Analyzing samples of fluids containing microplastics

Abstract

The disclosure notably relates to a method for analyzing samples of fluids containing microplastics. The method comprises inputting a sample of fluid into a liquid input channel. The method also comprises dividing the sample of fluid from the liquid input channel into at least two branch channels. The microplastics are distributed among the branch channels based on their size. The method also comprises by each branch channel, conveying the sample of fluid to a respective microfluidic slide arranged facing a respective camera. The method also comprises capturing with each respective camera a respective flux of images. The method also comprises processing each respective flux of images to analyze the sample of fluid.

Description

[0001] ANALYZING SAMPLES OF FLUIDS CONTAINING MICROPLASTICS

[0002] TECHNICAL FIELD

[0003] The disclosure relates to the field of fluid sample analysis, and more specifically to a method and device for analyzing samples of fluids containing microplastics.

[0004] BACKGROUND

[0005] Systems for biowaste valorization have been gaining more and more importance for more adequate waste management systems and regulations. Anaerobic digestion process (AD) and composting process are two major pathways for biowaste valorization. In an AD process, biowaste can be transformed into biogas and digestate by microorganisms in an oxygen-free environment. In a composting process, biowaste can be transformed into compost by microorganisms in an oxygenrequiring environment. Digestate and compost are both nutrient-rich substances that may be returned to agricultural soils to be used as organic amendments. Moreover, digestate may also be recirculated as a methanization input, which may be used for biogas production. The quality of a digestate or compost derived from biowaste (and naturally the future quality of the soil to which it is returned), depends on the one hand, on the quality of biowaste and, on the other hand, on the conditions of the AD process and the composting process.

[0006] A major contaminant of all types of biowaste, independent of their source, is plastic. This includes conventional plastics (fossil-based non-biodegradable plastics) from food packaging products and bioplastics (bio-based and / or biodegradable plastics) from growingly popular bioproducts (e.g., biobags, bio-coffee capsules, biocups).

[0007] The amount of microplastic in biowaste before being transformed by AD or composting will depend on the source-separation system or the post-separation system and any process used for plastic removal. Moreover, microplastic that enters the AD process or composting process may not be biodegraded (as is the case of conventional microplastics). Thus, there is the need for verifying the quantity and physical characteristics of the microplastics after the AD or composting process, so that the output of the AD or composting process complies with quality regulations. For example, in European countries, there are limits for microplastics larger than 2 mm.

[0008] Hence, there is a need to assess the risks and impact due to the presence of microplastic on fluid samples. To this day, there are no standardized methods for measuring microplastics in fluids such as composts, digestates, food waste or other biowastes. Moreover existing analytical methods for the quantification of microplastics require long analysis time and are labor intensive.

[0009] Within this context, there is still a need for an improved method for analyzing samples of fluids containing microplastics.

[0010] SUMMARY

[0011] It is therefore provided a method for analyzing samples of fluids containing microplastics. The method comprises inputting a sample of fluid into a liquid input channel. The method also comprises dividing the sample of fluid from the liquid input channel into at least two branch channels. The microplastics are distributed among the branch channels based on their size. The method also comprises by each branch channel, conveying the sample of fluid to a respective microfluidic slide arranged facing a respective camera. The method also comprises capturing with each respective camera a respective flux of images. The method also comprises processing each respective flux of images to analyze the sample of fluid.

[0012] The method may comprise one or more of the following: dividing the sample of fluid from the liquid input channel into at least two branch channels is performed according to at least two consecutive ranges of microplastics sizes; at least one range of the at least two consecutive ranges of microplastics sizes has an upper boundary lower than or equal to 100 micrometers, and at least one other range has a lower boundary higher than or equal to the upper boundary of the at least one range and an upper boundary higher than 100 micrometers; each respective camera is a digital camera, and wherein the resolution of the capturing increases as the size of the microplastics decreases per branch channel; - the resolution of the capturing is below 8.5 pixels per micrometer for at least one respective camera and / or above 8.625 pixels per micrometer for at least one respective camera; at least one optical path in front of the respective camera comprises two crossed polarizers; at least one optical path in front of the respective camera comprises an infrared detector and / or a monochromatic filter; and / or at least one optical path in front of the respective camera comprises an arrangement of a phase contrast microscope; processing each respective flux of images comprises determining one or more portions of each image comprising a respective class of microplastic on the flux;

[0013] - the method further comprises estimating, on each respective flux of images, one or more physical properties of the microplastics on the fluid sample, among: o a quantity of the microplastics; o a shape of the microplastics such as a spheric shape; o a volume of the microplastics; o a distribution of the size of the microplastics; o a rugosity of the microplastics; and / or o a crystallinity of the microplastics; estimating the physical properties of the microplastics comprises applying one or more neural networks on each respective flux of images, the one or more neural networks being configured to determine the physical properties of the microplastics on the respective flux of images; processing each respective flux of images comprises transmitting at least one respective flux of images;

[0014] - the method further comprises concentrating microplastics in the sample of fluid from the input channel before dividing the sample of fluid; - the dividing of the sample of fluid from the liquid input channel is performed by centrifugal force or fractionation;

[0015] - the centrifugation is performed by inputting the sample of liquid into a spiral centrifugal separator;

[0016] - the at least two channels are aligned on a same reference plane;

[0017] - the sample of fluid is of a compost liquid or of a methanization digestate.

[0018] It is further provided a device configured for analyzing samples of fluids containing microplastics according to the method. The device comprises the liquid input channel. The device also comprises a separator configured for performing the dividing of the sample of fluid. The device also comprises the at least two branch channels and each respective microfluidic slide. The device also comprises the at least two cameras.

[0019] The device may comprise one or more of the following:

[0020] - the device further comprises a concentrator configured for concentrating the sample of fluid;

[0021] - the at least two channels are aligned on a same reference plane;

[0022] - the separator is a spiral centrifugal separator.

[0023] BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Non-limiting examples will now be described in reference to the accompanying drawings, where:

[0025] FIG. 1 shows a flowchart of an example of the method;

[0026] FIG. 2 shows a schematic representation of an example of the device; and FIG.s 3 to 16 illustrate examples of the method and device.

[0027] DETAILED DESCRIPTION

[0028] With reference to the flowchart of FIG. 1, it is proposed a method for analyzing samples of fluids containing microplastics. The method comprises inputting S10 a sample of fluid into a liquid input channel. The method also comprises dividing S20 the sample of fluid from the liquid input channel into at least two branch channels. The sample of fluid circulates / flows into and inside the liquid input channel, and the liquid stream is then divided or separated into two branches, the liquid thereafter ci rculati ng / f lowi ng into and inside the at least two branch channels. The microplastics are distributed among the branch channels based on their size. The method also comprises, by each branch channel, conveying or flowing S30 the sample of fluid to a respective microfluidic slide arranged facing a respective camera. Each respective microfluidic slide is arranged across an optical path of the respective camera, for example perpendicular or substantially perpendicular to the optical path, thus facing the respective camera. The method also comprises capturing S40 with each respective camera a respective flux of images (as the liquid stream flows into and inside each microfluidic slide). The method also comprises processing S50 each respective flux of images to analyze the sample of fluid.

[0029] Such a method improves the analysis of samples of fluids containing microplastics.

[0030] Notably, as the microplastics are distributed among the branch channels based on their size, the method improves the discrimination of microplastics captured from each respective camera. Indeed, as small-sized microplastics conveyed on a respective microfluidic slide are separated from larger sized microplastics conveyed to other respective microfluidic slide, the small-sized microplastics are not or less obstructed from the field of view of the respective camera by larger sized microplastics. In addition, as the larger sized microplastics are separated from the small sized microplastics, there is a significant reduction of the probability of the microplastics clogging up.

[0031] At least some steps of the method may be computer implemented. This means that steps (e.g., the capturing S40 and / or the processing S50) of the method are executed by at least one computer, or any system alike. Thus, steps of the method are performed by the computer, possibly fully automatically, or, semi-automatically.

[0032] With reference to FIG. 2, it is further provided a device 200 configured for analyzing samples of fluids containing microplastics according to the method. The device comprises the liquid input channel 210. The liquid input channel 210 may be connected to (i.e., brought in fluid communication with) a liquid conduit 212. The liquid conduit 212 may be part of or external to the device. The device may include a liquid input port 211 which may be configured for connecting or disconnecting the liquid conduit 212 from the liquid input channel 210. The liquid input port 211 may be arranged on the enclosure of the device, for example on a housing of the device which encloses other components of the device. The housing may be made of metal and / or plastic material, for example biodegradable polymer such as EN13432. The liquid input channel may be enclosed in the housing. The liquid input port 211 may be connected at one end to the liquid conduit 212, and at the other end to the liquid input channel 210. The liquid input channel 210 may be made of a flexible or rigid material.

[0033] The device also comprises the at least two branch channels 231, 232 and each respective microfluidic slide 241, 242. The device may also comprise the at least two cameras 251, 252. Each branch channel 231, 232 and each microfluidic slide 241, 242 may be enclosed in the housing. Each branch channel 231, 232 may be made of a flexible or rigid material. Each microfluidic slide 241, 242 may be made of a transparent material, for example a transparent plastic material or a transparent glass material.

[0034] The device also comprises a separator 220 configured for performing the dividing of the sample of fluid. The separator 220 may be enclosed in the housing. By "separator" it is meant any device configured for receiving the sample of fluid as input and to output portions of the sample of fluid. For example the separator 220 may comprise obstacles that separate the microplastics in the sample of fluid. Alternatively, the separator 220 may be configured to perform centrifugation on the sample of fluid.

[0035] The separator may in particular be a spiral centrifugal separator, as explained below.

[0036] The liquid input channel 210, the separator 220, and / or the at least two branch channels 231, , may have controlled surface properties for avoiding the adhesion of microplastics. For example, the liquid input channel, the separator, and / or the at least two branch channels may have at least one surface made of or coated with fluorinated silanes or fluorinated materials such as Teflon.

[0037] The device may optionally further comprise a concentrator configured for concentrating microplastics in the sample of fluid. By "concentrator" it is meant any device configured for receiving the sample of fluid and outputting the sample of fluid with a higher density of microplastics. The concentrator may have a spiral geometry. The concentrator may comprise a central input for receiving the sample of fluid. The concentrator may comprise a curved channel extending in a spiral shape from the central input in an outward direction. The curved channel may comprise two output ports, one output port may output the concentrated microplastics, the other may output the fluid without the microplastics. The curved channel may comprise an inner central wall (facing the central input) and an outer central wall (facing the downward direction). The inner central wall may comprise a directional tangential filter, that is, a physical barrier (e.g., comprising fins or walls in a tangential direction with respect to the inner central wall) that separates particles from the fluid. This improves the flexibility for sorting particles as the separator may receive a more consistent flux of microplastics. The use of the concentrator may depend on initial microplastic concentration in the samples to be analyzed.

[0038] The liquid input channel may be of any geometry (e.g., straight, or allowing a degree of curvature) adapted for allowing the flow of the sample of fluid in a small scale (e.g., sub-millimeter scale). The liquid input channel may have a cross-sectional diameter for allowing the entry of the sample of fluid. The diameter may be of a predetermined length, for example 1mm or more, for example between 2mm and 5mm or more. Additionally or alternatively, the diameter may be dependent on the size of the largest microplastic to be sorted. For example, the diameter may be of a ratio between two to ten times larger than the largest size (e.g., the largest cross- sectional length) of microplastics to be sorted, for example 5 times yields good results. For example, when the microplastics have a maximum size of 400 micrometers, a diameter of 2mm may give a ratio of 5, a diameter of 5mm may give a ratio of 11, a diameter of 1mm may give a ratio of 2.5.

[0039] The device may comprise one or more outlets for evacuating liquids, for example one outlet for evacuating liquid extracted by the optional concentrator and deprived of microplastics, and one outlet 260 for evacuating liquid outputted by each microfluidic slide (after an image thereof has been captured). Inputting S10 the sample of fluid may comprise conveying the sample of fluid from an external liquid medium such as composts, digestates, food wastes or other biowastes, resulting for example from an anaerobic digestion process or composting process. Thus, the method may be included in a process comprising: providing the external liquid medium; extracting a sample of fluid from the external liquid medium; performing the method to perform the analysis.

[0040] In an example, the process may be adapted based on a result of the analysis. For example, the process may estimate a quantity of microplastics of the sample of fluid. The process may estimate the environmental impact of the presence of microplastics on the anaerobic digestion process or composting process, and perform corrective actions so as to improve the quality of the anaerobic digestion process or composting process

[0041] The sample of fluid is divided S20 from the liquid input channel into at least two branch channels. In other words, each of the at least two branch channels receives a portion of the sample of fluid. Each branch channel may be of a rigid material or a flexible material.

[0042] The microplastics are distributed among the branch channels based on their size. In other words, microplastics on the sample of fluid of a respective size are driven into a respective branch channel. For example, microplastics of a first size may be driven into a first branch channel, microplastics of a second size, the second size being larger than the first size, may be driven into a second branch channel. For example, the first size may be between 5 and 100 micrometers, the second size may be between 100 micrometers and 400 micrometers.

[0043] By each branch channel, the method conveys S30 the sample of fluid to a respective microfluidic slide. The respective microfluidic slide may comprise a thin flat conduit made of a transparent material and formed between or enclosed by a pair of parallel plates. The thin flat conduit may allow the flow of the sample of fluid. The pair of parallel plates may be of transparent material such as glass, thereby allowing the observation of the sample of fluid passing through the thin flat conduit. The parallel plates may be substantially flat. The microfluidic slide may comprise an entry connector configured to receive the conveyed sample of fluid. The microfluidic slide may have at another output connector configured for throwing out the sample of fluid. The thin flat conduit may have a predetermined depth (e.g., relative to a perpendicular direction of the sample of fluid along the slide) having a thickness between 50pm and 1mm and of width between 200pm and 10mm. The microfluidic slide may have frontal area dimensions of 75 mm x 25 mm or more, for example 75 mm x 50 mm. A microfluidic slide may have a predetermined spacing so as to allow the flow of the sample of fluid through the thin flat conduit. That is, the spacing may depend on the size of the microplastics to observe, for example 30 micrometers or more, for example 200 micrometers or even 600 micrometers or more.

[0044] The microfluidic slide is arranged facing a respective camera. In other words, the microfluidic slide is arranged so that the conveyed sample of fluid is captured by the respective camera, that is, so that there exists at least one optical path between the sample of fluid conveyed into the microfluidic slide and the respective camera. For example, one of the pair of parallel plates may face the respective camera. At least a portion of the parallel plate may be arranged so that the conveyed sample of fluid is visible by the respective camera, for example so that the sample of fluid passing through the conduit can be seen by the respective camera.

[0045] The method captures S40 with each respective camera a respective flux of images. In other words, the method captures with each respective camera a succession of images (i.e. a time-series of images). The flux of images is a set of successive 2D images representing the sample of fluid on the microfluidic slide as captured from the respective camera, each 2D image corresponding to a respective frame of the video flux. The capturing may be performed continuously, such that two successive frames of the video flux may be close one to another (e.g. separated by a time period less than 1 or 0.1 second). Each respective camera may be configured to capture the flux of images at a respective capturing frequency, e.g., 15 frames per second (fps) or more, for example, 30 fps, 60 fps or 120 fps or more. Each respective camera may function time-synchronously or time-asynchronously with at least one other respective camera (e.g., all cameras being synchronous in an example). By "time-synchronously", it is meant is that the capturing of the flux of images by the respective camera and the at least other respective camera (for example all of the cameras) is performed at the same time; for example, at a time t, the respective camera and the at least other respective camera capture a frame of a respective flux. By "time-asynchronously", it is meant that capturing of the flux of images by the respective camera is performed at times which are uncorrelated from the capturing of the flux of images of the other respective camera. Each image may be timestamped, such that images captured from different cameras may be synchronized.

[0046] The method processes S50 each respective flux of images to analyze the sample of fluid. In other words, the method may execute one or more set of operations on each respective flux of images, for example, compressing each respective flux, applying image enhancement methods, image processing methods and / or storing each respective flux of images. Processing the respective flux of images may comprise synchronizing the flux of images, meaning that, based on the dimensions of the microplastics conveyed by each respective branch channel, the method may take into account the time spent by the sample of fluid conveyed from the separator into to each microfluidic slide, and use this information to present the flux of images of different branch channels into a one-to-one correspondence, so that the method presents the respective flux of images taken from different cameras as having been captured from the same initial liquid from which the conveyed microplastics where distributed.

[0047] The method thus improves the analysis of samples of fluids containing microplastics. As the method divides the sample of fluid from the liquid input channel into at least two branch channels, microplastics are separated based on their size and thus microplastics on a given channel are not obstructed by microplastics (for example of a larger size) of a different channel when captured by a corresponding respective camera. In addition, as the sample of fluid is conveyed by each branch channel, this improves the creation of sediments that may obstruct the microfluidic slides. Consequently, the method improves the efficiency of the analysis, as there is less need to perform maintenance to the microfluidic slides due to the presence of sediments. In addition, as the method performs the processing on each respective flux of images capturing microplastics distributed based on their size, the method allows the application of image enhancement image processing methods and / or storing each respective flux of images taking into account the size of the microplastics conveyed on each branch.

[0048] In other words, the method allows an efficient flow of the microplastics contained in the sample of fluid, and captures continuously (e.g., "on the fly") respective flux of images which corresponding to a stream of microplastics (separated based on their respective sizes) as these traverse a respective microfluidic slide. This allows an improved discrimination and analysis of the microplastics. Thus, the device allows a processing on-the-fly of the images of the liquids, thereby reducing waiting times and improving reliability against the formation of sediments or clogs as the sample of fluid is conveyed.

[0049] Dividing the sample of fluid from the liquid input channel into at least two branch channels may be performed according to at least two consecutive ranges of microplastics sizes. In other words, dividing the sample into at least two branch channels may be performed by conveying microplastics of the sample of fluid into a respective branch channel depending on the range of microplastics sizes. By consecutive it is meant that the at least two ranges followed each other continuously among one or more predetermined size thresholds between a lower boundary threshold and an upper boundary threshold. Each branch may correspond to a range of sizes between two consecutive size thresholds or to a range of sizes above (respectively below) the lower (respectively upper) boundary threshold. For example, microplastics having a size in one of the ranges may be conveyed to one respective branch channels, and microplastics having a size on the other consecutive range may be conveyed (independently from microplastics of a size outside the respective range) to at least other of the respective branch channels. The at least two ranges may be separated by a predetermined threshold. For example, at least one range of the at least two ranges may be below the predetermined threshold, at least one other range (e.g., all of the ranges) are above the predetermined threshold.

[0050] Thus, the method separates microplastics on the sample of fluid such that microplastics having a size between a respective consecutive range is efficiently captured by a respective camera, as the method allows to discern microplastics according to their respective size.

[0051] At least one range of the at least two consecutive ranges of microplastics sizes may have an upper boundary lower than or equal to 100 micrometer. The upper boundary may be any predetermined size, for example 80 micrometers. At least one other range may have a lower boundary higher than or equal to the upper boundary of the at least one range. The lower boundary may be of a given predetermined size, for example 90 micrometers. The at least one other range may also have an upper boundary higher than 100 micrometers. In examples, the upper boundary may be of 200 micrometers. In such examples, the at least one range my have a range between 0 and 80 micrometers and the other range is between 90 micrometers and 200 micrometers In other examples, the maximal value of a first range of sizes may be lower than or equal to 100 micrometers, and minimal value of a second range of sizes is higher than or equal to 100 micrometers. The 100 micrometer upper threshold may separate the two consecutive ranges (100 micrometers being thus a boundary of the two ranges). Or alternatively, at least one of the two ranges may be separate from the 100 micrometer value). In other words, microplastics having a size below the 100 micrometer upper threshold (for example 40 micrometers or smaller, e.g., 5 micrometers) may be conveyed to a first branch channel, microplastics having a size above the 100 micrometer upper threshold may be conveyed to at least one other channel, for example a second branch channel. The first branch channel may have a depth between 200 micrometers and 300 micrometers. The first branch channel may have a depth between 500 micrometers and 600 micrometers. This results in a good compromise for capturing microplastics according to a respective size in current applications, where microplastics may range in sizes between 40 micrometers and 400 micrometers, and where 100 micrometers is a good compromise for analyzing microplastics (e.g., smaller than 80 micrometers) that would be otherwise be obstructed from microplastics Having a size over 100 micrometer. In other examples, a first consecutive range of microplastic sizes may be below a 20 micrometer upper threshold, a second consecutive range of microplastic sizes may be below a 80 micrometer upper threshold. The method may convey microplastics having a size below 20 micrometers to a first branch channel. The method may also convey microplastics having a size between 20 micrometers to 80 micrometers to a second channel. The method may convey microplastics having a size above 80 micrometers (for example between 80 micrometers and 400 micrometers) to a third branch channel.

[0052] Each respective camera may be a digital camera. Each respective digital camera may have a respective resolution, for example, between, 2 megapixels and 12 megapixels or more, for example 64 megapixels. The resolution of the capturing may increase as the size of the microplastics decreases per branch channel. The method may increase the resolution of the camera by assigning a predetermined number of pixels per micrometer.

[0053] This results in that the method improves the accuracy of the analysis of the samples of fluids with respect to the sizes being conveyed on each respective microfluidic slide. Indeed, the method improves the recognition of visual characteristics of microplastics according to a respective size, thanks to the method adapting the resolution of a respective camera depending on the size of microplastics.

[0054] In examples, a first branch channel may convey a first portion of the sample of fluid containing microplastics of a first size to a first respective microfluidic slide facing a first respective camera. For example, the first size may be between 5 and 100 micrometers. The second branch channel may convey a second portion of the sample of fluid containing microplastics of a second size to a second respective microfluidic slide facing a second respective camera, the second size being larger than the first size; for example, the second size may be between 100 micrometers and 400 micrometers. The method may assign a first resolution of pixels per micrometer for the first respective camera, and a second resolution of pixels per micrometer for the respective camera. The first resolution is larger than the second resolution. Thereby, in this example, the method improves the recognition of the microplastics of the first size, as the increase in resolution improves the recognition of visual features of the microplastics compared to the second resolution. The cameras may capture digital images, and the resolution of the capturing may be below 8.5 pixels per micrometer for at least one respective camera. For example, the resolution may be of about 5.75 pixels per micrometer when the microplastics have a size between 100 and 400 micrometers ("about" meaning + / - 10% in the disclosure). Alternatively or additionally, the resolution of the capturing may be above 8.625 pixels per micrometer for at least one respective camera. For example, the resolution may be of about 11.5 pixels per micrometer when the microplastics have a size between 5 and 100 micrometers.

[0055] This results in a good compromise for capturing microplastics according to a respective size. A higher resolution helps better analyze smaller particles, while for larger particles a lower resolution is needed so that the particle does not fill the image captured completely.

[0056] At least one optical path in front of the respective camera may comprise two crossed polarizers (part of the device). By "optical path" it is meant a linear trajectory that a light ray (e.g., emitted by a light source) follows as it propagates in the space from the light source (for example through an optional filter, a condenser) passing through the microfluidic slide, through another optional filter and the respective camera. The two crossed polarizers may be configured for performing cross polarization, that is, only allowing the pass of light with a predetermined orientation, e.g., so as to allow discerning the microplastics conveyed on the respective microfluidic slide facing the camera. In this configuration, one polarizer may be disposed between the light source and the sample, a second polarizer is disposed between the other side of the sample and the detector (camera). The two polarizers are rotated at 90 degrees with respect to each other. The cross-polarizers enable to differentiate materials based on their birefringence, which most microplastics exhibit.

[0057] At least one optical path in front of the respective camera may comprise an infrared detector. Additionally or alternatively, at least one respective camera may comprise a monochromatic filter (part of the device). This allows to detect microplastics based on their infrared spectra, or microplastics that may be difficult to discern from the background under visible light. Additionally or alternatively, at least one optical path in front of the respective camera may comprise a phase contrast microscope (part of the device). The phasecontrast microscope may be configured to convert phase shifts in light passing through the microplastics conveyed in the microfluidic slide to brightness changes when represented in the respective flux of images. This enables the visualization of components of the microplastics that would be difficult to see under visible light.

[0058] Processing each respective flux of images may comprise determining one or more portions of each image. Each portion may comprise a respective class of microplastic on the flux.

[0059] In other words, the method may take the flux of images as input and output one or more portions of the image enclosing the respective class of microplastic. For example, the method may determine at least one position or localization (e.g., in x,y- coordinates on each respective image) indicative of the position of the microplastic on each respective image. A respective portion may be a group of pixels of each image enclosing pixels representing respective microplastic enclosing the image. For example, the respective portion may be a bounding box (e.g., comprising pixels of the microplastic and pixels of the background) or a segmentation of the image comprising only pixels representing the respective microplastic. The method may apply image processing methods (e.g., object localization and / or image segmentation techniques) to obtain the one or more portions of the image enclosing the respective microplastic.

[0060] By "class" it is meant a category of the microplastic. In other words, microplastics of a respective class may share similar physical properties or physical constraints (e.g., a relative size or shape).

[0061] In examples, determining the one or more portions may comprise applying one or more neural networks on each respective flux of images. The one or more neural networks may be configured to determine the one or more portions of the image.

[0062] This results in a precise determination of the microplastics on the respective flux of images, as the method may concentrate the processing on the portions enclosing the respective microplastic (so that portions on the background are discarded). The method may further comprise estimating, on each respective flux of images, one or more physical properties of the microplastics on the fluid sample. By "estimating" it is meant obtaining a respective value from the one or more physical properties of a respective microplastic represented in the flux of images. Performing the estimation may comprise applying a set of computations on the flux of images, for example on the portions of each image comprising the respective class of microplastics on the flux. Performing the estimation may comprise extracting a distribution of the area and / or color of a respective microplastic, and performing statistics operations on moments of the distribution of area and / or color. The one or more physical properties may also comprise a quantity of the microplastics, that is, the number of microplastics captured on a respective flux of images. The one or more physical properties comprise a class of the microplastics, e.g., conventional plastics (for example fossil-based non-biodegradable plastics), bioplastics. The one or more physical properties may also comprise a shape of the microplastics. The shape of a given microplastic may be the spatial distribution of the microplastic as represented in the respective flux of images, for example a spheric shape. The one or more physical properties may also comprise a volume of the microplastics, e.g., represented as a ratio of pixels representing microplastics of the flux of images with respect to the background. The one or more physical properties may also comprise a distribution of the size of the microplastics. The one or more physical properties may also comprise a rugosity of the microplastics, and / or a crystallinity of the microplastics.

[0063] This further improves the accuracy of the analysis of samples of fluids containing microplastics. Indeed, thanks to the estimation of the one or more physical properties, the device extracts more accurate information from the microplastics present in each respective flux of images. For example, a respective flux of images may be enriched with information such as the class of the microplastic, the quantity of the microplastics represented on the respective flux of images, and / or distribution (or density) of the microplastics in the flux. This information thus provides an accurate indicator of the pollution of the sample of fluid, e.g., when the sample of fluid comes from an anaerobic digestion process or a composting process. In examples, the method may detect one or more objects on a respective flux of images. The method may extract the contours of respective detected one or more objects. The method may perform the estimation of one or more physical properties on the detected one or more objects, for example by performing mathematical operations on the area enclosed by the contours, for example obtaining a distribution of the area of the contours of each detected object and / or the colors. The method may extract the data and statistics from the detected objects.

[0064] Estimating the physical properties of the microplastics may comprise applying one or more neural networks on each respective flux of images. By "estimating" it is meant that the neural network outputs a value indicative of a respective physical property on the microplastics represented by the respective flux of images. For example, the one or more neural networks may compute for each respective localization of a microplastic on the flux of images (e.g., a localization being a bounding box enclosing the microplastic, represented by coordinates (x, y) and a size, such as width and height) a respective output representing values of the one or more physical properties. In other words, for a location of at least one microplastic represented by a respective flux of images, the one or more neural networks measure the value of the one or more physical properties of the at least one respective microplastic on its respective localization. The one or more neural networks may provide each outputted value of the one or more biological attributes in the form of one or more labels.

[0065] A neural network is a function comprising a collection of connected nodes, also called "neurons". Each artificial neuron receives an input and outputs a result to other neurons connected to it. The artificial neurons and the connections linking each of them have weights, which are adjusted via a training process. Each respective flux of images may be provided raw (e.g., as acquired from the respective camera) to the neural network, or alternatively after having been processed (e.g., after being compressed or image-processed).

[0066] The one or more neural networks may be configured to determine the physical properties of the microplastics on the respective flux of images. In other words, weights on each respective neural networks may comprise weights configured so that, when inputting the respective flux of images, the one or more neural networks output a value representative of the physical properties of the microplastics on the respective flux of images. In examples, the one or more neural networks may be applied to the determined one or more portions of each image. Alternatively, the one or more neural networks may be configured to determine the one or more portions of the image.

[0067] Thus, the method leverages from the accuracy achieved by the one or more neural networks for analyzing samples of fluids containing microplastics. The use of (trained) one or more neural networks results in relatively fast processing times, all while yielding an accurate estimation of the one or more physical properties of the microplastics represented in the flux of images.

[0068] Each neural network (of the one or more neural networks) may have been trained according to a machine-learning method (i.e., the value of its weights and parameters ensure a level of prediction / inference which reflects such training). The machine-learning method may comprise providing a dataset comprising a set of training samples.

[0069] As known per se from the field of machine-learning, the dataset impacts the speed of the learning of the one or more neural networks and the quality of the learning, that is, the accuracy of the trained one or more neural networks to analyze the flux of images. The dataset may be provided with a set of training sample that depends on the contemplated quality of the learning for performing the determination of the physical properties of the microplastics. This set may comprise a number of training samples higher than 1 000, 10 000, or yet 100 000 training samples. Each training sample may comprise a respective flux of images, one or more portions of each image comprising a respective microplastic on the flux, each microplastic being associated to one or more annotations containing values representative of a respective physical properties of the microplastics (e.g., in the form of respective one or more labels). The machine-learning method may also leverage from other training samples of the dataset not included in the set, e.g., training samples containing any microplastic, e.g., such as a training sample comprising images of liquid medium or training sample comprising images of microplastics together with other objects such as pollutants and / or microorganisms. The quantity of the training samples in the dataset contemplated for the training thus follows a tradeoff between the accuracy to be achieved by the one or more neural networks, and the speed of the training.

[0070] The machine-learning method may also comprise training the neural network based on the provided dataset. As known per se from the field of machine-learning, the training proceeds to adjust the weights of the neural network according to the computed output. As known per se, the output is compared to the values of the annotations in the training samples and the weights may be adjusted according to such comparison. The performance of the accuracy due to learning may be tracked using standard machine-learning methods.

[0071] The one or more neural networks may comprise a YOLOv7 neural network or a YOLOv8 neural network. In experiments, the inventors have appreciated that the specific neural networks yields good results for estimating the physical properties of the microplastics.

[0072] Processing each respective flux of images may comprise transmitting at least one (e.g., all of) respective flux of images (that is, captured from a respective camera). The method may transmit the at least one respective flux of images through a network. The transmission may a wired transmission, such as via an Ethernet connection or wireless, for example via Wi-Fi, Bluetooth or Radio transmission.

[0073] The transmission allows to further analyze the at least one respective flux of images, for example on a client terminal in a laboratory.

[0074] The method may further comprise concentrating microplastics on the sample of fluid from the input channel before dividing the sample of fluid. In other words, the method may increase the density of the microplastics on the fluid before dividing the sample of fluid.

[0075] As the microplastics are concentrated when dividing the sample of fluid, the transfer of the microplastics to respective microfluidic slides (based on their size) is performed faster, as the concentration allows a rapid flow of the microplastics. This results in that the time for analyzing the sample of fluid is reduced. The dividing of the sample of fluid from the liquid input channel may be performed by centrifugal forces for separating microplastics with different sizes. Alternatively, the dividing may be performed by fractionation, e.g., pinched flow fractionation, that is, a continuous separation of sizes of particles, for example by employing a microchannel comprising a laminar flow profile.

[0076] It has been found that microplastics of a first range of microplastic sizes are distributed on a direction more outward than microplastics of a second (consecutive) range of microplastic sizes when the first range is smaller than the second range, when the sample of fluid undergoes centrifugal forces.

[0077] The centrifugal force may be applied by inputting the sample of liquid into a spiral centrifugal separator. The spiral centrifugal separator may comprise a central input and respective outputs. Each respective output may be configured for outputting microplastics of a respective size. The spiral centrifugal separator may comprise a curved channel extending in a spiral shape from the central input in an outward direction. The curved channel may comprise an inner central wall (facing the central input) and an outer central wall (an interior wall inside the curved channel facing the downward direction). When inputting the sample of liquid into the spiral centrifugal separator, the sample of fluid may undergo centrifugal acceleration as the sample is pushed along the curved channel to a respective output, for example leading to the formation of inertial flows comprising microplastics of different sizes.

[0078] It has been found that microplastics tend to stat closer to the inner central wall of the curved channel of the spiral centrifugal separator as the size increases. For example, microplastics of a first range of microplastic sizes are distributed on a direction more outward (which are distributed towards the outer wall of the spiral centrifugal separator) than microplastics of a second (consecutive) range of microplastic sizes when the first range is smaller than the second range (which are distributed towards the inner central wall of the spiral centrifugal separator compared to the microplastics of the first range), when the sample of fluid undergoes centrifugal forces. The respective outputs may be distributed so as to separate the microplastics as these are conveyed through the spiral centrifugal separator. Each respective branch channel may be connected to a respective output.

[0079] Each respective branch channel may have a respective depth for confining microplastics of the respective range of sizes. The at least two branch channels may be aligned, e.g., from the output of the spiral centrifugal separator on a same reference plane. For example, one channel may be aligned with the outward wall of the curved channel. The other channels may be aligned sequentially towards the inner central wall of the curved channel. Thus, each branch channel is positioned so that the images taken with the respective camera is of high quality for each respective branch channel, as the alignment compensates from the difference of sizes on each respective branch channel.

[0080] The sample of fluid may be of a compost liquid (for example taken from an anaerobic digestion process) or of a methanization digestate (for example taken from an composting process).

[0081] The method hence allows the analysis and monitoring of microplastics throughout the anaerobic digestion process and / or composting process, serving as a tool to evaluate the anaerobic digestion process and / or composting process in its capacity to biodegrade conventional and biodegradable plastics and the environmental impact of the anaerobic digestion process and / or composting process (in terms of its contribution to leakage of microplastic). Additionally or alternatively, the sample of fluid may be sea water, fresh water and wastewater.

[0082] Examples of the method are now discussed with reference to FIG.s 3 to 13.

[0083] FIG. 3 illustrates examples 300 of the spiral centrifugal separator 310-340. The spiral centrifugal separator 310 comprises a central input port 311 and output port 312. The central input port 311 may receive for example the output of the concentrator (not shown) while the output port 312 may distribute the microplastics divided on set of ranges to the branch channels based on their size. Similarly, the spiral centrifugal separator 320 comprises an central input port 321 and output port 322, comprising less turns than the spiral centrifugal separator 310. In the same manner, the spiral centrifugal separator 320 comprises a central input port 321 and output port 322 and the spiral centrifugal separator 340 comprises a central input port 341 and output port 342. The spiral centrifugal separators comprise different widths (for example between 0.8mm and 2mm) and may have a different number of turns.

[0084] FIG. 4 shows an experiment 400 for dividing the sample of fluid with a spiral centrifugal separator 410. In the experiment, it is shown a sample of fluid being input through the input port 411. The separated fluid is shown in the output port 412.

[0085] FIG. 5 shows an enlarged view 500 of the experiment. The enlarged view 500 illustrates how the microplastics are separated in the spiral centrifugal separator when the sample of fluid undergoes centrifugal acceleration as it its pushed along the curved channel 510 of the spiral centrifugal separator. Microplastics of a smaller size 511 tend to be on an inner central wall of the curved channel 510 while microplastics of a larger size 512 tend to be on an outer central wall of the curved channel 510.

[0086] FIG. 6 shows paths followed by the microplastics on the enlarged view 500. The microplastics having the size 511 follow a path 610 while the microplastics having the size 512 follow the path 620. As the paths separate through the spiral centrifugal separator, the microplastics may be conveyed to different branch channels (not shown).

[0087] FIG. 7 illustrates a fractionator 700 for dividing the sample of fluid by pinched flow fractionation. The fractionator 700 comprises an input port for receiving the sample of fluid, and a port 720 for receiving liquid without particles, e.g., a dilutant. The fractionator 700 comprises a pinched segment 730. The output of the pinched segment 730 is connected to a broadened segment 740 and a detection line 750. An enlarged view 760 shows how microplastics of sizes 780 and 770 are separated from the output of the pinched segment 730.

[0088] FIG. 8 shows an experimental example of the fractionator 810 and separated microplastics 820.

[0089] FIG. 9 shows an enlarged view 900 of the output from the fractionator 810. Microplastics of a size 910 follow a path separated from smaller microplastics 920.

[0090] FIG. 10 shows an example of the concentrator 1000. The concentrator comprises an input central port 1010, a first output port 1020 for outputting the concentrated microplastics, and a second output port 1030 for outputting a continuous phase (that is, fluid) without microplastics (in other words, filtered out). The figure also shows an enlarged view of the concentrator. The concentrator comprises a channel having an inner wall. The inner wall has fins 1060 disposed tangentially, thereby forming a directional tangential filter, that is, a physical barrier that separates microplastics from the continuous phase, that is the fluid following the flow 1040. The output 1020 may provide the concentrated microplastics to the input of the separator 311 321, 331, 341,710, 1110.

[0091] FIG. 11 shows an example of the device 1100 for analyzing samples of fluids containing microplastics. The device comprises the separator 1110, a first microfluidic slide 1120 and a second microfluidic slide 1130. The first microfluidic slide faces a first camera 1140, the second microfluidic slide 1130 faces a second camera 1130.

[0092] FIG. 12 illustrates transversal views of a first microfluidic slidel210 as viewed by a respective camera, the transversal view thereby forms a first observation chamber and second microfluidic slide 1220 as viewed by other camera, the transversal view thereby forms a second observation chamber. The first microfluidic slide observation chamber 1210 is of low magnification. The first observation chamber 1210 has a field of view 1211 of 1380x1035 micrometers. The first observation chamber may comprise a channel having a first depth D1, where 500 pm >D > 400 pm, that is, a depth configured for allowing the flow of microplastics with a maximum size of 400 pm . The first observation chamber also comprises a channel having second depth£)2, where 400 > D2>110 pm, that is, a depth configured for allowing the flow of microplastics with a maximum size of approximately 80 pm. The second observation chamber is of high magnification. The second observation chamber 1220 has a field of view 1221 of 455x340 micrometers. .The second observation channel 1220 has a third depth D3where 10 pm < D3< 100 pm, that is, a depth configured for allowing the flow of smaller microplastics, e.g., 29 micrometers or less.

[0093] FIG. 13 illustrates alignments 1310, 1320 of the two channels. In a first alignment 1310, the two channels 1311 and 1312 are centered on a same reference plane 1313 so that microplastics are centered on that reference plane, so that images are sharp as the microplastics.

[0094] In a second alignment 1320, if the two channels 1321 and 1322 are not centered on the same reference plane 1323, the microplastics may be confined in the center of each channel, but not on the focal plane of the microscope objective. The microplastics in the right channel may be blurred.

[0095] FIG. 14 illustrates an example 1400 of a microfluidic slide 1410 arranged to be facing a respective camera 1440. The microfluidic slide 1410 is arranged across an optical path 1420. The optical path 1420 comprises two crossed polarizers 1430. In other words, the two crossed polarizers 1430 are placed between the camera 1440 and the microfluidic slide 1410 and along the optical path 1420

[0096] FIG. 15 illustrates an example 1500 of a microfluidic slide 1510 arranged to be facing a respective camera 1540. The microfluidic slide 1510 is arranged across an optical path 1520. The optical path 1520 comprises an infrared detector and a monochromatic filter 1530. In other words, the infrared detector and the monochromatic filter 1530 are placed between the camera 1540 and the microfluidic slide 1510 and along the optical path 1520.

[0097] FIG. 16 illustrates an example 1600 of a microfluidic slide 1610 arranged to be facing a respective camera 1640. The microfluidic slide 1610 is arranged across an optical path 1620. The optical path 1620 comprises an arrangement of a phase contrast microscope 1630. In other words, the phase contrast microscope 1630 is placed between the camera 1640 and the microfluidic slide 1610 and along the optical path 1620.

Claims

CLAIMS1. A method for analyzing samples of fluids containing microplastics, the method comprising:- inputting (S10) a sample of fluid into a liquid input channel;- dividing (S20) the sample of fluid from the liquid input channel into at least two branch channels, the microplastics being distributed among the branch channels based on their size;- by each branch channel, conveying (S30) the sample of fluid to a respective microfluidic slide arranged facing a respective camera;- capturing (S40) with each respective camera a respective flux of images; and- processing (S50) each respective flux of images to analyze the sample of fluid.

2. The method of claim 1, wherein dividing (S20) the sample of fluid from the liquid input channel into at least two branch channels is performed according to at least two consecutive ranges of microplastics sizes.

3. The method of claim 2, wherein at least one range of the at least two consecutive ranges of microplastics sizes has an upper boundary lower than or equal to 100 micrometers, and at least one other range has a lower boundary higher than or equal to the upper boundary of the at least one range and an upper boundary higher than 100 micrometers.

4. The method of any one of claims 1 to 3, wherein each respective camera is a digital camera, and wherein the resolution of the capturing increases as the size of the microplastics decreases per branch channel.

5. The method of claim 4, wherein the resolution of the capturing is below 8.5 pixels per micrometer for at least one respective camera and / or above 8.625 pixels per micrometer for at least one respective camera.

6. The method of any one of claims 1 to 5, wherein:- at least one optical path in front of the respective camera comprises two crossed polarizers;- at least one optical path in front of the respective camera comprises an infrared detector and / or a monochromatic filter; and / or- at least one optical path in front of the respective camera comprises an arrangement of a phase contrast microscope.

7. The method of any one of claims 1 to 6, wherein processing (S50) each respective flux of images comprises determining one or more portions of each image comprising a respective class of microplastic on the flux.

8. The method of claim 7, further comprising estimating, on each respective flux of images, one or more physical properties of the microplastics on the fluid sample, among:- a quantity of the microplastics;- a shape of the microplastics such as a spheric shape;- a volume of the microplastics;- a distribution of the size of the microplastics;- a rugosity of the microplastics; and / or- a crystallinity of the microplastics.

9. The method of claim 8, wherein estimating the physical properties of the microplastics comprises applying one or more neural networks on each respective flux of images, the one or more neural networks being configured to determine the physical properties of the microplastics on the respective flux of images.

10. The method of any one of claims 1 to 9, wherein processing (S50) each respective flux of images comprises transmitting at least one respective flux of images.

11. The method of any one of claims 1 to 10, further comprising concentrating microplastics in the sample of fluid from the input channel before dividing the sample of fluid.

12. The method of any one of claims 1 to 11, wherein the dividing (S20) of the sample of fluid from the liquid input channel is performed by centrifugal force or fractionation.

13. The method of claim 12, wherein the centrifugation is performed by inputting the sample of liquid into a spiral centrifugal separator.

14. The method of any one of claims 1 to 13, wherein the at least two channels are aligned on a same reference plane.

15. The method of any one of claims 1 to 14, wherein the sample of fluid is of a compost liquid or of a methanization digestate.

16. A device configured for analyzing samples of fluids containing microplastics according to the method of any one of claims 1 to 15, comprising:-the liquid input channel (210);-a separator (220) configured for performing the dividing of the sample of fluid;- the at least two branch channels (231, 232) and each respective microfluidic slide (241, 242);- the at least two cameras (251, 252).

17. The device of claim 17, further comprising a concentrator configured for concentrating the sample of fluid.18 The device of claim 16 or 17, wherein the at least two channels are aligned on a same reference plane.

19. The device of any one of claims 16 to 18, wherein the separator is a spiral centrifugal separator.

Images