Device for detecting bacteria in a sample

Abstract

The invention relates to a device for detecting bacteria in a sample, the device comprising an optical unit (4) arranged to produce sets of multispectral images of portions of a sample, these being obtained by means of illumination eliciting an autofluorescence response and by means of illumination eliciting a reflectance response. These sets of multispectral images are fed to an analyser (6) comprising a classifier that produces an image, each pixel of which is drawn from an identical pixel in the images of the set of multispectral images and is associated with an element identifier indicating whether this pixel is associated with a bacterium or with another element. This image is then used by a computer to determine bacteria detection data.

Description

Device for detecting bacteria in a sample

[0001] The invention relates to the field of microbiology and in particular the detection of bacteria in samples.

[0002] In the agri-food sector, bacterial detection is a major challenge. All stakeholders are subject to extremely rigorous hygiene standards to prevent the sale of food containing bacteria, regardless of its level of processing. Similar issues exist in the pharmaceutical and cosmetic industries.

[0003] Indeed, when the presence of bacteria is detected too late, it is common for many consumers to suffer the consequences, which can range from more or less significant digestive discomfort to poisoning and / or infections that can cause death. A. State of the art

[0004] The detection of the presence of bacteria relies, in the state of the art, on taking samples at regular intervals on the production lines, and on counting the bacteria present in the samples after cultivation.

[0005] This conventional approach involves cultivating bacteria for several days in a nutrient medium, followed by manual colony counting to determine the number of bacteria, usually measured in colony-forming units per milliliter (CFU / mL). However, while considered the gold standard in microbiology, this method has several drawbacks, notably the significant time required to obtain results, which can range from two to five days depending on the specific microorganisms being quantified. This duration is even greater for sterility tests, which can take up to 15 days in the pharmaceutical and cosmetic industries.

[0006] In recent years, alternative methods have been explored to try to replace the Petri dish culture technique: polymerase chain reaction (PCR), most probable number approach, ATP measurement, flow or solid phase cytometry, microscopy-based methods, methods based on the analysis of hyperspectral measurements, or the use of excitation and emission matrices, etc. A.1 PCR Methods

[0007] Among these exploratory methods, PCR offers the fastest results, generally in less than an hour. However, PCR has limitations because it can only detect specific genetic sequences (such as the 16S rRNA gene), as demonstrated by the QuickScan quantitative test (registered trademark). This gene is not present in all bacterial species, which can introduce bias. Generally speaking, each PCR test focuses on a specific genetic sequence, making these tests insufficiently "generalist." Furthermore, the use of reagents in these methods makes them expensive and less accessible for routine use compared to other methods. A.2 "Most probable number" method

[0008] The most probable number approach is a culture-based approach. BioMérieux has optimized this approach using microfluidics and combining it with a fluorescent dye to develop a solution called Tempo (registered trademark). To test the suspension, it is loaded and incubated in a microchamber with a culture medium coupled to a fluorescent dye. The limit of detection of this method ("Limit of Detection" or "LOD" in English) is 1 CFU / mL (see the article by Cayer et al. "Evaluation of the Tempo® System: Improving the Microbiological Quality Monitoring of Human Milk", Frontiers in Pediatrics, Volume 8, 2020, https: / doi.org / 10.3389 / fped.2020.00494) and a dynamic range of 3.7 log, allowing the detection of the total number of viable bacteria ("Total Viable Count" or "TVC" in English) between 24 and 48 hours.

[0009] This detection time is too long and does not constitute a sufficient improvement compared to Petri dish culture. A.3 Methods based on ATP measurement

[0010] ATP measurement amplifies the signal generated by the presence of adenosine triphosphate (ATP). This provides rapid results in seconds, but the method is highly sensitive to organic contamination from chemicals or detergents, which can lead to numerous false positives. Because ATP measurement is easy to deploy in the field and relatively inexpensive, it is frequently used, but it remains unsuitable as a reliable, single measure of bacterial presence. A.4 Flow cytometry methods

[0011] This method relies on illuminating a capillary containing the sample with a laser. It can use dyes to label bacteria, and an optical sensor is used to collect the photons emitted by the labeled bacteria.

[0012] Flow cytometry still has significant limitations. First, it counts both live and dead cells and requires the analysis of large sample volumes, which can take several hours and is too time-consuming. Furthermore, it is only compatible with liquid matrices, and the equipment is expensive (several hundred thousand euros per machine). A.5 Solid-phase cytometry methods

[0013] The solution to be analyzed is filtered through a membrane selected to retain particles at least the size of a bacterium (e.g., 0.4 µm). This is combined with staining the bacteria using a dye. The membrane is then scanned with a laser to excite the fluorochrome, and the number of bacteria is counted. As an alternative to laser detection, epifluorescence imaging can be performed.

[0014] Various implementations based on this principle have been developed by companies such as Red Berries (see for example patent application EP3861125A1), Microbs, Innosieve Diagnostics (see for example patent application EP2440331A1) or Mibic.

[0015] These methods require a filterable solution and specific reagents, making them expensive. Furthermore, after the detection and counting of microorganisms, the reagents used can have toxic effects on cellular metabolism, potentially hindering subsequent culture and identification of these microorganisms.

[0016] Finally, these methods require significant pretreatment steps and incubation periods of at least 20 minutes, and they exhibit a high level of false negatives due to dust or impurities affecting fluorescence, or dye absorption. A.6 Optical Microscopy Methods

[0017] Optical microscopy methods have attracted interest for the identification and enumeration of bacteria due to their potential for rapid and non-destructive detection that requires minimal sample preparation.

[0018] Characterization and enumeration of bacteria are required for a wide range of concentrations and different types of samples, which significantly complicates these methods.

[0019] Current optical methods applied to the detection of bacteria only work with a highly concentrated suspension, making them useless in practical applications. A.7 Epifluorescence Microscopy Methods

[0020] Fluorescence microscopy is an imaging technique that occurs when a fluorophore interacts with light of chosen energy, leading to the excitation of a fluorophore from the ground state to a higher energy state, before returning to a lower energy level by emitting a photon.

[0021] Epifluorescence microscopy typically uses a single wavelength band and lacks the ability to differentiate bacteria from other objects without the use of optimized bandpass filters. This is because the multi-band approach requires switching between bands using a dichroic wheel, which is a slow process.

[0022] These methods also require sample preparation with optical dyes, such as DAPI, for counting bacteria (see the article by T. Muthukrishna et al. "Evaluating the Reliability of Counting Bacteria Using Epifluorescence Microscopy", Journal of Marine Science and Engineering. 2017; 5(1):4, https: / doi.org / 10.3390 / jmse5010004).

[0023] The main cases where fluorescent dyes are unnecessary are known in ophthalmology, where autofluorescence of ocular cells can provide valuable information on patient-specific pathologies (see the article by A. Gakamsky et al., "Tryptophan and Non-Tryptophan Fluorescence of the Eye Lens Proteins Provides Diagnostics of Cataract at the Molecular Level," Scientific Reports 7, 40375 (2017), https: / / doi.org / 10.1038 / srep40375). In such cases, fluorescence microscopy can provide a rapid and reliable detection system using intrinsic fluorescence.

[0024] However, there is currently no optimized system that can distinguish objects from bacteria at this level of precision, with a proven application in microbiology. A.8 Hyperspectral Imaging Methods

[0025] Hyperspectral imaging solutions have been studied for the quantification of bacteria. These solutions generally use white light illumination in the VNIR spectrum ("Visible and Near-infrared", between 400nm and 1100nm), combined with machine learning ("ML" or "ML" models) or deep learning ("DL" models).

[0026] These models are monolithic, meaning that feature extraction is performed to form the input for the machine learning (ML), or a digital learning (DL) model is used to process hyperspectral images that may have undergone preprocessing. This makes it impossible to determine which part of the images is associated with a bacterium, thus limiting validation possibilities during training, or requiring imaging times incompatible with industrial implementation.

[0027] Two main approaches are competing: - one approach is based on measuring adulteration. For example, in the article by Zhenget et al., "ANondestructiveReal-TimeDetectionMethod of Total Viable Count in Pork by Hyperspectral Imaging Technique," Applied Sciences 7, no. 3: 213, 2017, https: / / doi.org / 10.3390 / app7030213, a model that correlates the spectral signature of pork with the total viable cell count (TVC) is trained to rapidly quantify the TVC in a piece of pork. However, this measurement primarily detects product adulteration rather than the actual quantity of bacteria. It is effective for matrices damaged by bacteria but is not suitable for low bacterial loads (e.g., less than 100 CFU / mL) or matrices such as environmental samples and milk.One approach is based on sorting bacteria according to their spectral signature on a microscope slide, as described in the article by Michael et al., "Hyperspectral imaging of common foodborne pathogens for rapid identification and differentiation," Food Science & Nutrition, 2019, 7: 2716–2725, https: / / doi.org / 10.1002 / fsn3.1131. This approach uses white light and transmittance microscopy to differentiate Salmonella spp., E. coli, and Cronobacter sakazakii. While these hyperspectral solutions allow for differentiation between bacterial species, they require significant sample preparation and high-resolution optics. Furthermore, they do not allow for the direct calculation of colony-forming units per milliliter (CFU / mL).

[0028] Beyond the problems mentioned above, hyperspectral imaging-based methods suffer from acquisition speed issues with existing hyperspectral cameras, which make them unusable in contexts of detecting low levels of bacteria. A.9 Multispectral Imaging Methods

[0029] Multispectral imaging solutions have been studied for the investigation of sample adulteration. Generally speaking, methods for studying adulteration are unsuitable for the problem addressed by this invention.

[0030] As an example, the article by Spyrelli et al., "Implementation of Multispectral Imaging (MSI) for Microbiological Quality Assessment of Poultry Products," Microorganisms 2020, 8, 552. https: / / doi.org / 10.3390 / microorganisms8040552, describes the capture of reflectance at multiple wavelengths between 405 nm and 970 nm. These measurements are processed to form input vectors for a machine learning process, which returns a TVC.

[0031] A similar approach is presented in the article by Fengou et al. "Detection of Meat Adulteration Using Spectroscopy-Based Sensors", Foods 2021, 10, 861. https: / / doi.org / 10.3390 / foods10040861. In this article, the performance of adulteration detection is validated against an ultraviolet fluorescence approach.

[0032] All these approaches have the drawback of being unable to detect low levels of bacteria. Furthermore, they are based solely on reflectance. By their very nature, they are unsuitable for detecting single bacteria, both due to the resolution of the processed image and the signal used.

[0033] A.10 Excitation and Emission Matrix Methods

[0034] Excitation and emission matrices (EEMs) are used for multi-component analysis and provide a molecular fingerprint for different types of samples. EEM spectroscopy has been used in various applications, particularly in the wine and oil industry, to predict taste by determining specific compounds of interest (see the article by B. Quintanilla-Casa et al., "Using fluorescence excitation-emission matrices to predict bitterness and pungency of virgin olive oil: Afeasibility study", Food Chemistry, Volume 395, 30 November 2022, 133602, https: / / doi.org / 10.1016 / j.foodchem.2022.133602).

[0035] Compared to conventional fluorometry methods, EEM spectroscopy has the advantage of not depending on concentration and of accurately identifying certain samples.

[0036] Some applications in microbiology have enabled the detection of bacteria in water at levels as low as 10 CFU / mL (see, for example, the article by Nakaret et al., "Quantification of bacteria in water using PLS analysis of emission spectra of fluorescence and excitation-emission matrices", Water Research, Volume 169, 2020, 115197, https: / / doi.org / 10.1016 / j.waters.2019.115197). However, in the case of food, these methods are only capable of detecting adulteration, and in particular, CFU / mL levels below 100 CFU / mL.

[0037] Finally, they require tryptophan calibration to obtain accurate results and the use of expensive equipment such as a fluorometer. B. Summary of the invention

[0038] In general, the Petri box method and the exploratory methods described above are unsatisfactory for one or more of the following reasons: implementation time, complexity, cost, quality of detection, detection threshold, ability to determine the TVC.

[0039] The invention improves the situation. To this end, it proposes, according to a first aspect, a device for detecting bacteria in a sample, comprising an optical bench including an illumination block comprising an optical source and a distribution optic for illuminating a sample, and a measurement block including a collection optic and a measurement optic, the collection optic being arranged to redirect a measurement radiation from the sample in response to illumination towards the measurement optic.The optical source is arranged to emit a first radiation with a wavelength approximately equal to 365nm, and a second radiation with wavelengths in a range between 375nm and 1000nm for the distribution optics to illuminate the sample. The distribution optics include a magnifying objective. The collection optics further include one or more splitters arranged to separate the measurement radiation into at least a first beam and a second beam. The first beam presents a portion of the measurement radiation having one or more wavelengths in a first range between 400nm and 414nm, and the second beam presents a portion of the measurement radiation having one or more wavelengths in a second range between 414nm and 490nm.The measurement optics includes at least a first optical sensor for the first beam, and a second optical sensor for the second beam arranged to return an image in which each pixel is associated with an area of ​​the sample illuminated by the magnification objective and with a measured light amplitude value respectively, the first optical sensor being arranged in a position suitable for the focal plane of the first measurement beam, and the second optical sensor being arranged in a position suitable for the focal plane of the second measurement beam.

[0040] The device is arranged to control a relative displacement of the sample with respect to the illumination block in order to illuminate successive portions of the sample with the first radiation and with the second radiation, the illumination of a given portion by the first radiation on the one hand and by the second radiation on the other hand being carried out without overlap, and to obtain a set of multispectral images for each of the successive portions of the sample.

[0041] The device further includes an analyzer comprising a classifier and a computer, the classifier comprising a deep learning neural network arranged to receive the set of multispectral images associated with a portion of the sample and to return an image in which each pixel is associated with an element type identifier chosen from a list comprising at least two elements, one of which designates a bacterium, and the computer being arranged to determine bacterium detection data from the images returned by the deep learning neural network.

[0042] This device is particularly advantageous because it provides a device that allows for the detection of bacteria at very low detection levels, and in times compatible with an industrial production line.

[0043] According to various embodiments, this device may have one or more of the following characteristics: - the optical source comprises two separate sources, one of which is arranged to emit the first radiation and the other is arranged to emit the second radiation, or a single source capable of emitting both the first and second radiations and which is arranged to emit the first and second radiations separately either by controlling its emission or by combining it with a controlled optical filter, - said one or more beam splitters of the collection optics are further arranged to separate the measurement radiation into a third beam distinct from the first and second beams, this third beam having a portion of the measurement radiation having one or more wavelengths in a third range between 470nm and 1000nm,The measurement optics include a third optical sensor to measure the third beam; the analyzer is further arranged to receive the measurements from the third optical sensor along with the measurements from the first and second optical sensors to derive the bacteria detection data; the optical bench further includes an autofocus system positioned between the optical source and the magnifying objective; the magnifying objective and the autofocus system are included in the distribution optics and the collection optics; the deep learning neural network of the classifier is a convolutional neural network, a transformer network, or a combination of both; the computer is arranged to determine a ratio between the area occupied by pixels associated with a bacteria identifier and a reference area, and to calculate the bacteria detection data from this ratio.- The classifier is arranged to associate a pixel with an identifier of a type of element chosen from a list including a bacterium, a matrix element, an air bubble, a membrane element, or a foreign element; and - the calculator is arranged to determine the reference surface from the pixels whose identifiers are associated with a membrane element.

[0044] According to a second aspect, the invention also relates to a method for detecting bacteria in a sample comprising the following operations: a) determining one or more sets of multispectral images, each set of multispectral images of a given sample portion comprising images obtained by measuring, in a first wavelength range between 400nm and 414nm and in a second range between 414nm and 490nm, an autofluorescence beam emitted by the given sample portion when illuminated by radiation with a wavelength substantially equal to 365nm, and on the other hand a reflectance beam of the given sample portion when illuminated by radiation with wavelengths in a range between 375nm and 1000nm, the illuminations giving rise to the autofluorescence beam and the reflectance beam being non-overlapping,b) provide each set of multispectral images to a deep learning neural network to produce an image in which each pixel is associated with an element type identifier chosen from a list comprising at least two elements, one of which designates a bacterium, etc.) calculate bacteria detection data from at least some of the images from operation b).

[0045] According to various embodiments, this method may have one or more of the following characteristics: - operation a) comprises determining a set of multispectral images for a set of non-overlapping sample portions whose union covers the entire surface of the sample; - operation a) is performed for a sample portion, then operation b) is performed for that sample portion, and operations a) and b) are repeated until a measurement end condition, including exceeding a threshold or determining a set of multispectral images for the entire sample, is met; - operation a) is performed for a chosen number of sample portions, then a necessary number of sample portions is determined by applying operation b) to the resulting sets of multispectral images.Then operations a) and b) are performed with a stopping condition that takes into account the number of sample portions required; the number of sample portions required is re-evaluated each time operations a) and b) are performed, and in which the stopping condition takes into account the re-evaluated number of sample portions required; and operation c) includes determining a ratio between the area occupied by pixels associated with a bacterial identifier and a reference area, and calculating the bacterial detection data from this ratio.

[0046] Other features and advantages of the invention will become more apparent from the following description, drawn from illustrative and non-limiting examples taken from the drawings in which: - a schematic view of a device according to the invention, - a representation of an embodiment of the optical bench, - a representation of an example of an image obtained at the output of the optical block according to two respective measurement channels, - a generic view of an analyzer implemented in the embodiment, - a block diagram of an implementation of the analyzer, - a representation of an example of an image obtained at the output of the classifier, - a diagram illustrating an operating method of the analyzer according to an embodiment in operational mode, - results of bacterial detection obtained with the matrix shown in Figures 3 and 6 and a transformer as a classifier.- lare represents images similar to those in Figures 3 and 6 obtained with another matrix, and - lare represents results of bacterial detection obtained with the original matrix and a transformer as a classifier.

[0047] The drawings and description below contain, for the most part, elements of a definite nature. They can therefore not only serve to better explain the present invention, but also contribute to its definition, if necessary. C. Examples of implementation of the invention

[0048] Lare represents a schematic view of an embodiment of a device 2 according to the invention.

[0049] Device 2 includes an optical bench 4 and an analyzer 6. The optical bench 4 includes an illumination block 8 and a measurement block 10.

[0050] The optical bench 4 is configured to perform multispectral imaging on a sample 12 obtained by grinding and depositing a sample 14 onto a membrane, for which a bacterial detection value 16 is sought. In the embodiment described here, the sample can be wheat, flour, or even a liquid reflux (for example, to test for the presence of bacteria in water). Hereafter, the sample 14 may also be referred to as the "matrix".

[0051] At the output of the optical bench 4, a plurality of multispectral images 18 (hereafter referred to as the multispectral image set) are transmitted to the analyzer 6 to determine bacterial detection data 16, which may be a bacterial quantity value. In some embodiments, the bacterial quantity value may be a number of colony-forming units per milliliter (CFU / mL). In other embodiments, this bacterial quantity value may be a Boolean value indicating whether a threshold of colony-forming units per milliliter (CFU / mL) has been exceeded. Although reference is made here to colony-forming units per milliliter, any other relevant CVT measurement may be used, such as colony-forming units per gram, colony-forming units per square centimeter, etc.As will be seen below, in some embodiments, the bacteria detection data 16 may be data enabling the determination of a quantity value of bacteria by a treatment or a normalization.

[0052] In this embodiment, the analyzer 6 implements a machine learning model (ML). For this reason, dashed arrows are used to show the elements for training the analyzer 6, via a Petri dish 20 in which bacteria from sample 14 are conventionally cultured and then counted to provide data for training the analyzer 6. Once training is complete, only the solid arrows are used by device 2 to calculate a detection value for bacteria 16. The operation of the analyzer 6, its training, and its embodiment variants will be described further in the embodiments shown in Figures 4 to 7. C.1 Optical bench

[0053] Laillustre un mode d’incarnation du bench optique 4.

[0054] Optical bench 4 includes an illumination block 22 and a measurement block 24.

[0055] The illumination unit includes an optical source 26 and a distribution optic 28 for illuminating the sample 12. In the example described here, the distribution optic 28 includes a magnifying lens 30 and an autofocus 32. As will be seen below, the magnifying lens 30 and the autofocus 32 also serve as a collection optic for performing measurements. In some embodiments, the autofocus 32 may be omitted.

[0056] In the example described here, the optical source 26 is a polychromatic light source. Generally, the optical source 26 can exhibit an emission spectrum encompassing wavelengths between 300 nm and 1200 nm. The reason for this choice is that the Applicant has discovered that it is possible, using a multispectral approach, to study the spectrum in autofluorescence (also called "intrinsic fluorescence") and reflectance in order to detect the presence of bacteria in matrices such as water, flour, wheat, or other products to be tested.

[0057] This is highly unusual. While reflectance is a useful signal for detecting adulteration at the macroscopic level, it is not relevant when detection levels are much lower, such as at the level of individual bacteria. Traditionally, reflectance is ignored in microscopy in favor of sample transmittance measured by phase-contrast imaging. However, studying transmittance using phase-contrast imaging is impossible in this case.

[0058] Against all expectations, the Applicant discovered that not only do bacteria exhibit a useful signal in reflectance, but also that, unlike all existing approaches, the combination of autofluorescence and reflectance makes it possible to detect individual bacteria and to exclude foreign bodies that could distort the measurements.

[0059] The invention therefore combines the non-classical use of reflectance with autofluorescence, whereas the only known methods – which do not address the problem targeted by the invention – at best use only one type of signal.

[0060] In the example described here, the optical source 26 is a light-emitting diode (LED) source. Alternatively, any optical source capable of emitting radiation that induces autofluorescence within the range mentioned above and reflectance will be suitable, provided that the distribution optics 28 allow the radiation to be concentrated onto the sample 12 to obtain a suitable size for imaging. Thus, the optical source 26 could be a laser source or any other suitable light source.

[0061] In particular, the optical source 26 is arranged so as to be able to deliver the radiation causing autofluorescence and the radiation causing reflectance in a non-overlapping manner in time.

[0062] Put another way, the optical source 26 can be controlled to first illuminate a given portion of the sample with radiation that induces autofluorescence, and then to illuminate that given portion of the sample with radiation that induces reflectance.

[0063] The Applicant discovered that this is particularly advantageous because it allows for better measurement of autofluorescence and reflectance. Indeed, since these two phenomena are very different, this temporal separation makes it advantageous to be able to emit the most appropriate excitation radiation without causing masking or noise.

[0064] To achieve this, the optical source 26 can be split, meaning it comprises, for example, two LED sources, each specific to a particular radiation pattern, with, for instance, a dichroic mirror allowing one of the radiation patterns to pass through. This maintains collimated illumination beams that consistently illuminate the same area. Alternatively, the optical source 26 could be a broad-spectrum LED or laser source controlled to emit selectively according to the radiation that induces autofluorescence or the radiation that induces reflectance. Also alternatively, instead of controlling the emission spectrum, the optical source 26 can include at its output a filter or a set of filters controlled so that a broadband radiation is reduced to either the radiation that induces autofluorescence or the radiation that induces reflectance, as required.

[0065] Since autofluorescence is a low-intensity phenomenon, the optical source 26 is configured to emit autofluorescence radiation with a wavelength approximately equal to 365 nm, that is, centered on the 365 nm wavelength, which corresponds to the autofluorescence excitation of NADH (the reduced form of nicotinamide adenine dinucleotide). Thus, illumination radiation does not interfere with the autofluorescence measurement. The situation is different with regard to reflectance. Indeed, reflectance can be measured over a broad spectrum, from 300 nm to over 1000 nm. However, in order to limit the risk of photobleaching, the optical source is configured to emit reflectance radiation with a wavelength between 375 nm and 1000 nm. Preferably, the reflectance radiation can have a wavelength between 400nm and 1000nm.

[0066] Furthermore, the multispectral approach is advantageous compared to a hyperspectral approach because it allows for a faster imaging rate for a given resolution. More specifically, the multispectral approach, with its faster acquisition rate, allows for higher-resolution images to be obtained within a constant time budget, and therefore for the detection of higher colony-forming unit (CFU) counts per milliliter (CFU / mL). Thus, the 30x magnification objective in the example described here is a 20x magnification objective, which allows for images of approximately 0.5 mm x 0.5 mm.

[0067] Alternatively, other lenses could be used. Generally, the combination of the lens and cameras used results in a resolution of approximately 84 pixels for a bacterium. In general, any combination that achieves a resolution between 1 pixel and 1500 pixels (representing a magnification of approximately 100X) allows for detection with a quality and imaging time that remain compatible with the desired performance, namely the ability to perform a reliable measurement in less than one hour.

[0068] Sample 12 is received on a movable platform 33 of the optical block 4. The movable platform 33 allows the sample 12 to be moved in an XY plane relative to the measuring block 22 and the measuring block 24, thus illuminating and measuring a different portion of the sample 12 each time. The autofocus 32 can also be controlled based on the movement of the platform 33 to perform the Z-axis focus adjustment. As will be seen below, the movement of the movable platform 33 can be controlled by the analyzer 6. Alternatively, the movable platform 33 can be controlled by a control module of the optical bench 4. Also alternatively, the objective lens 30 could be movable relative to the sample 12.

[0069] As mentioned above, the sample is exposed to the radiation that induces autofluorescence separately from its exposure to the radiation that induces reflectance. When each portion of the sample is first exposed to one type of radiation and then to the other, the acquisition time is not significantly affected because the illumination-measurement time for reflectance is approximately 5 ms, while the illumination-measurement time for autofluorescence is approximately 500 ms. Consequently, decoupling the illuminations results in a difference of only about 1%. The illumination-measurement time for autofluorescence could be shortened with a higher illumination power.

[0070] It would also be possible to scan the entire sample to perform one type of measurement (i.e., reflectance or autofluorescence), and then scan the entire sample again to perform the other type of measurement. In this case, some processing may be required to ensure proper calibration of the measurements for each portion, and there would be a slightly greater time loss because the sample is scanned twice, meaning that the travel times, on the order of 20 ms between each portion, are doubled.

[0071] The measurement block 6 includes a collection optic 34 and a measurement optic 36. The collection optic 34 has the role of collecting the autofluorescence and reflectance radiation from the sample 12 and redirecting it to the measurement optic 36.

[0072] As mentioned above, the optical bench 4 is designed in a compact manner, and the magnifying objective 30 and the autofocus 32 serve both the distribution optics 28 and the collection optics 34. Thus, the light beam from the light source 26 is focused and its size is reduced on the sample 12 by the objective 30 and the autofocus 32, and the radiation which comes from the sample 12 passes back through the objective 30 and the autofocus 32 which restores its initial size for the optical measurement.

[0073] In the example described here, the collection optics 34 further includes three separators 38 and three lenses 39 which each redirect a portion of the radiation from the sample 12 to the measuring optics 36.

[0074] Thus, a first separator 38 comprises a dichroic mirror that isolates a portion of the radiation from sample 12 with wavelengths between 400 nm and 425 nm, and preferably between 400 nm and 414 nm, to form a first beam. This dichroic mirror directs the remainder of the radiation from sample 12 towards a second separator 38.

[0075] Similarly, the second beam splitter 38 includes a dichroic mirror 38 which isolates a portion of the radiation from sample 12 with wavelengths between 414 nm and 490 nm, and preferably between 414 nm and 470 nm, to form a second beam. This dichroic mirror directs the remaining radiation from sample 12 to a third beam splitter 38.

[0076] Finally, the third beam splitter 38 collects the remaining radiation, which therefore has wavelengths beyond 470 nm, to form a third beam. Alternatively, the third beam splitter 38 can include a dichroic mirror that selects a portion of the radiation from sample 12 with wavelengths preferably between 470 nm and 520 nm to form the third beam.

[0077] The first beam, the second beam and the third beam are each received by a respective lens 39 of the collection optics 34.

[0078] The measuring optics 36 in the example described here includes a first camera 40, a second camera 42 and a third camera 44. Each lens 39 redirects respectively the first beam to the first camera 40, the second beam to the second camera 42 and the third beam to the third camera 44.

[0079] The first camera 40, the second camera 42 and the third camera 44 are arranged to measure spectral bands which correspond to the wavelength range of respectively the first beam, the second beam and the third beam.

[0080] In some embodiments, the third separator 38 may be omitted, and the analyzer 6 will operate on measurements of the first and second beams only.

[0081] In the example described here, the first camera 40, the second camera 42, and the third camera 44 are Phoenix 8.1 MP Model (IMX566), and the resulting image from each shot has a resolution of 1420 x 1420, as shown in the figure. Other camera models or optical sensors can, of course, be used, as well as other resolutions for the corresponding images. However, it must be taken into account that the diffraction limit sets a maximum usable resolution.

[0082] In each image, each pixel corresponds to an illuminated area of ​​the sample 12. The value associated with each pixel is the sum of the intensities measured for the wavelengths of the camera at that location. To represent the value associated with each pixel, a color can be used that depends on the sum of the measured intensities.

[0083] Thus, in operation, the optical source 26 will successively illuminate portions of the sample 12. The illumination of each portion of the sample 12 is carried out in two stages.

[0084] A first illumination of approximately 500 ms at a wavelength of 365 nm is used to elicit the autofluorescence radiation from sample 12, which is then measured, yielding images of the sample portion similar to the one on the right. Subsequently, a second illumination of approximately 5 ms is performed at approximately 405 nm, and the reflectance radiation from sample 12 is measured, resulting in images of the sample portion similar to the one on the left.

[0085] Because autofluorescence and reflectance are measured temporally disjointly, it is possible to dedicate certain cameras to either reflectance radiation or autofluorescence radiation. This allows the cameras to be positioned so that they are in the focal plane of the autofluorescence radiation on one hand and in the focal plane of the reflectance radiation on the other. As a result, the images captured from these radiations are more precise and have better resolution, thus improving the quality of the detection.

[0086] Conversely, if all cameras are used to measure both autofluorescence and reflectance, a compromise must be made in their placement relative to the focal plane of the autofluorescence radiation on the one hand, and the focal plane of the reflectance radiation on the other. This compromise will necessarily impact the measurement quality, either for autofluorescence, reflectance, or both.

[0087] The images obtained from the cameras for each given portion form, for that given portion, a multispectral image set used by analyzer 6 as described below. In the case described here, the multispectral image set is a triplet comprising one reflectance image and two autofluorescence images.

[0088] Alternatively, more than three cameras can be used. For example, one camera can be dedicated to reflectance measurement, and five cameras can be dedicated to autofluorescence. In these configurations, the number of images in the multispectral image set will vary with the number of cameras used. In this case, six-part image sets (one reflectance image and five autofluorescence images) are processed by the analyzer 6 to determine bacteria detection data 16, which then returns a bacteria detection value. C.2 Analyzer

[0089] Analyzer 6 and its processing of measurements from cameras 40 to 44 will now be described with reference to the.

[0090] The analyzer 6 includes a data storage of 100 and a calculator of 102.

[0091] Data storage 100 receives sample image data, deep neural network data, and bacteria detection data.

[0092] Each sample image data is associated with coordinates indicating which part of sample 12 it originates from, as well as camera data indicating which of the first, second, or third beams it is associated with. As explained above, the sample image data comprises pixels, each associated with an intensity value representing the sum of the intensities measured by the camera for the location of sample 12 designated by the coordinates and the pixel in question. Additionally, a supplementary camera could be used to detect the absolute position of sample 12. Alternatively, the sample 12 holder can be adapted for the same purpose.

[0093] Data storage 100 can be any type of data storage suitable for receiving digital data: hard drive, flash memory hard drive, flash memory in any form, RAM, optical disc, locally distributed or cloud-based storage, etc.

[0094] The computer (PC) 102 may include one or more processors (P) 104, a network interface 106 comprising a transmitter (Tx) 108 and a receiver (Rx) 110 to enable the device to transmit and / or receive data to other computing devices connected to a network 112 (for example, an IP network) to which the network interface 106 is connected.

[0095] In some embodiments where the computer 102 includes a programmable processor, a computer program product (CPP) 116 may be provided to implement the phase correction processing. The CPP 116 stores a computer program (CP) 118 that includes computer instructions (CRI) 120. The CPP 116 may be stored on a computer-readable medium (CRM) 122, which may be a non-transitory computer-readable medium, such as magnetic media (hard drive, SSD, magnetic tape, etc.), optical media (CD, DVD, Blu-ray disc, etc.), memory (RAM, flash memory, etc.), distributed or cloud storage, etc.Computer-readable media (CRM) 122 can also be stored in data storage 100.

[0096] The processor(s) (P) 104 can be any processor suitable for the calculations described below. Such a processor can be implemented in any known form, such as a microprocessor for a personal computer, laptop, tablet, or smartphone; a dedicated digital signal processor (DSP); a dedicated FPGA or SoC chip; a computing resource on a grid or in the cloud; a graphics processing unit (GPU) cluster; a microcontroller; or any other form capable of providing the computing power necessary for the implementation described below. One or more of these elements can also be implemented as specialized electronic circuits such as an ASIC. A combination of processor(s) and electronic circuits can also be considered. Processors dedicated to machine learning may also be considered.

[0097] In the exploratory methods described in the introduction, particularly in some of the multispectral or hyperspectral methods, a deep learning neural network is used to directly extract the value of colony-forming units per millilitre (CFU / mL), from a single shot of the sample 12.

[0098] The Applicant explored several ways of processing the sample image data and determined that using a deep-learning neural network on a single image is not desirable.

[0099] On the contrary, the Applicant discovered that, in order to calculate an industrially usable bacterial detection value—that is, on the order of 1 CFU / mL—a high spatial resolution for data collection is desirable, requiring more than 3116 exposures for a sample approximately 2 cm in diameter. Advantageously, concentrating the light from the optical source also increases the surface intensity of illumination and thus the corresponding autofluorescence response.

[0100] It should first be noted that this number of images makes the use of hyperspectral cameras impossible, as the acquisition time for each image is 2.5 minutes. Beyond this, the Applicant realized that the lower the target colony-forming units per milliliter (CFU / mL), the less reliable direct detection by deep learning neural networks becomes. Indeed, with increasing resolution, the information about the presence of bacteria becomes diluted, or even spread across several images, making it difficult for the neural network to learn reliably.

[0101] The Plaintiff thus discovered that it is preferable to carry out the detection in two operations.

[0102] For this reason, as shown on the, the analyzer 6 includes a classifier 50 which includes a deep learning neural network which receives a set of multispectral images as input and returns an image 52 as output, and a computer 54 which receives the image 52 as input and returns the detection data 16.

[0103] It goes without saying that the distinction between classifier 50 and calculator 54 is essentially functional, but that analyzer 6 can be seen as made of a single block.

[0104] Thus, a first detection operation implemented by the classifier 50 consists of associating a label with each pixel of a given area of ​​sample 12, based on all the images concerning that area. Indeed, sample 12 is typically composed of a membrane and the atomized sample 14. Consequently, the optical measurement can correspond, for each pixel, to the following types: membrane material, matrix element, a bacterium, an air bubble, or a foreign element such as dust. In its simplest form, the label can be reduced to "bacteria" or "other element".

[0105] Put another way, each pixel receives a value that indicates whether it is considered that pixel belongs to the membrane of sample 12, to a bacterium, to a matrix element that is the subject of the measurement of bacterium detection, to an air bubble, or to a foreign element.

[0106] Alternatively, the deep learning neural network can return a vector indicating the probability of each pixel belonging to a given category (for example, [0.4; 0.6] would indicate a 40% chance that a given pixel belongs to a bacterium and a 60% chance that it does not, in the case of a binary label described above). In this case, the first modulo 50 can be arranged to determine a unique label for each pixel, either by choosing the label with the highest value, or by partially using the values ​​of neighboring pixels, and so on. Alternatively, the probability of belonging to the "bacteria" category could be taken into account in the subsequent calculation of the bacteria detection data.

[0107] Lare represents an example of image 52 obtained after applying the labels starting from the measurements shown on the.

[0108] The Applicant tested several deep learning neural networks for this task, and found that convolutional neural networks (CNNs), recurrent neural networks (RNNs), and transformers or self-attentive models are useful for performing this first operation.

[0109] To train the deep neural network, multispectral image sets were measured for numerous samples from each matrix, and each pixel in each multispectral image set was annotated with its corresponding label. By each pixel in each multispectral image set, we mean that a label is associated with a pixel coordinate for each image in the multispectral image set corresponding to the same shot.

[0110] The Applicant tested several of these CNN models, including Unet, Unet++, MAnet, Linknet, FPN, PSPNet, PAN, DeepLabV3, and DeepLabV3+. Access to these CNNs is available at https: / / web.archive.org / web / 20231021175139 / https: / / smp.readthedocs.io / en / latest / . Among the transformers, SegViTv2 (see https: / / arxiv.org / pdf / 2306.06289.pdf) and Segmenter (see https: / / huggingface.co / docs / transformers / model_doc / segformer) were also identified. The Applicant also determined that a combination of CNNs, RNNs, and / or transformers can be useful.

[0111] A second part of the detection process is carried out by the computer 54 and consists of quantifying the presence of bacteria based on the images 52 on which the pixels have been labeled. This can be done in various ways.

[0112] According to one embodiment of the invention, the pixel area associated with a bacterium is calculated and accumulated image by image to obtain a total area occupied by the bacteria on the sample 12, which allows for the detection of a TVC value by relating it to the detected membrane area. To determine the TVC from the ratio between the area occupied by the bacteria and the area occupied by the membrane, cultures in Petri dishes were performed to measure the actual TVC for these samples, using standardized sampling quantities 14.

[0113] The determined TVC value obviously depends on the quantity of sample 14 used to prepare sample 12. Therefore, if the quantity of sample 14 is not normalized, then the analyzer 6 can receive as input the value in grams, liters, or square centimeters, allowing the TVC value to be converted to CFU / g, CFU / mL, or CFU / cm². Alternatively, this normalization can be performed later.

[0114] By "frame by frame," we mean the images obtained after processing by the first operation above, each corresponding to a given portion of the sample that was illuminated and whose autofluorescence and reflectance response was measured. As suggested earlier, the cumulative TVC value can also be compared to a threshold, and the operation can stop as soon as the threshold has been exceeded or the entire sample has been illuminated.

[0115] This type of processing is illustrated on the diagram, which shows a loop of operation of device 2.

[0116] This loop begins with an operation 60 in which a set of multispectral images of a portion of the sample 12 is obtained by imaging with the optical bench 4.

[0117] Next, in operation 62, analyzer 6 calculates a ratio of bacteria surface area to membrane surface area, and a bacteria detection value is calculated in operation 64.

[0118] In operation 64, a bacterial detection value can be obtained by determining a sample portion CTV for the multispectral image set from operation 60. This sample portion CTV can then be summed over all sample portions. Alternatively, the bacterial detection value can be re-evaluated for each sample portion; that is, the bacterial surface area to membrane surface area ratio for the current measurement is updated at each iteration of the loop, and the CTV is evaluated from this total ratio (up to the current loop).

[0119] Next, in operation 66, it is determined whether the loop should continue. If a threshold is detected, this involves comparing the result of operation 64 with that threshold. Otherwise, this operation determines whether there is still another portion of the sample to be illuminated. Finally, the loop ends in operation 68 with the sending of the bacteria detection data thus determined.

[0120] Lamontre demonstrates the performance of the measurement carried out by the lamethode loop with water as a matrix and a transformer as a deep learning neural network.

[0121] Lamontre respectively shows the images corresponding to the (left) and the (right) obtained with wheat as the matrix, and lamontre shows the performance of the measurement carried out by this method with wheat as the matrix and a transformer as the deep learning neural network.

[0122] The Applicant also discovered that it is possible to speed up the process by reducing the number of shots taken.

[0123] More specifically, in one particular embodiment, 5 non-overlapping images (for example) are obtained to initialize the measurement loop. Based on these initial values, a number of shots are taken to calculate the TVC value with an accuracy of ±0.5 LOG.

[0124] The table below represents the number of shots as a function of TVC values. TVC Value Shots 0 19 60 118 46 22 123 34 25 26 27 2

[0125] In a first variant, a loop similar to that of the may be implemented, but with a fixed number of shots taken from the table above as a stopping condition.

[0126] In a second variant, the loop can be slightly modified by evaluating the number of shots at each loop iteration based on the evolution of the TVC value of the previous loop, and by defining a loop termination condition based on the comparison between the updated number of shots needed and the number of loops already performed and / or exceeding a TVC value threshold if applicable.

[0127] The Applicant's work suggests that there is no preferred route for taking photographs. Nevertheless, to limit any risk, it is preferable to traverse the sample by selecting photographs that are roughly evenly distributed. For example, it is possible to define non-overlapping zones, take a measurement in each zone, and then repeat the process.

[0128] In other variants, some of the 52 images could be ignored.

[0129] Thus, the device described above makes it possible to carry out a TVC measurement with high precision in very low detection ranges, with a measurement time that is less than 2 hours, and is regularly less than 10 minutes.

[0130] As mentioned above, the Applicant also experimented with the use of transformers as a deep learning neural network. Indeed, despite their qualities demonstrated above, CNNs exhibit narrow effective receptive fields.

[0131] Transformers introduce a non-local self-attention mechanism that can perform operations combining distant pixels in the image. This significantly increases the effective receptive field and improves model performance for a given computing power.

[0132] To validate this theory, different machine learning models were tested on a dataset of 133 images (20 for validation and 113 for training). Within this dataset, 24 samples contained only dust, and 109 samples contained only bacteria. These samples were annotated at the pixel level.

[0133] The performance of transformer models of varying sizes and that of a convolutional neural network model are compared. The transformer used is of the Segformer type (b0, b1, b2, b3, and b4 – b0 requiring the least computing power and b4 the most), while the convolutional neural networks used are DeeplabV3+ coupled with Resnet 101. In the latter case, the DeeplabV3+ neural network is used as the "backbone" to extract a feature set, which is then processed by the Resnet neural network used as the "head" to determine the TVC value. The backbone transforms the image into features, which are then processed by the head. In typical applications, this would involve classification, regression, object detection, or, in the example described here, semantic segmentation (one class per pixel).

[0134] It is also possible to combine a convolutional or recurrent neural network with a transformer. In this case, the transformer will always be used as the "head", while the other neural network will be used as the "backbone".

[0135] The models are powered by an Nvidia T4 GPU with 4 Intel Cascade Lake virtual CPUs and 16 GB of RAM.

[0136] This comparison comprises two parts:

[0137] 1. Intersection over Union (IoU) on the validation set (the 10 images mentioned above)

[0138] 2. End-to-end prediction of the total viable count (TVC) on 5 unused training samples, each composed of 5 images mixing wheat and bacteria (remembering that the model was trained only on pure samples). The resulting total viable count is then compared with that measured in the laboratory using Petri dishes in terms of absolute error and precision. Precision is defined as the number of samples for which the absolute error is less than 0.5.

[0139] The table below summarizes the results of the evaluations of the first part: Model Best I / O Iterations per minute TVE Accuracy Average TVC Error SegFormer b462.524774%0.46

[0140] The results show that the use of transformer models offers a clear advantage.

[0141] Lamontre presents the results of the evaluations in the second part. This figure specifically shows the results obtained using the SegFormer b4 transformer, as well as their comparison to Petri dish measurements.

Claims

Device for detecting bacteria in a sample, comprising an optical bench (4) including an illumination block (22) comprising an optical source (26) and a distribution optic (28) for illuminating a sample, and a measurement block (24) including a collection optic (34) and a measurement optic (36), the collection optic (34) being arranged to redirect measurement radiation from the sample (12) in response to illumination towards the measurement optic (36), characterized in that the optical source (26) is arranged to emit a first radiation with a wavelength substantially equal to 365nm, and a second radiation with wavelengths in a range between 375nm and 1000nm towards the distribution optic (28) in order to illuminate the sample (12), the distribution optic (28) including a magnification objective (30),the collection optics (34) further includes one or more splitters (38) arranged to separate the measurement radiation into at least a first beam and a second beam, the first beam having a portion of the measurement radiation having one or more wavelengths in a first range between 400nm and 414nm, the second beam having a portion of the measurement radiation having one or more wavelengths in a second range between 414nm and 490nm, the measurement optics (36) including at least a first optical sensor (40) for the first beam and a second optical sensor (42) for the second beam arranged to return an image in which each pixel is associated with an area of ​​the sample (12) illuminated by the magnifying lens (30) and with a measured light amplitude value respectively,the first optical sensor (40) being arranged in a position adapted to the focal plane of the first measurement beam, and the second optical sensor (40) being arranged in a position adapted to the focal plane of the second measurement beam, the device (2) being arranged to control a relative displacement of the sample (12) with respect to the illumination block (22) in order to illuminate successive portions of the sample (12) with the first beam and with the second beam, the illumination of a given portion by the first beam on the one hand and by the second beam on the other hand being carried out without overlap, and to obtain a set of multispectral images for each of the successive portions of the sample (12), the device (2) further comprising an analyzer (6) comprising a classifier (50) and a computer (54),the classifier (50) comprising a deep learning neural network arranged to receive the set of multispectral images associated with a portion of the sample (12) and to return an image (52) in which each pixel is associated with an element type identifier chosen from a list comprising at least two elements, one of which designates a bacterium, and the computer (54) being arranged to determine bacterium detection data (16) from the images returned by the deep learning neural network. Device according to claim 1, wherein the optical source (26) comprises two separate sources, one of which is arranged to emit the first radiation and the other is arranged to emit the second radiation, or a single source capable of emitting both the first and second radiation and which is arranged to emit the first and second radiation separately either by controlling its emission or by combining it with a controlled optical filter. Device according to claim 1 or 2, wherein said one or more separators (38) of the collection optics (34) are further arranged to separate the measurement radiation into a third beam distinct from the first beam and the second beam, this third beam having a portion of the measurement radiation having one or more wavelengths in a third range between 470nm and 1000nm, the measurement optics (36) comprising a third optical sensor (44) for measuring the third beam, the analyzer (6) being further arranged to receive the measurements from the third optical sensor (44) with the measurements from the first optical sensor (40) and the second optical sensor (42) to derive the data for the detection of bacteria (16). Device according to any one of the preceding claims, wherein the optical bench (4) further comprises an autofocus (32) disposed between the optical source (26) and the magnifying lens (30). Device according to claim 4, wherein the magnifying lens (30) and the autofocus (32) are included in the distribution optics (28) and in the collection optics (34). A device according to any one of the preceding claims, wherein the deep learning neural network of the classifier (50) is a convolutional neural network, a transformer, or a combination of both Device according to any one of the preceding claims, wherein the calculator (54) is arranged to determine a ratio between the area occupied by pixels associated with a bacteria identifier and a reference area, and to calculate bacteria detection data (16) from this ratio. Device according to any one of the preceding claims, wherein the classifier (50) is arranged to associate a pixel with an identifier of element type selected from a list comprising a bacterium, a matrix element, an air bubble, a membrane element or a foreign element. Device according to claim 7 in combination with claim 8, wherein the computer (54) is arranged to determine the reference surface from pixels whose identifiers are associated with a membrane element. A method for detecting bacteria in a sample comprising the following operations: a) determining one or more sets of multispectral images, each set of multispectral images of a given sample portion comprising images obtained by measuring, in a first wavelength range between 400nm and 414nm and in a second range between 414nm and 490nm, an autofluorescence beam emitted by the given sample portion when illuminated by radiation with a wavelength substantially equal to 365nm, and on the other hand a reflectance beam of the given sample portion when illuminated by radiation with wavelengths in a range between 375nm and 1000nm, the illuminations giving rise to the autofluorescence beam and the reflectance beam being non-overlapping,b) provide each set of multispectral images to a deep learning neural network to produce an image (52) in which each pixel is associated with an element type identifier chosen from a list comprising at least two elements, one of which designates a bacterium, etc.) compute bacteria detection data from at least some of the images from operation b). A method according to claim 10, wherein operation a) comprises determining a set of multispectral images for a set of non-overlapping sample portions whose union covers the entire surface of the sample. A method according to claim 10, wherein operation a) is performed for a portion of the sample, then operation b) is performed for that portion of the sample, and operations a) and b) are repeated until a measurement end condition including exceeding a threshold or determining a set of multispectral images for the whole sample is met. A method according to claim 10, wherein operation a) is performed for a chosen number of sample portions, then a required number of sample portions is determined by applying operation b) to the resulting multispectral image sets, and then operations a) and b) are performed with a stopping condition taking into account the required number of sample portions. A method according to claim 13, wherein the number of sample portions required is re-evaluated each time operations a) and b) are carried out, and wherein the stopping condition takes into account the re-evaluated number of sample portions required. A method according to any one of claims 10 to 14, wherein operation c) comprises determining a ratio between the area occupied by pixels associated with a bacteria identifier and a reference area, and calculating bacteria detection data (16) from this ratio.

Images