be entered into the receptacle 228 together with the carrying liquid without the carrying liquid having to be ejected through the outlet 252.

[0166] Once entered into the apparatus 200, the oocytes are sequentially and individually analysed.

[0167] With further reference Figs. 5-9, also to Fig. 10, a method of operating the apparatus 200 will now be discussed in more detail.

[0168] Operation of the apparatus 200 involves, for each oocyte, an initial step 302 of entering the oocyte into the conveyor system 212 and bringing the oocyte to the imaging system 210. This involves rotating the disc 228 and allowing the oocyte, at this point in time being suspended in the carrying liquid in the receptacle 206, to be captured by an opening 230 in the rotating disc 228 when the opening is immersed in the carrying liquid.

[0169] The rotating disc 228 raises the captured oocyte to the position of the first picking arrangement 232 and the first suction tube 234 is extended along axis 246 towards the disc 228 by the first motor and pneumatic arrangement (not visible). A distal end of the suction tube 234 is brought into contact with the oocyte which, at this point in time, is positioned in an opening 230, and a negative pressure established in the suction tube 234 attaches the oocyte to the distal end. The suction tube 234 is then retracted to position the distal end and the captured oocyte 100 in front of the first camera 220 such that the oocyte 100 becomes positioned on the optical axis 236 of the first camera 220 (see Figs. 8 and 9).

[0170] In a next step 304, the first camera 220 captures a first set of images of the oocyte from a plurality of directions. In the present embodiment, the images are captured through a protective glass sheet 254 (see Fig. 9). Between each image capture, the suction tube 234 is rotated a predetermined angle about its axis 246, thus allowing images of the oocyte 100 to be captured from a plurality of directions in a plane which is orthogonal to the axis of rotation 234 of the first suction tube 234. In other words, the first set of images represents images taken at a predetermined number of rotational positions about a first axis of rotation, i.e. the axis of rotation 246 of the first suction tube 234.

[0171] Preferably, the first set of images includes images of the oocyte 100 when rotated an entire revolution, i.e. throughout 360°. For example, the first set of images may comprise four images taken from four orthogonal directions, in which case the suction tube 234 and the attached oocyte 100 is rotated 90° between each image capture, thus resulting in the first set of images comprising four images taken at first rotational positions 0°, 90°, 180° and 270°.

[0172] However, the first set of images may involve a different configuration, e.g. one image captured every 180° (resulting in the first set of images totalling two images), one image captured every 120° (resulting in the first set of images totalling three images), one image captured every 72° (resulting in the first set of images totalling five images), etc.

[0173] In a preferred embodiment the attached oocyte 100 may be rotated 45° between each image capture. Consequently, if the attached oocyte is rotated an entire revolution, i.e. throughout 360°, the first set of images will comprise eight images taken at first rotational positions 0°, 45°, 90°, 135°, 180°, 225°, 270° and 315°.

[0174] The images of the first set of images are forwarded to the analysis system 214.

[0175] In a next step 306, the oocyte 100 is transferred from the first imaging station 216 to the second imaging station 218. In the present embodiment this is accomplished by the second motor and pneumatic arrangement 244 extending the second suction tube 240 towards the first second suction tube 234 until a distal end of the second suction tube 240 is brought into contact with the oocyte 100 which, at this point in time, is still attached to the distal end of the first suction tube 234. A negative pressure is then established in the second suction tube 240 attaching the oocyte 100 to the distal end of the second suction tube 240. The negative pressure in the first suction tube 234 is then released and the oocyte 100 is transferred to the second suction tube 240. The second suction tube 240 is then retracted to position the distal end of the second suction tube 240 and the captured oocyte 100 in front of the second camera 224 such that the oocyte 100 becomes positioned on the optical axis 242 of the second camera 224.

[0176] In a next step 308, the second camera 224 captures a second set of images of the oocyte 100. As in the first imaging station 216, the images are captured through a protective glass sheet 256 (see Fig. 9). Between capture of each image in the second set of images, the suction tube 240 is rotated a predetermined angle about its axis 248, thus allowing images of the oocyte 100 to be captured from a plurality of directions in a plane which is orthogonal to the axis of rotation of the second suction tube 240. Since the second suction tube 240 extends substantially orthogonally to the first suction tube 234, the plane in which the second set of images are captured is substantially orthogonal to the plane in which the first camera 220 captured the first set of images.

[0177] As the first set of images, the second set of images may advantageously include images of the oocyte 100 when rotated an entire revolution, i.e. throughout 360°. For example, as the first set of images, the second set of images may comprise four images taken from four orthogonal directions, in which case the suction tube 253b and the attached oocyte 100 is rotated 90° between each image capture. Again, however, the second set of images may involve a different configuration, e.g. one image captured every 180° (resulting in the second set of images totalling two images), one image captured every 120° (resulting in the second set of images totalling three images), one image captured every 72° (resulting in the second set of images totalling five images), etc.

[0178] In a preferred embodiment the attached oocyte 100 may be rotated 45° between each image capture also in the second imaging station 218. Consequently, if the attached oocyte is rotated an entire revolution, i.e. throughout 360°, the second set of images will comprise eighth images taken at second rotational positions 0°, 45°, 90°, 135°, 180°, 225°, 270° and 315°.

[0179] As the images of the first set of images, the images of the second set of images are forwarded to the analysis system 214.

[0180] The lighting systems 222 and 226 ensure that sufficient lighting of the oocyte is provided during image capture. In the disclosed embodiment the lighting systems 222 and 226 are arranged to provide lighting to the oocyte from a lighting axis forming an angle to the optical axis 236, 242 of approximately 45 degrees. However, in an alternative embodiment the lighting systems may be arranged opposite the oocyte as compared to the respective camera 220, 224, thus back-lighting the oocyte. In a next step 310, the analysis system 214, in a first sub-step 310a, identifies measures and / or evaluates at least one characteristic of lipid droplet coalescence in the images of the first and second set and, in a second sub-step 310b, categorises the oocyte as belonging to one of a plurality of predefined categories based on the at least one characteristic. In other words, the analysis system 214 is configured to attribute one of the predefined categories to the oocyte based on the at least one characteristic of the lipid droplets.

[0181] Steps 302-310 are repeated for all oocytes to be analysed.

[0182] In a next and final step 312, the analysis system 214 attributes a quality parameter to the sample of oocytes based on the distribution of the sampled oocytes in the predefined categories.

[0183] Since previously discussed image capturing steps 304 and 308 involve capturing images of the oocyte from a plurality of rotational angels when the oocyte is rotated about two orthogonal axes, i.e. axes 246 and 248 in the present embodiment (see Fig. 8), the images forwarded to the analysis system 214 from the first and second imaging stations 216, 218 will provide a comprehensive picture and overview of the distribution of the lipid droplets in the oocyte, thus facilitating sub-steps 310a and 310b.

[0184] As previously discussed, the analysis system 214 may advantageously utilise a trained image classification system for identifying, measuring and / or evaluating the at least one characteristic of the lipid droplet coalescence (in sub-step 310a) and classifying each oocyte according to one of the predefined categories (in sub-step 310b). Known machine learning techniques and artificial intelligence software, e.g. based on TENSORFLOW, may be used to train and verify such machine learning system.

[0185] As previously discussed, identifying, measuring and / or evaluating the at least one characteristic of the lipid droplet coalescence may alternatively comprise measuring and evaluating the area size of the lipid droplet coalescence region. Optionally, as previously discussed, this can also comprise measuring or evaluating the area size of the oocyte.

[0186] As previously discussed, the analysis system 214 is preferably a computer implemented analysis system comprising a central processing unit (CPU), volatile and / or non-volatile memory and data input and output devices.

[0187] It is appreciated that certain features of the invention, which, for clarity, have been described above in the context of separate embodiments, may also be provided in combination in a single embodiment. Conversely, various features of the invention, which, for brevity, have been described in the context of a single embodiment, may also be provided separately or in any suitable sub-combination. In particular, it will be appreciated that features described in relation to one particular embodiment may be interchangeable with features described in relation to the other embodiments. In the preceding description, various aspects of the apparatus according to the invention have been described with reference to the illustrative embodiment. For purposes of explanation, specific numbers, systems and configurations were set forth in order to provide a thorough understanding of the apparatus and its workings. However, this description is not intended to be construed in a limiting sense. Various modifications and variations of the illustrative embodiment, as well as other embodiments of the apparatus, which are apparent to person skilled in the art to which the disclosed subject-matter pertains, may lie within the scope of the present invention as defined by the following claims.

Claims

Claims1. A method of analysing a sample of a plurality of fish oocytes, comprising:- for each oocyte in the sample:- capturing a first set of images of the oocyte (100) from a plurality of rotational angels when rotated about a first axis of rotation (246);- capturing a second set of images of the oocyte (100) from a plurality of rotational angels when rotated about a second axis of rotation (248), which second axis of rotation (248) is orthogonal to the first axis of rotation (246);- identifying at least one characteristic of lipid droplet coalescence in the captured first and second sets of images; and- based on the identified at least one characteristic of lipid droplet coalescence, classifying the oocyte as belonging to one of a plurality of predefined categories, and- based on the distribution of the sampled oocytes in the plurality of categories, attributing a quality parameter to the sample of oocytes, the quality parameter being any one of: a predicted fertilisation rate of the sampled oocytes; a predicted eyed- embryo success rate of the sampled oocytes; and a predicted hatching success rate of the sampled oocytes.

2. The method according to claim 1, wherein said step of classifying the oocyte as belonging to one of the plurality of predefined categories comprises supplying the first and second sets of images to a machine learning system having been trained on a dataset of oocyte images comprising:- at least one first class (C1-C3) of oocytes images depicting oocytes being associated with at least one of: a first fertilisation rate; a first eyed-embryo success rate; and a first hatching success rate; and- at least one second class (C4-C5) of oocytes images depicting oocytes being associated with at least one of: a second fertilisation rate; a second eyed-embryo success rate; and a second hatching success rate, wherein at least one of said first fertilisation rate, said first eyed-embryo success rate, and said first hatching success rate is higher than said second fertilisation rate, said second eyed-embryo success rate, and said second hatching success rate, respectively.

3. The method according to claim 2, wherein said at least one first class (C1-C3) of oocytes images comprises images displaying oocytes having a lipid droplet coalescence region (104) covering less than a predetermined ratio of the total area of the depicted oocyte,and wherein said at least one second class (C4-C5) of oocytes images comprises images displaying oocytes having a lipid droplet coalescence region covering more than the predetermined ratio of the total area of the depicted oocyte.

4. The method according to claim 3, wherein said predetermined ratio is any one of: within the range of 11% - 14%; within the range of 12% - 13%; and 12%.

5. The method according to claim 2, wherein said at least one first class (C1-C3) of oocytes images comprises images displaying oocytes having a lipid droplet coalescence region (104) having an area size less than a predetermined area size value, and wherein said at least one second class (C4-C5) of oocytes images comprises images displaying oocytes having a lipid droplet coalescence region (104) having an area size larger than the predetermined area size value.

6. The method according to claim 5, wherein said predetermined area size value is any one of: within the range of 1.3 mm2- 1.7 mm2; within the range of1.4 mm2- 1.6 mm2; and 1.5 mm2.

7. The method according to claim 1, wherein the at least one characteristic of lipid droplet coalescence comprises the area size of a lipid droplet coalescence region (104) in the oocyte, and wherein the method comprises, for each oocyte in the sample, measuring or evaluating the area size of the lipid droplet coalescence region in the captured first and second sets of images.

8. The method according to claim 7, wherein said plurality of categories comprise a first category for oocytes having a lipid droplet coalescence area size being less than a predetermined area size value and a second category of oocytes having a lipid droplet coalescence area size being larger than the predetermined area size value.

9. The method according to claim 8, wherein said predetermined area size value is any one of: within the range of 1.3 mm2- 1.7 mm2; within the range of 1.4 mm2- 1.6 mm2; and1.5 mm2.

10. The method according to claim 1, wherein the at least one characteristic of lipid droplet coalescence comprises a relative area size of a lipid droplet coalescence region (104) in the oocyte, and wherein the method comprises, for each oocyte in the sample:- measuring or evaluating the area size of the lipid droplet coalescence region and the area size of the oocyte in the captured first and second sets of images; and- calculating the relative area size of the lipid droplet coalescence region (104) by dividing the measured or evaluated area size of the lipid droplet coalescence region with the measured or evaluated area size of the oocyte.

11. The method according to claim 10, wherein said plurality of categories comprise a first category for oocytes having a lipid droplet coalescence relative area size being less than a predetermined relative area size value and a second category for oocytes having a lipid droplet coalescence relative area size being greater than the predetermined relative area size value.

12. The method according to claim 11, wherein said predetermined relative area size value is any one of: within the range of 0.11 to 0.14; within the range of 0.12 to 0.13; and 0.12.

13. A computer-implemented method of attributing a quality measure to a sample of fish oocytes, comprising the steps of:- for each oocyte in the sample:- obtaining at least one image, taken by an imaging device, of an oocyte of the sample depicting lipid droplets in the oocyte; and- inputting the obtained at least one image to a trained learning model and classifying the oocyte as belonging to one of a plurality of predefined categories based on coalescence of the lipid droplets;- based on the distribution of the sampled oocytes in the plurality of categories, attributing the quality parameter to the sample of oocytes, the quality parameter being any one of: a predicted fertilisation rate of the sampled oocytes; a predicted eyed-embryo success rate of the sampled oocytes; and a predicted hatching success rate of the sampled oocytes.

14. A computer program having instructions which when executed by a computing device or system (214) causes the computing device to perform the method according to claim 13.

15. A fish oocyte analysis apparatus (200) comprising a processor and a memory (214) and being configured to perform the method according to claim 13.

16. A computer-implemented method for training a neural network for use in a system for analysing a sample of fish oocytes, the method comprising the steps of:- inputting training data to the neural network to train the neural network through machine learning; wherein the training data comprises:- data representing coalescence of lipid droplets in fish oocytes; and- a scoring data on any one of; an oocyte fertilisation rate of; an oocyte eyed-embryo success rate; and an oocyte hatching success rate.

17. The method according to any one of claims 1-13 and 16, wherein said fish oocytes are any one of: Salmonidae oocytes; and oocytes of the species Salmo salar.

Images