Method and apparatus for analysing fish eggs

Abstract

A method of analysing a fertilised fish egg (100) having incubated to a spine stage is disclosed. The method comprises: capturing a first set of images of the fish egg when the fish egg is rotated about a first axis of rotation (225a); capturing a second set of images of the fish egg when the fish egg is rotated about a second axis of rotation (225b) which is orthogonal to the first axis of rotation (225a); for each image in the first and second set of images, identifying the vertebral column depicted in the image and measuring or evaluating at least one characteristic of the depicted vertebral column; and based on the at least one measured or evaluated characteristic, classifying the fish egg as viable or non-viable. A related apparatus is also disclosed.

Description

[0001] METHOD AND APPARATUS FOR ANALYSING FISH EGGS

[0002] The present disclosure relates to a method of analysing fish eggs, especially with the purpose of determining whether incubating fish egg are viable. The present disclosure also relates to an associated apparatus.

[0003] The present disclosure also relates to a method and to an associated apparatus for determining the ratio of viable fish eggs in a batch of incubating fish eggs.

[0004] In aquaculture fish farming, spermatozoa and eggs are mixed to allow the eggs to be impregnated and fertilized. In order to determine the success rate of the fertilization process, a sample of eggs may be drawn from the fertilized batch and analysed to determine the ratio between the number of fertilized eggs and the number of unfertilized ones, e.g. the fertilization rate.

[0005] WO 2021 / 228993 Al discloses a method of identifying fertilized fish eggs in a sample of eggs taken from a batch of incubating fish eggs. The method includes the steps of treating the sample with a fixative bringing an outer membrane of the eggs at least partly transparent and / or a blastodisc of the eggs at least partly opaque. The method further includes the step of capturing at least one image of individual eggs of the sample depicting the blastodisc of the egg, analysing the at least one image for detecting divided cells in the blastodisc, and identifying each egg associated with an image showing divided cells in the blastodisc as a fertilized egg.

[0006] Although the system and apparatus discussed in WO 2021 / 228993 Al functions well, there is a need not only to analyse if eggs are fertilised but also to estimate how viable they are.

[0007] It is known to manually study fertilised fish eggs having reached the spine stage, i.e. at a stage where the vertebral column, or spine, has emerged in the egg to discern whether the eggs appear to be viable. Figs, la-le schematically illustrate fertilized eggs from a teleost fish species at the spine stage has emerged in the egg. The period of time necessary for the vertebral column to develop differs between fish species. Also, the period of time necessary for the vertebral column to develop will depend of the temperature of the environment in which the eggs incubate. For Atlantic salmon (Salmo salar), the spine stage typically occurs at approximately 120 to 150 degree-days post fertilisation. As is known in the art of fish aquaculture, degree-days (DD) or cumulative degree-days (CDD; °C days) is a measure which takes into account the temperature in the incubator when timing fish egg development. In particular, degree-days divided by the temperature in the incubator in centigrade will give a time measure in days. For example, if the temperature in the incubator is 10°C, the spine stage for S. salar can be expected at 120 to 150 degree-days / 10°C = 12 to 15 days post fertilisation.

[0008] For other fish species the incubation time necessary for the vertebral column to develop may be different. For example, for brown trout (Salmo trutta) the spine stage may typically occur at 90 to 100 degree-days, rainbow trout (Oricorhyrichus mykiss) approximately 77 degree- days, Chinook salmon (Oncorhynchus tshawytscha) approximately 135 degree-days and brook trout (Salvelinus fontinalis) approximately 88 degree-days. Atlantic halibut (Hippoglossus hippoglossus) will typically reach the spine stage at approximately 50 degree-days.

[0009] Fig. la illustrates a fish egg 100a having a straight, normal vertebral column 102a. Such an egg can be expected to develop to a healthy hatchling. Figs. Ib-le, by contrast, illustrate eggs lOOb-lOOe having abnormal vertebral columns 102b-102e. In particular, Figs. Ib-le schematically illustrates a Y or V shaped vertebral column (Fig. lb), a corkscrew or curled vertebral column (Fig. 1c), a crooked vertebral column (Fig. d), and a stubby or short vertebral column (Fig. le). Such eggs are likely to die prior to hatching or, if they hatch, to produce hatchlings with an arrested development. Consequently, Fig. la typically represents a viable egg, whereas Figs. Ib-le typically represent non-viable eggs.

[0010] In this disclosure, unless otherwise stated, “viable” means more likely than not to develop to a healthy hatchling and “non-viable means less likely than not to develop to a healthy hatchling.

[0011] Analysing spine stage fish eggs manually to determine viability is time-consuming and the result can vary depending on the operator's subjective assessment of what characterizes a viable and non-viable egg, respectively. Therefore, there is a need for an automated method and system that reliably can distinguish between viable and non-viable fish eggs.

[0012] One problem when analysing spine stage fish eggs is that the orientation of the egg may influence how the spine is perceived. For example, a fish egg having a Y or V shaped vertebral column (see Fig. lb) may appear to have a straight vertebral column if the egg is viewed in the plane spanned by the Y or V shaped vertebral column and, therefore, may be mistaken for a viable fish egg. During manual analysis this is rarely a problem since the operator performing the analysis can usually manipulate the egg and determine and pick the viewing direction that is the most suitable to determine the features of the vertebral column.

[0013] WO 2019 / 001604 Al discloses an apparatus and a method for examining fish eggs, in particular the development of fish eggs in fish farms. According to the disclosed method, a sample of fish eggs is taken, placed in a sample holder and examined using optical coherence tomography (OCT). The measuring head of the optical coherence tomography system is positioned over each fish egg and in the measurement area data from the fish blastodisc and / or embryonic structure of the fish eggs is taken. The data is evaluated and from the sample the developmental state of each examined fish egg is determined.

[0014] However, the apparatus and method is primarily designed to study the blastodisc of the eggs and is less suitable for studying eggs having reached the spine stage. Also, since the apparatus and method discussed in WO 2019 / 001604 Al relies on OCT, the apparatus and method are relatively slow and complicated. One object of the present disclosure is to address this problem and provide a useful method and an associated apparatus and system for analysing fish eggs, especially with the purpose of determining the viability of fish eggs incubating in an incubator.

[0015] Another object is to provide a method and an apparatus and system enabling fast and efficient analysis of early development of the vertebral column in fish eggs, in particular with the aim of allowing efficient detection of abnormal vertebral columns in fish eggs.

[0016] Yet a further object is to provide a method and an apparatus and system for establishing the ratio of viable fish eggs in a batch of incubating fish eggs.

[0017] Yet another object of the present disclosure is to address the shortcomings of prior art apparatuses and methods and provide an alternative approach to analysing fish eggs, especially with the aim of determining the ratio of viable fish eggs in a batch of incubating fish eggs.

[0018] Yet a further object of the present disclosure is to provide a method and an associated apparatus which is suited for analysing fish eggs from salmonoid species, such as Atlantic salmon, brown trout and Chinook salmon.

[0019] According to a first aspect, the present disclosure provides a method of analysing a fertilised fish egg having incubated to a spine stage. The method comprises the steps of:

[0020] (a) capturing a first set of images of the fish egg when the fish egg is rotated about a first axis of rotation;

[0021] (b) capturing a second set of images of the fish egg when the fish egg is rotated about a second axis of rotation which is orthogonal to the first axis of rotation;

[0022] (c) for each image in the first and second set of images, identifying the vertebral column depicted in the image and measuring or evaluating at least one characteristic of the depicted vertebral column; and

[0023] (d) based on the at least one measured or evaluated characteristic, classifying the fish egg as viable or non-viable.

[0024] The at least one characteristic may comprise at least one of: a curvature of the depicted vertebral column, a length of the depicted vertebral column and an aspect ratio of the depicted vertebral column.

[0025] Step (c) may involve employing an image analysis system, e.g. a computer implemented image analysis system, for identifying the vertebral column depicted in respective image and for measuring or evaluating the least one characteristic of the depicted vertebral column.

[0026] An image analysis system may be involved also in step (d). For example, step (d) may involve employing computer vision and algorithms implemented in the computer implemented image analysis system and built by a machine learning system, e.g. the TensorFlow (TM) toolkit, utilising mathematical models based on sample or training images of fish eggs known to be viable or non-viable.

[0027] In particular, step (d) may preferably involve using a machine learning system having been trained on a training set of images depicting fish eggs having been classified as viable or non-viable. In addition to a class comprising normal or viable fish eggs, i.e. fish eggs with a normal or viable vertebral column, the training set may contain a plurality of classes of abnormal or non-viable fish eggs, e.g. any combination of: a class comprising eggs with a Y or V shaped vertebral column; a class comprising eggs with a corkscrew or curled vertebral column; a class comprising eggs with a crooked vertebral column; and a class comprising eggs with a stubby or short vertebral column. In this latter case, step (h) may involve not only classifying the fish as non-viable but also classifying the non-viable fish egg according to type of spinal deformation.

[0028] The machine learning system may comprise a trained neural network model having been trained on a training set of images depicting fish eggs having been classified as viable or non-viable.

[0029] Step (d) may preferably involve using a machine learning system having been trained on a training set of images depicting fish eggs of the same species as being analysed according to the method.

[0030] Depending on how the vertebral column is depicted in each image of the first and second set of images, the machine learning system may possibly classify one individual image as viable and another image as non-viable. For example, a curled vertebral column (e.g. see Fig. 1c) could, from one viewing direction, appear rectilinear and, therefore, could erroneously be classified as viable. However, taking into account all of the images in the first and second set of images from the first and second sets of images, a correct classification of the egg is possible.

[0031] According to one alternative embodiment, the method may comprise the steps of: identifying one of the images of the first and second set of images as a representative image; and classifying the identified representative image as being associated with one of a viable egg and a non-viable egg.

[0032] For example, a first computer implemented image analysis system, e.g. comprising a first machine learning system, may be used to identify one of the images of the first and second set of images as being a representative image, and a second computer implemented image analysis system, e.g. comprising a second machine learning system, may be used to classify the identified representative image as either viable or non-viable.

[0033] The step of identifying one of the images of the first and second sets of images as a representative image may comprise measuring or evaluating the rectilinearity of the vertebral column of the first and second sets of images and choosing the image displaying the least rectilinear vertebral column as the representative image. The image displaying the least rectilinear vertebral column is the image most likely to reveal a spinal defect if present in the egg and, therefore, could be considered to be most suitable candidate for the subsequent classification step. If this “worst case images” is classified as viable, the egg is also classified as viable.

[0034] At least one of the first and second machine learning system may comprise a trained neural network model.

[0035] The above-discussed step (a) may involve capturing the first set of images when the egg is rotated in a first rotational plane and step (b) may involve capturing the second set of images when the egg is rotated in a second rotational plane which is orthogonal to the first plane.

[0036] Capturing the first set of images may comprise capturing the first set of images in a first imaging station comprising a first camera having an optical axis located in the first rotational plane, and capturing the second set of images may comprise capturing the second set of images in a second imaging station having a second camera having an optical axis located in said second rotational plane. The optical axes of the first and second cameras may be parallel.

[0037] Alternatively, the first set of images and the second set of images may be captured using the same camera.

[0038] Capturing any one of the first set of images and the second set of images may comprise capturing images of the egg when it is rotated an entire revolution about respective axis of rotation. Advantageously, capturing any one of the first set of images and the second set of images may comprise rotating said individual egg between image captures within the range of 45° to 135°.

[0039] The method is suited for analysing fish eggs from salmonoid species but can also be used for analysing the eggs of other fish species.

[0040] According to a second aspect, the present disclosure provides a method of determining a viability rate of a batch of fertilised fish eggs having incubated to spine stage, comprising the steps of:

[0041] - drawing a sample comprising a plurality of eggs from the batch of incubating fish eggs;

[0042] - performing the steps (a)-(d) according to the first aspect on each egg of the drawn sample; and

[0043] - calculating the viability rate as a ratio between the number of eggs classified as being viable and the total number of eggs in the sample.

[0044] According to a third aspect, the present disclosure provides a method comprising the step of allowing a batch of fertilised fish eggs to incubate until the eggs reach a spine stage, i.e. until features of the vertebral column becomes visible and or identifiable in the eggs. For Atlantic salmon (Salmo salar) this may involve letting the fertilised batch incubate until approximately 120 to 150 degree-days. It is understood, however, that the incubation time for other fish species may differ. The method further comprises the steps of drawing a representative sample of fish eggs from the incubating batch fish eggs when the eggs have reached said spine stage and analysing each egg in the drawn sample according to the steps (a)-(h) of the first aspect. The method may comprise performing the steps (a)-(h) sequentially for each egg in the drawn sample, e.g. by bringing the sample eggs in single file passed an imaging system, i.e. guiding the eggs one-by-one through the imaging system.

[0045] The method may comprise the step of treating the drawn sample of fish eggs with a fixative bringing the outer membrane of the eggs at least partly transparent and / or the vertebral column of the eggs more visible. Such fixatives are known in the art and may include acetic acid-alcohol-formaldehyde solutions, such as Serra's fixative. Other known fixtures are Stockard's solution and AquaBoost® Quattro. However, any fixative capable of bringing the eggs at least partly transparent and / or the vertebral column visible can be used. The step of treating the egg with a fixative is preferably effectuated prior to steps (a)-(h).

[0046] According to a fourth aspect, the present disclosure provides a fish egg analysis apparatus comprising:

[0047] - an imaging system configured for:

[0048] - capturing a first set of images of a fish egg when the fish egg is rotated about a first axis of rotation; and

[0049] - capturing a second set of images of the fish egg when the fish egg is rotated about a second axis of rotation which is orthogonal to the first axis of rotation;

[0050] - an image analysis system configured for:

[0051] - identifying, in each image in the first and second set of images, a vertebral column depicted in the image;

[0052] - measuring or evaluating at least one characteristic of the depicted vertebral column; and

[0053] - based on the at least one measured or evaluated characteristic, classifying the fish egg as viable or non-viable.

[0054] The apparatus may comprise a receptacle for receiving a sample of fish eggs, e.g. taken from a batch of incubating fish eggs, and a conveyor system configured fortransporting eggs from the receptacle to the imaging system. The conveyor system may comprise a rotatable disc comprising, along its periphery, a plurality of openings for receiving the eggs.

[0055] The conveyor system may comprise a transfer device which is configured to transfer eggs from the disc to the imaging system. The transfer device may comprise a picking element which is configured to pick the eggs from the openings and position them in the imaging system. The picking element may comprise a suction tube having a distal end which is configured, by means of a negative pressure in the suction tube, to pick an egg from an opening of the disc.

[0056] 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.

[0057] 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.

[0058] Following drawings are appended to facilitate the understanding of the disclosure and the invention as it is defined in the claims:

[0059] Figs, la-le are schematic illustrations of fertilized eggs from a teleost fish species having developed to the spine stage;

[0060] Fig. 2 illustrates a fish egg analysis apparatus according one embodiment of the disclosure;

[0061] Fig. 3 shows a detailed part-view of the apparatus according to Fig. 2 with covers removed for clarity;

[0062] Fig. 4 shows a detailed part-view of the apparatus according to Fig. 2 from an alternative angle;

[0063] Fig. 5 shows a detailed part-view of the apparatus according to Fig. 2 with a disc of a conveyor system of the apparatus removed for clarity;

[0064] Fig. 6 shows a further detailed part-view of the apparatus according to Fig. 2;

[0065] Fig. 7 is a flow chart schematically illustrating a method according to one embodiment of the disclosure.

[0066] 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.

[0067] In the following, one or more specific embodiments of the disclosure will be described in more detail with reference to the drawings. However, it is specifically intended that the disclosure 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.

[0068] Figs. 2-6 illustrate an embodiment of a fish egg analysis apparatus 200 which may be used in a system for determining the viability rate of a batch of incubating fish eggs. Fig. 2 shows the apparatus 200 in an isometric view and Fig. 3 shows a detailed part-view of the apparatus 200 with cover panels 202 and 204 removed for clarity.

[0069] The apparatus 200 comprises a receptacle 228 including an inlet 229 for receiving a sample of fish eggs, e.g. taken from a batch of incubating fish eggs, and a liquid carrying the fish eggs. The liquid carrying the eggs may be a liquid in which the fish eggs survive, e.g. fresh or sea water. In the present embodiment, the receptacle 228 has the form of a substantially rectangular open-top tray.

[0070] The apparatus 200 also comprises an imaging system 208 configured for capturing images of the eggs (see Fig. 3) and a conveyor system 214 configured for transporting the eggs from the receptacle 228 to the imaging system 208.

[0071] The apparatus also comprises an image analysis system 216 configured for analysing images captured by the imaging system 208.

[0072] The imaging system 208 comprises a first imaging station 212a and a second imaging station 212b. With reference to Fig. 4, which shows a detailed part- view of the apparatus 200 from a second viewing angle, the first imaging station 212a comprises a first camera 222a and an associated first lighting system 224a. The second imaging station 212b comprises a second camera 222b and an associated second lighting system 224b.

[0073] As previously stated, the apparatus 200 comprises a conveyor system 214 configured to transport eggs from the receptacle 228 to the imaging system 208. The conveyor system 214 comprises a rotatable disc 230 (see Fig. 4) having, along its periphery, a plurality of through- openings 232. The disc 230 is rotatably arranged about a substantially horizontal axis extending above and across the receptacle 228. The disc 230 has a diameter which is sufficient to allow the disc 230 to extend down into the receptacle 228 and allow eggs fed to the receptacle to be captured by the openings 232 when the openings are immerged in the liquid carrying the eggs.

[0074] The receptacle 228 may comprise one or a plurality of guide groves configured for guiding the eggs towards and into the openings and a spillway passage allowing the liquid in which the eggs are transported into the receptacle to be separated from the eggs and conducted to an outlet and ejected. The disc 230 may be recessed in a side-wall of the receptacle 228.

[0075] For eggs of the species Salmo salar, experiments have shown that the openings 232 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 eggs. Generally, however, it is understood that the dimensions of the openings 732 should be adapted to the species of eggs to be analysed.

[0076] With reference to Fig. 5, which shows a detailed part-view of the apparatus 200 with the disc 230 removed for clarity, the conveyor system 214 further comprises a first picking element 252a which is configured to pick an egg from an opening 232 in the disc 230 and position the egg in front of the first camera 222a in the first imaging station 212a. In the present embodiment, the first picking element 252a comprises a substantially rectilinear first suction tube or straw 253a which extends substantially orthogonal to the optical axis 223a of the first camera 222a and also, in this embodiment, substantially parallel to the rotational axis of the disc 230, i.e. substantially orthogonal to the disc 230.

[0077] The conveyor system 214 further comprises a second picking element 252b which is configured to pick an egg from the first picking element 252a and position the egg in front of the second camera 222b in the second imaging station 212b. In the present embodiment, the second picking element 252b comprises a substantially rectilinear second suction tube or straw 253b which extends substantially orthogonal to the optical axis 223b of the second camera 222b and also substantially orthogonal to the first suction tube 252a, i.e. substantially parallel to the disc 230 in this embodiment.

[0078] The first picking element 252a is operable by means of a first motor and pneumatic arrangement. 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 element 252b is actuatable by means of a second motor and pneumatic arrangement 254b (see e.g. Fig. 4).

[0079] By means of the first motor and pneumatic arrangement, the first suction tube 253a is movable in its axial direction 225a and is also rotatable about this axis 225a, as is illustrated in Fig. 5. Likewise, by means of the second motor and pneumatic arrangement 254b, the second suction tube 253b is movable in its axial direction 225b and also rotatable about this axis 225b, as is also illustrated in Fig. 5.

[0080] A chute 244 (see Fig. 5) is arranged to receive eggs released by the second suction tube 253b.

[0081] With further reference to Fig. 7, an embodiment of operating the apparatus 200 will now be discussed in more detail.

[0082] Operation of the apparatus 200 and the system in which it is arranged may involve an initial step 702 of allowing a batch of fertilised fish eggs to incubate to the spine stage. As previously discussed, this stage may typical be reached at 120 day-degrees for Salmo salar, although for other fish species the incubation time necessary for the vertebral column to develop may be different.

[0083] A subsequent step 704 involves drawing a sample comprising a plurality of eggs from the batch of fish eggs having incubated to the spine stage. In a subsequent step 706, which is optional, the drawn sample may be treated with a fixative bringing the outer membrane of the eggs of the sample at least partly transparent and / or the vertebral column of the eggs more visible. Such fixatives are known in the art and may include acetic acid-alcohol-formaldehyde solutions, such as Serra's fixative. Other known fixtures are Stockard's solution and AquaBoost® Quattro. However, any fixative capable of bringing the eggs transparent and / or the vertebral column visible can be used.

[0084] In a next step 708, the sampled eggs and a carrying liquid, e.g. water, are entered into the receptacle 228 through the inlet 229 (see Fig. 2).

[0085] In a next step 710, the eggs are entered into the conveyor system 214 (see Fig. 3) to be brought through the imaging system 208. In this step the eggs are captured by the openings 232 in the rotating disc 230 when the openings are immersed in the carrying fluid. The carrying fluid in which the eggs are transported into the receptacle 228 is conducted to an outlet 237 (see Fig. 2) and ejected.

[0086] The rotating disc 230 raises the captured eggs to the position of the first picking element 252a and the first suction tube 253a is extended along axis 225a towards the disc 230 by the first motor and pneumatic arrangement (not visible). A distal end of the suction tube 253a is brought into contact with an egg positioned in an opening 232 and a negative pressure in the suction tube 253a attaches the egg to the distal end. The suction tube 253a is then retracted to position the distal end and the captured egg 100 in front of the first camera 222a such that the egg 100 becomes positioned on the optical axis 223a of the first camera 222a.

[0087] In a next step 712, the first camera 222a captures a first set of images of the egg. In the present embodiment, the images are captured through a protective glass sheet 226a (see Fig. 6). Between the capture of each image in the first set of images, the suction tube 253a is rotated a predetermined angle about its axis 225a, thus allowing images of the egg 100 to be captured from a plurality of directions in a plane which is orthogonal to axis of rotation 225a of the first suction tube 253a. In other words, the first set of images represents images taken at a predetermined number of rotational positions about the first axis of rotation 225a.

[0088] Advantageously, the first set of images includes images of the egg 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 253a and the attached egg 100 is rotated 90° between each image capture, thus resulting in the first set of images comprising images taken at first rotational positions 0°, 90°, 180° and 270°.

[0089] However, the first set of images may involve a different configuration, e.g. one image captured every 180° (totalling 2 images in the set), one image captured every 120° (totalling 3 images in the set), one image captured every 72° (totalling 5 images in the set), etc.

[0090] In a preferred embodiment the attached egg 100 may be rotated 45° between each image capture. Consequently, if the attached egg 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°.

[0091] The images of the first set of images are forwarded to an image analysis system 216, either individually one by one as they are captured or as a complete set in a single transfer after all the images of the set have been captured.

[0092] In step 714, the first set of images is analysed by the image analysis system. This analysis may comprise, for each image in the first set of images, a first sub-step 714a of identifying the vertebral column depicted in the image and a subsequent, second sub-step 714b of measuring or evaluating one or a plurality of characteristics of the depicted vertebral column, e.g. the curvature, the length and / or the aspect ratio of the depicted vertebral column. The image analysis system 216 may advantageously employ a computer implemented image analysis system for identifying the vertebral column depicted in respective image and for measuring or evaluating the characteristic(s) of the depicted vertebral column.

[0093] In a next step 716, the egg 100 is transferred from the first imaging station 224a to the second imaging station 224b. This is accomplished by the second motor and pneumatic arrangement 254b extending the second suction tube 253b towards the first second suction tube 253a until a distal end of the second suction tube 253b is brought into contact with the egg 100 which, at this point in time, is still attached to the distal end of the first suction tube 253a. A negative pressure is then established in the second suction tube 253b attaching the egg 100 to the distal end of the second suction tube 253b. The negative pressure in the first suction tube 253a is then released and the egg 100 is transferred to the second suction tube 253b. The second suction tube 253b is then retracted to position the distal end of the second suction tube 253b and the captured egg 100 in front of the second camera 222b such that the egg 100 becomes positioned on the optical axis 223b of the second camera 222b.

[0094] In a next step 718, the second camera 222b captures a second set of images of the egg 100. In the present embodiment, the images are captured through a protective glass sheet 226b (see Fig. 6). Between capture of each image in the second set of images, the suction tube 253b is rotated a predetermined angle about its axis, thus allowing images of the egg 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 253b. Since the second suction tube 253b extends substantially orthogonally to the first suction tube 252a, the plane in which the second set of images are captured is substantially orthogonal to the plane in which the first camera 222a captured the first set of images.

[0095] As the first set of images, the second set of images may advantageously include images of the egg 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 egg 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° (totalling 2 images in the set), one image captured every 120° (totalling 3 images in the set), one image captured every 72 (totalling 5 images in the set), etc.

[0096] In a preferred embodiment the attached egg 100 may be rotated 45° between each image capture also in the second imaging station 212b. Consequently, if the attached egg is rotated an entire revolution, i.e. throughout 360°, the second set of images will comprise eighth images taken at first rotational positions 0°, 45°, 90°, 135°, 180°, 225°, 270° and 315°.

[0097] The images of the second set of images are forwarded to the image analysis system 216, either individually one by one as they are captured or as a complete set in a single transfer after all the images of the set have been captured.

[0098] In a next step 720, the second set of images is analysed by the image analysis system in a similar manner as the first set of images. This analysis may comprise, for each image in the second set of images, a first sub-step 720a of identifying the vertebral column depicted in the image and a subsequent, second sub-step 720b of measuring or evaluating one or a plurality of characteristics of the depicted vertebral column, e.g. the curvature, the length and / or the aspect ratio of the depicted vertebral column. As in step 714, the image analysis system 216 may advantageously employ a computer implemented image analysis system for identifying the vertebral column depicted in respective image and for measuring or evaluating the characteristic(s) of the depicted vertebral column.

[0099] In a subsequent step, 722, the characteristic(s) of the depicted vertebral column of the images of the first and second images are analysed in the image analysis system 216 to classify the egg as viable or non-viable. Because the two sets of images are captured from two substantially orthogonal planes and because the egg is rotated between images captures, the set of images will yield a sufficiently clear picture of the characteristics of the vertebral column to allow the image analysis system to classify the egg as either having a normal vertebral column, i.e. being viable, or an abnormal vertebral column, i.e. being non-viable.

[0100] In order to clearly discern the characteristics of the vertebral column it has been found that each set of images, i.e. each of the first set of imagens and the second set of images, should preferably contain at least two images captured at two different rotational angles about respective axis of rotation 225a, 225b. Preferably, however, each set of images should contain at least four images, and more preferably at least six images, e.g. eight images, captured at even intervals when rotated about respective axis of rotation.

[0101] To classify the egg, the image analysis system may advantageously employ computer vision and algorithms built by machine learning systems utilising mathematical models based on sample or training images of eggs known to be either viable or non-viable.

[0102] The image analysis system may comprise a central processing unit (CPU), volatile and / or non-volatile 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. 2) 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.

[0103] In said image capturing steps 712 and 718, lighting systems 224a and 224b (see Fig. 4) ensure that sufficient lighting is provided.

[0104] 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.

[0105] 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 fertilised fish egg (100) having incubated to a spine stage, characterised by the steps of:(a) capturing a first set of images of the fish egg (100) when the fish egg (100) is rotated about a first axis of rotation (225a);(b) capturing a second set of images of the fish egg (100) when the fish egg (100) is rotated about a second axis of rotation (225b) which is orthogonal to the first axis of rotation (225a);(c) for each image in the first and second set of images, identifying the vertebral column (202a-e) depicted in the image and measuring or evaluating at least one characteristic of the depicted vertebral column; and(d) based on the at least one measured or evaluated characteristic, classifying the fish egg as viable or non-viable.

2. The method according to claim 1, wherein said at least one characteristic comprises at least one of: a curvature of the depicted vertebral column, a length of the depicted vertebral column and an aspect ratio of the depicted vertebral column.

3. The method according to any one of the preceding claims, wherein step (c) involves employing a computer implemented image analysis system for identifying the vertebral column depicted in respective image and for measuring or evaluating the least one characteristic of the depicted vertebral column.

4. The method according to any one of the preceding claims, wherein step (d) involves employing computer vision and algorithms built by a machine learning system utilising mathematical models based on sample or training images of fish eggs known to be viable or non-viable.

5. The method according to claim 4, wherein step (d) involves utilising a machine learning system having been trained on a training set of images depicting fish eggs having been classified as viable or non-viable.

6. The method according to claim 5, wherein said training set comprises a plurality of classes of abnormal or non-viable fish eggs comprising at least two of: a class comprising eggs with a Y or V shaped vertebral column; a class comprising eggs with a corkscrew or curled vertebral column; a class comprising eggs with a crooked vertebral column; and a class comprising eggs with a stubby or short vertebral column.

7. The method according to any one of claims 5 and 6, wherein step (d) comprises classifying a non-viable fish egg according to type of spinal deformation.

8. The method according to any one of claims 5-7, wherein said machine learning system comprises a trained neural network model having been trained on a training set of images depicting fish eggs having been classified as viable or non-viable.

9. The method according to any one of the preceding claims, wherein step (d) involves using a machine learning system having been trained on a training set of images depicting fish eggs of the same species as being analysed according to the method.

10. The method according to any one of the preceding claims, comprising the step of:- identifying one of the images of the first and second set of images as a representative image; and- classifying the identified representative image as being associated with one of a viable egg and a non-viable egg.

11. The method according to claim 10, wherein the step of identifying one of the images of the first and second sets of images as a representative image may comprise measuring or evaluating the rectilinearity of the vertebral column of the first and second sets of images and choosing the image displaying the least rectilinear vertebral column as the representative image.

12. The method according to any one of the preceding claims, comprising the step of, prior to steps (a)-(d), treating the fish egg (100) with a fixative bringing the outer membrane of the egg at least partly transparent and / or the vertebral column of the egg more visible.

13. The method according to any one of the preceding claims, wherein said step (c) comprises measuring or evaluating the diameter of the egg (100), and wherein said step (d) comprises classifying the fish egg as viable or non-viable based on the at least one measured or evaluated characteristic and the measured or evaluated diameter of the egg.

14. A method of determining a viability rate of a batch of fertilised fish eggs having incubated to spine stage, comprising the steps of:- drawing a sample comprising a plurality of eggs from the batch of incubating fish eggs;- performing the steps (a)-(d) according to any one of claims 1-13 on each egg of the drawn sample; and- calculating the viability rate as a ratio between the number of eggs classified as being viable and the total number of eggs in the sample.

15. A fish egg analysis apparatus (200) comprising:- an imaging system (208) configured for:- capturing a first set of images of a fish egg (100) when the fish egg is rotated about a first axis of rotation (225a); and- capturing a second set of images of the fish egg when the fish egg is rotated about a second axis of rotation (225b) which is orthogonal to the first axis of rotation (225a);- an image analysis system (216) configured for:- identifying, in each image in the first and second set of images, a vertebral column depicted in the image;- measuring or evaluating at least one characteristic of the depicted vertebral column; and- based on the at least one measured or evaluated characteristic, classifying the fish egg as viable or non-viable.

Images