METHOD FOR THE ANALYSIS OF A BIOLOGICAL SAMPLE USING ARTICLE MASK

DE602021021540T2Active Publication Date: 2024-11-06BIOMERIEUX SA
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
DE602021021540
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Priority Date
2020-09-25
Filing Date
2021-09-23
Publication Date
2024-11-06
Estimated Expiration
2041-09-23

AI Technical Summary

Technical Problem

Current methods for analyzing biological samples by imaging struggle to accurately determine biomass parameters due to artifacts such as air bubbles and foreign elements, which affect optical measurements and limit sensitivity, especially in holographic images.

Method used

A method using a holographic imaging system to acquire images, determine an image mask by comparing pixel intensity values with masking thresholds, and associate active or inactive states with pixels to exclude artifacts, allowing for the calculation of biomass parameters from active pixels only.

Benefits of technology

This approach significantly improves the analysis quality by excluding artifacts, enhancing the robustness of measurements and enabling more accurate determination of biomass parameters, even in the presence of foreign elements like air bubbles.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader
Need to check novelty before this filing date? Find Prior Art

Description

Technical field

[0001] The present invention relates to the field of analysis of biological samples by imaging, and more particularly relates to the masking of artifacts which may be present on an image, when determining a biomass parameter representative of the quantitative spatial distribution of biological agents in a biological sample for the purposes of analysis of said biological sample. Technological background

[0002] The analysis of biological samples by imaging uses an optical analysis instrument into which the biological samples to be analyzed are introduced. A biological sample consists of a suspension of biological agents. Biological agents are, for example, microorganisms (bacteria, yeasts, molds, etc.). The analysis of a biological agent in the biological sample may include the identification of said biological agent or the determination of a characteristic of this biological agent, such as, for example, the minimum inhibitory concentration of an antibiotic that would be effective against said biological agent.

[0003] The biological sample, called inoculum in its initial state, is placed in a receptacle, or well, at least partially transparent through which the analysis instrument can carry out measurements of optical properties of the biological sample. The well contains a nutrient medium and also one or more reagents, such as an enzymatic substrate or antibiotics, intended to allow the analysis of the biological agents present in the biological sample by their possible interactions with them. Generally, a plurality of wells are provided to receive each of the inoculum, each of the wells being provided with different reagents or the same reagent at different concentrations. Depending on the nature of the biological agents present in the inoculum, these react with certain reagents, and not with other reagents, or at certain concentrations and not at others.For example, in an antibiogram to test antibiotic sensitivities, the reagents consist of different antibiotics at different concentrations, and the biological agents will multiply in the wells containing the antibiotics to which they are not sensitive or in which the antibiotic concentration is insufficient, or on the contrary will see their development more or less hindered in the wells containing the antibiotics to which they are sensitive at sufficient concentrations.

[0004] These differences in interactions between biological agents and reagents therefore result in different changes in biomass in the wells. Biomass, that is, the quantity of biological matter present in each well, directly influences the optical properties of the biological sample present in each well, since the biological agents themselves have optical properties that are different from the solution in which they are suspended.

[0005] In particular, the transmittance of the biological sample is affected by the evolution of the concentration of biological agents. For this reason, methods for analyzing biological samples have been developed based on determining the evolution over time, during an incubation phase, of the overall transmittance (or absorbance, which is equivalent) of a well filled with the biological sample, in order to determine a turbidity measurement, typically expressed in McFarland (McF). This turbidity measurement is directly representative of the biomass of biological agents in the biological sample. To do this, an emitting diode illuminates the sample with a light beam of known intensity, and a point photodiode placed opposite the emitting diode relative to the sample makes it possible to determine the light intensity received after the light beam has passed through the biological sample.However, such a transmittance measurement has a fairly low sensitivity, so that it is not possible to measure turbidity below 0.05 McF, or even below 0.1 McF. In addition, biomass does not always allow the concentration of biological agents to be determined: if the volume of biological agents increases, for example by elongation for bacteria, the biomass increases for the same number of biological agents.

[0006] In fact, this is a global measurement, which does not take into account the spatial distribution of biological agents inside the analysis receptacle, nor any possible artifacts that may affect the measurements carried out. Such artifacts can have various origins. For example, the presence of air bubbles in the analysis receptacle or folds in the film closing the analysis receptacle can significantly alter the propagation of light waves, and therefore alter the optical measurements. Artifacts can also come from elements foreign to the biological sample, i.e. different from the biological agents and the saline solution, and present in the analysis receptacle, such as external dust, nutrient crystals used for bacterial growth, reagents not dissolved in the saline solution, etc. These foreign elements can absorb part of the light rays, and therefore also alter the transmittance measurements.

[0007] US Patent 5,566,249 A proposes a method for detecting bubbles in an image acquired by a microscope. A grayscale image is binomially filtered, then the grayscale image is subtracted from the filtered image. Edge detection is implemented to detect the edges of air bubbles, the system filling in the edges thus detected to map the presence of the bubbles. However, this system is not suitable for holographic images, but only for images acquired by a microscope in which the bubbles appear as edges. Furthermore, this method does not allow for the detection of artifacts other than air bubbles.

[0008] Document US 2018 / 052425 A1 discloses a method for analyzing a biological sample comprising acquiring a holographic image of the biological sample and determining from the acquired holographic image the pixels of the holographic image having a phase intensity value greater than at least one predetermined threshold so as to identify the spatial coordinates of artifacts and mask them from the optical illumination field using a sigmoid function, making it possible to recalculate the holographic image in the absence of the artifacts.

[0009] Document US 5,566,249 A discloses a method of analyzing a biological sample comprising acquiring an image of the biological sample and determining an image mask obtained by edge detection of particles identified in the image which is based on a contrast measurement using the comparison of a variation in intensity of the pixels associated with the boundary of the particle to determine whether the particle is an artifact to be masked or not.

[0010] WO 2019 / 135060 A1 discloses a method of analyzing a biological sample comprising acquiring a holographic image of the biological sample and using variations of statistical parameters in the distribution of gray levels in the holographic image to identify artifacts in the image. Presentation of the invention

[0011] The invention therefore aims to make it possible, during the analysis of a biological sample, to determine a value of a biomass parameter of the biological agents present in the biological sample, without alteration of the measurement by possible artifacts resulting from the presence of foreign elements such as air bubbles.

[0012] To this end, the invention proposes a method for analyzing a biological sample using an analysis instrument, the biological sample comprising biological agents and being arranged in an analysis receptacle in a field of view of a holographic imaging system, the method comprising: an acquisition of a holographic image of the biological sample at a measurement time, the holographic image associating an intensity value for each pixel of said holographic image, the intensity values ​​being included in an intensity range extending between a minimum intensity value and a maximum intensity value, characterized by a determination from the acquired holographic image, of an image mask associating an active or inactive state with each pixel of the holographic image as a function of the intensity values ​​of the pixels of the holographic image, so that an inactive state is associated with pixels of the holographic image corresponding to artifacts caused by elements present in the field of view other than biological agents,the image mask being determined so that the inactive state is associated at least with pixels of a related set of pixels having intensity values ​​in a masking range constituting a subset of the intensity range, said masking range being delimited by at least one masking threshold, and the determination of the image mask comprises the comparison of intensity values ​​of the pixels of the holographic image with the at least one masking threshold and the association with each pixel of a first mask value translating the inactive state of said pixel or of a second mask value translating the active state of said pixel according to the result of the comparison, a determination of a value of at least one biomass parameter representative of the quantitative spatial distribution of biological agents in the field of view, from only the pixels of the holographic image having an active state defined by the image mask,and providing said biomass parameter value among the analysis results.

[0013] Thanks to the image mask, pixels in the holographic image corresponding to artifacts caused by elements present in the field of view other than biological agents are not taken into account in the subsequent analysis of the holographic image, which significantly improves the quality of this analysis and makes it more robust to foreign elements such as air bubbles in the analysis receptacle.

[0014] The invention is advantageously supplemented by the following various characteristics taken alone or according to their various possible combinations: the intensity values ​​are within an intensity range extending between a minimum intensity value and a maximum intensity value, and a low masking threshold corresponds to a value between 5% and 25% of the extent of the intensity range, and preferably between 10% and 20% of the extent of the intensity range; the intensity values ​​are within an intensity range extending between a minimum intensity value and a maximum intensity value, and a high masking threshold corresponds to a value between 75% and 95% of the extent of the intensity range, and preferably between 80% and 90% of the extent of the intensity range; the state of a pixel of the mask is determined by the intensity value of said pixel with respect to at least one masking threshold;the determination of the image mask takes into account size or shape criteria of related sets of pixels having intensity values ​​in the masking range to assign an active or inactive state to the pixels of said sets of pixels and / or takes into account size or shape criteria of related sets of pixels having similar active or inactive states to modify an active or inactive state to the pixels of said sets of pixels; the determination of the image mask comprises the implementation of a mathematical morphology, and in particular a morphological dilation is applied to the image mask, followed by a morphological erosion; the determination of the image mask comprises the use of a neural network previously configured to associate an inactive state with pixels of the holographic image corresponding to artifacts caused by elements present in the field of view other than biological agents;the value of the biomass parameter is derived from a number of biological agents appearing in the active pixels of the holographic image; determining the value of the biomass parameter comprises determining, for each of a plurality of active pixel areas of the holographic image, whether biological agents are present in said active pixel area; whether biological agents are present in an active pixel area is determined by comparing a grayscale value of the area with a threshold, or by comparing a pattern of the area with reference patterns in a database; the image mask associates an active state with at least 25% of the pixels of the holographic image, and preferably with at least 35% of the pixels of the holographic image. ;

[0015] The invention also relates to an analysis instrument comprising a holographic system with a field of view configured to acquire a holographic image and data processing means, the analysis instrument being configured to receive a biological sample in an analysis receptacle in the field of view of the holographic system and to implement the steps of the analysis method according to the invention. Presentation of figures

[0016] Other characteristics, aims and advantages of the invention will emerge from the following description, which is purely illustrative and non-limiting, and which must be read in conjunction with the appended drawings in which: there figure 1 shows an example of an analysis card comprising a plurality of receptacles in the form of wells which can be used for the placement of a biological sample to be analyzed, according to a possible embodiment of the invention; the figure 2schematically shows an example of a holographic imaging system that can be used in an analytical instrument according to a possible embodiment of the invention; figure 3 is a diagram illustrating steps of the analysis method according to a possible embodiment of the invention; figure 4 is an example of a holographic image in which bubbles present in the analysis receptacle appear; the Figure 5 is an example of an image mask determined from the holographic image of the figure 4 ; there figure 6 shows the result of applying a morphological dilation followed by a morphological erosion applied to the image mask of the Figure 5 ; there figure 7 shows the result of applying the image mask of the figure 6 to the holographic image of the figure 4 ; there figure 8 shows a detail of the figure 7to illustrate a set of active pixels in which biological agents are searched. Detailed description

[0017] The method of analyzing a biological sample is carried out using an analysis instrument comprising a holographic imaging system with a field of view, the analysis instrument being configured to receive a biological sample in an analysis receptacle within the field of view of the holographic imaging system.

[0018] There figure 1shows an example of an analysis card 1 comprising a plurality of analysis receptacles 2 in the form of wells that can be used for placing a biological sample to be analyzed. The analysis receptacles 2 are here organized according to a two-dimensional network on a plane, each receptacle 2 being associated with different analysis conditions, typically by means of different reagents present in the analysis receptacles 2. For example, in the context of an antibiogram for testing antibiotic sensitivities, the reagents consist of different antibiotics at different concentrations. The use of an analysis card 1 is not required, but such an analysis card 1 makes it possible to carry out a plurality of tests in a standardized manner during the same analysis period.

[0019] Each analysis receptacle 2 is at least partially transparent to at least one wavelength of light, visible or not, and preferably is at least partially transparent for the visible spectrum. This transparency allows the analysis of the biological sample contained therein by optical means such as the holographic imaging system. Preferably, an analysis receptacle 2 has at least two opposite transparent faces, so as to have a transparent axis for the propagation of light. These two opposite transparent faces are for example less than 10 mm apart, and preferably less than 5 mm apart. Typically, the opposite transparent faces are transparent films defining the analysis receptacle 2. It is common for the reagents to be fixed on at least one of the transparent faces.The reagents can thus be introduced into the analysis receptacle 2 by placing them on the film intended to form a transparent face, before the latter is applied to the analysis card 1.

[0020] In order to allow the filling of the analysis receptacles 2, such an analysis card 1 may for example comprise a conduit 5 intended to be immersed in a volume 3 of inoculum 3 prepared in a tube 4. The inoculum is prepared by an operator who introduces biological agents, for example taken from a culture in a Petri dish by means of a rod or a swab, in suspension in a saline solution, with a dilution corresponding to a determined range of turbidity, for example between 0.5 and 0.63 McF for bacteria as biological agents or between 1.8 and 2.2 McF for yeasts as biological agents, the range depending on the type of analysis carried out and the measuring instrument. This preliminary suspension is then diluted further, for example by a factor of 20, or even 100, to analyze Gram- bacteria or by a factor of 10, or even 100, to analyze Gram+ bacteria.This subsequent dilution can be automated, and therefore carried out by the measuring instrument after inserting tube 4 into the analysis instrument. Of course, other specific turbidity ranges can be used, depending on the protocols used. The desired dilution can be obtained in one go, or as in the example above, in several goes.

[0021] One end of the conduit 5 is then immersed in the volume 3 of inoculum resulting from the preparation in the tube 4, and the whole is introduced into the analysis instrument. Of course, all or part of these preparation steps can be automated. The inoculum travels through the conduit 5, then via a fluid circulation circuit provided in the analysis card 1, is distributed between the analysis receptacles 5. This movement of the inoculum in the conduit 5 and the analysis card 1 can be caused by capillarity and / or by depressurization of the air present at the open end of the tube 4. For example, with depressurization, the air present in the analysis card 1, which is at atmospheric pressure, leaves the analysis card 1 through the tube 5 through the inoculum 3 and leaves its place to the inoculum 3 which thus rises through the tube 5 into the analysis card 1.Conversely, it is possible to apply air pressure to the inoculum via the open end of the tube 4 to cause the inoculum 3 to rise up the tube 5. The biological sample consisting of the inoculum is then placed in an analysis receptacle 2.

[0022] The analysis instrument comprises a holographic imaging system with a field of view configured to acquire a holographic image of this field of view. The acquisition of a holographic image allows a significant depth of field, and therefore very good sensitivity for detecting biological agents. When acquiring a holographic image, the holographic imaging system is placed facing an analysis receptacle 2. By way of non-limiting example, the figure 2schematically represents an in-line holographic imaging system 10 arranged so that the field of view 11 of said holographic imaging system 10 is contained in the volume of biological sample contained in an analysis receptacle 2. The analysis card 1, and therefore the analysis receptacles 2 which it comprises, is placed in an object plane of the holographic imaging system 10. The holographic imaging system 10 defines an imaging axis 16, simplified here by a straight line corresponding to the optical axis but which may consist of a set of successive straight lines defining the light path, depending on the configuration of the optical components of the holographic imaging system 10.

[0023] On one side of the analysis receptacle 2, here on the optical axis 16, there is a light source 14 configured to illuminate the analysis receptacle 2 in the field of view (or "field-of-view") of the holographic imaging system 10 by means of an illumination beam of sufficiently coherent light. The light source 14 can produce the illumination light, or simply be the termination of an optical fiber carrying this illumination light, possibly provided with a diaphragm or iris. The illumination beam has the conventional characteristics for holographic imaging, without any particular additional constraints. The illumination beam can thus be monochromatic (for example with a wavelength around 640-670 nm) or possibly be composed of several wavelengths, for example used one after the other.

[0024] On the other side of the analysis receptacle 2, here on the optical axis 16, there is an image sensor 12, which is a digital sensor such as for example a CMOS or CCD sensor. The image sensor 12 is placed on an image plane of the holographic imaging system 10, and is configured to acquire a hologram, that is to say a spatial distribution of intensity of the interferences caused by interactions between the inoculum placed in the field of view 11 and the illumination beam.

[0025] The holographic imaging system 10 is here provided with a set of optical members 18 arranged between the analysis receptacle 2 and the digital image sensor 12 such as for example a microscope objective 18a, and a tube lens 18b, in the example illustrated. An optical member such as the microscope objective 18 is however optional, the invention not being limited to holographic microscopy with a lens. The arrangement described here is of course a non-limiting example. Any holographic imaging system 10 can be used, with different optical members (with or without a microscope objective, etc.). Thus, as soon as a holographic imaging system 10 can acquire an image in which the interference patterns generated by the biological sample appear, this holographic imaging system is suitable for implementing the method.Preferably, however, the holographic imaging system 10 is configured to define a depth of field of at least 100 µm in depth in the direction of the optical axis 16 around each acquisition focal plane 20, and preferably at least 150 µm, and more preferably at least 250 µm. Typically, the analysis receptacle 2 comprises two opposite transparent faces organized along the optical axis 16, and the depth of field extends over at least 100 µm between the two opposite transparent faces of the analysis receptacle, and preferably over at least 150 µm, and more preferably over at least 250 µm. The field of view 11 is understood to be the space in which the presence of biological agents can be determined from a hologram imaging said field of view 11.

[0026] The measuring instrument also includes components for processing data, such as a processor, memory, communication buses, etc. Since these other components are only specific in the process they implement and the instructions they contain, they will not be detailed below.

[0027] There figure 3 is a diagram illustrating steps of the analysis method, which follow a prior placement (step S 1) of the biological sample in an analysis receptacle 2 in the field of view 11 of the holographic imaging system 10, detailed above.

[0028] Each measurement cycle includes an acquisition (step S02) of a holographic image of the biological sample at a measurement time.

[0029] When acquiring a holographic image, the holographic imaging system acquires a hologram, which has the advantage of providing a large depth of field, and therefore a high sensitivity for detecting biological agents in the biological sample. When acquiring a hologram, the light source 14 emits a reference illumination beam, which can be translated by a reference plane wave propagating in the Z direction along the imaging axis 16. The biological agents present in the field of view 11 inside the analysis receptacle 2, by their diffraction properties, scatter the incident reference light. The wave scattered by the biological agents and the reference wave interfere on the image sensor 12 to form the hologram.Since a digital image sensor is only sensitive to the intensity of the electromagnetic field, the hologram corresponds to the spatial distribution of intensity of the total field corresponding to the addition of the scattered wave and the reference wave, which is translated by a gray level value for each pixel. The holographic image used can be the hologram or can be an image reconstructed by backpropagation calculation from the hologram, using a propagation algorithm for example based on the Rayleigh Sommerfeld diffraction theory. Like the hologram, such a reconstructed image is defined by gray levels for each pixel.Using the hologram without reconstruction allows for high detection sensitivity, because each biological agent appears in the hologram surrounded by rings corresponding to the interference patterns caused by the presence of said biological agents, thus facilitating the detection of the presence of these biological agents. In addition, non-reconstruction saves time and computing resources. However, using a reconstructed image has other advantages, such as allowing the precise localization, possibly in three dimensions, of the biological agents appearing in the reconstructed image.

[0030] In all cases, the holographic image associates an intensity value for each pixel of said holographic image. These intensity values ​​are included in an intensity range extending between a minimum intensity value and a maximum intensity value. The intensity value is typically a gray level value. When the gray level is coded on 8 bits, the intensity values ​​can take integers between the minimum intensity value 0, which corresponds to black, and the maximum intensity value 255, which corresponds to white. Of course, other encodings of the intensity values ​​can be used.

[0031] There figure 4shows an example of a holographic image thus acquired, in this case an image reconstructed from a hologram. It is a 2400 × 2500 pixel image covering a field of view 11 extending over approximately 1.5 mm 2< with a depth of field of approximately 200 µm. The gray levels are coded on 8 bits (256 gray levels), 0 being black and 255 white. The biological sample here consists of a saline solution of Pseudomonas aeruginosaas biological agents. In this image of a biological sample, artifacts appear that cover a significant part of the field of view 11. First, air bubbles 20 can be seen, which appear black in the image. Due to the difference between the refractive index of the air inside the bubble and the refractive index of the saline solution and the rounded shape of the interface between the air and the water forming the wall of the bubble, part of the incident light rays is reflected, and another part of the rays is scattered, so that only the light rays incident at the center of the bubble and parallel to the optical axis 16 can pass through the bubble and reach the image sensor 12. The air bubbles 20, in the absence of light intensity, therefore appear essentially as black shapes, with, however, some light traces.Dark traces 21 caused by the non-flatness of the transparent films closing the analysis receptacle 2 in order to form the two opposite transparent faces can also be seen in the image. These may, for example, be reliefs such as folds. Artifacts such as air bubbles 20 or these dark traces 21 appear in the holographic image although they are not the biological agents to be imaged. It is therefore necessary to avoid these artifacts in order to correctly analyze the biological sample using the holographic image.

[0032] For this purpose, once this holographic image has been acquired, an image mask is determined (step S03) from the holographic image. The image mask associates an active or inactive state with each pixel of the holographic image. The image mask therefore comprises as many pixels as the holographic image. The active or inactive state of a pixel of the holographic image is determined as a function of the intensity values ​​of the pixels of the holographic image. The image mask is determined so that an inactive state is associated with pixels of the holographic image corresponding to artifacts caused by elements present in the field of view other than biological agents.

[0033] Since artifacts in the holographic image often appear as related sets of pixels having intensity values ​​in a masking range that is a subset of the intensity range, the inactive state is associated with at least pixels in a set of adjacent pixels having intensity values ​​in that masking range. That masking range is bounded by at least one masking threshold, and preferably with at least two masking thresholds: a low masking threshold and a high mask threshold, higher than the low masking threshold.

[0034] For example, the low masking threshold corresponds to a value between 5% and 25% of the extent of the intensity range, and preferably between 10% and 20% of the extent of the intensity range. The intensity range extends between the minimum intensity value and the maximum intensity value. Taking the example of gray levels coded on 8 bits (therefore on a range 0-255), the low masking threshold is for example between 13 and 64, and preferably between 25 and 50. The high masking threshold corresponds to a value between 75% and 95% of the extent of the intensity range, and preferably between 80% and 90% of the extent of the intensity range. Taking the example of gray levels coded on 8 bits (therefore on a range 0-255), the high masking threshold is for example between 191 and 243, and preferably between 204 and 230.Preferably, the low masking threshold and the high masking threshold are separated by at least 60% of the extent of the intensity range, preferably by at least 70%, and preferably by at least 80% of the extent of the intensity range.

[0035] The determination of the image mask may for example comprise a succession of non-linear operations such as thresholding or the application of a mathematical morphology. Thus, the state of a pixel of the mask may be determined by the intensity value of said pixel with respect to at least one masking threshold. For example, during the generation of the image mask, depending on the result of the comparison between each intensity value of each pixel of the holographic image with at least one masking threshold, either a first mask value translating the inactive state of a pixel is associated with said pixel, or a second mask value translating the active state of said pixel is associated with said pixel. The image mask associates the inactive state with the pixels having intensity values ​​in the masking range, typically close to at least one end of the intensity range, the proximity to one end of the intensity range being defined by the masking threshold.Thus, if the intensity value of a pixel is below the low masking threshold, an inactive state is associated with said pixel. Similarly, if the intensity value of a pixel is above the high masking threshold, an inactive state is associated with said pixel. Conversely, an active state is associated with a pixel whose intensity value is outside the masking range, that is to say between the low masking threshold and the high masking threshold, or simply above the low masking threshold if only this is used, or simply below the high masking threshold if only this is used to define the masking range.

[0036] The image mask, associating either an active or an inactive state with each pixel, can therefore be interpreted as defining a binary image. The Figure 5 shows an example of a graphical representation of the image mask determined from the holographic image of the figure 4, and in which pixels in the inactive state are represented as black, and pixels in the inactive state are represented as white. In this example, the image mask was generated using a single low masking threshold of 40, or about 15.7% of the 0-255 range of grayscale values. Pixels with grayscale values ​​less than or equal to 40 were assigned an inactive state, as black in the image mask image, and pixels with grayscale values ​​greater than 40 were assigned an active state, as white in the image mask image.

[0037] In black on the image mask we find the rounded shapes 22 of the air bubbles as well as black traces 23 corresponding to the dark traces 21 of the holographic image of the figure 4. However, white traces 24 are also observed inside the rounded shapes 22 of the air bubbles, which may have been caused by irregularities in the optical path of the light rays, for example caused by folds, or corresponding to the rays which may have passed through the air bubble. In addition, the black traces 23 may have discontinuities resulting from the failure to take into account certain pixels affected by the dark traces 21 in the image mask, a consequence inherent in the use of a masking threshold.

[0038] It is possible that the determination of the image mask includes the implementation of a mathematical morphology, in particular after a comparison with at least one masking threshold as described above. The mathematical morphology operations may in particular aim at locally modifying the shapes of the connected sets of pixels to regularize them. For example, in order to improve the continuity of artifacts in the image mask, a morphological dilation is preferably applied to the image mask, followed by a morphological erosion, i.e. the application of a morphological closure. The size of the structuring element used in these morphological operations depends on the level of morphological closure that is desired to be obtained, as well as on the characteristics of the image mask. For example, a low masking threshold with a high value will require a less pronounced morphological closure operation than with a low value.

[0039] There figure 6 shows the image mask resulting from the application of morphological dilation, followed by morphological erosion, with a structuring element represented by a size window. It can be seen that the white traces 24 are inside the rounded shapes 22 of the air bubbles. Indeed, the morphological closure causes the filling in black of the small white traces which are isolated in a black shape. In addition, the black traces 23 are a little more regular.

[0040] The determination of the image mask may take into account size or shape criteria of related sets of pixels having intensity values ​​in the masking range to assign an active or inactive state to the pixels of said sets of pixels, and / or size or shape criteria of related sets of pixels having similar active or inactive states to modify an active or inactive state to the pixels of said sets of pixels. For example, in addition to the comparison with a masking threshold and / or the application of a mathematical morphology, it is possible to require, in order to associate a pixel with the inactive state, that this pixel be part of a related set of pixels having intensity values ​​in the masking range, which represents a certain area. For example, artifacts such as those of the air bubbles 20 are represented by extended and related shapes, which are also of rounded regular shape.By imposing a minimum size for connected sets of pixels to go into the inactive state or a shape regularity criterion, one can for example selectively restrict the inactive state to the air bubble pixels 20.

[0041] In order to determine the image mask, it is also possible to use a neural network previously configured to associate an inactive state with pixels of the holographic image corresponding to artifacts caused by elements present in the field of view other than biological agents. More precisely, the neural network is previously trained on a training database formed of pairs of holographic images and corresponding image masks, the image masks being for example determined as explained above, or by another way so that an inactive state is associated with pixels of the holographic image corresponding to artifacts caused by elements present in the field of view other than biological agents.Then, in use, the neural network receives as input the holographic image for which it is desired to determine the image mask, and provides as output the corresponding image mask, which associates an inactive state with pixels of the holographic image corresponding to artifacts caused by elements present in the field of view other than biological agents. As recalled above, artifacts in the holographic image often appear in the form of related sets of pixels having intensity values ​​in the masking range. Therefore, the inactive state is also associated with at least pixels of a related set of pixels having intensity values ​​in a masking range.

[0042] Once the image mask is obtained, the image mask is applied (step S04) to the holographic image. The mask is applied so that the active and inactive states of the pixels of the image mask are used to restrict the subsequent processing carried out on the holographic image: only the pixels of the holographic image corresponding to an active state in the image mask are then used, hence the active and inactive designations for the states of a pixel. For example, the pixels of the holographic image corresponding to an inactive state in the image mask can be removed from the holographic image: the coordinates of these removed pixels are then unassigned. Other ways of applying the image mask can be envisaged, since this application results in the pixels of the holographic image corresponding to an inactive state in the image mask not being taken into account.

[0043] The method then comprises a determination (step S05) of a value of at least one biomass parameter representative of the quantitative spatial distribution of biological agents in the field of view 11, from only the pixels of the holographic image having an active state defined by the image mask. The pixels in the inactive state are not taken into account to determine the value of the biomass parameter. It is therefore necessary to keep enough pixels in the active state to allow the implementation of the method. Consequently, the image mask preferably associates an active state with at least 25% of the pixels of the holographic image, and preferably with at least 35% of the pixels of the holographic image. An activity threshold corresponding to a minimum number of pixels in the active state, or to a maximum number of pixels in the inactive state, can be established.If the image mask does not meet this activity threshold, for example with too few pixels in the active state, several solutions are possible.

[0044] Failure to meet the threshold may mean that the holographic image is composed primarily of artifacts and therefore does not allow for reliable analysis of the biological sample. It is then possible to trigger the emission of a non-compliance alert to the operator. The non-compliance alert for the biological sample may take several forms. For example, the analysis instrument may include an electroacoustic transducer, and the emission of the non-compliance alert includes the emission of sounds to an operator to warn them of the non-compliance. Similarly, the emission of the non-compliance alert may include the emission of a light signal to the operator.The analysis instrument typically comprises a human-machine interface with a display screen, and the issuance of the non-conformity alert may comprise the display on the screen of a message warning an operator of the non-conformity of the biological sample. Such an alert, in particular when it occurs during a measurement instant at the start of the measurement duration, may make it possible to alert on a problem preventing correct analysis of the biological sample, such as for example incorrect filling of the analysis receptacle 2, and may allow the operator to replace the biological sample without waiting for the end of the measurement duration to detect the problem, the measurement duration possibly extending over several hours of incubation. The non-conformity alert may also take the form of a simple mention of the detection of the non-conformity, preferably with the indication of the measurement instant, among the analysis results.Indeed, it is common for air bubbles to grow during the measurement period, or incubation period, and it can be advantageous to know the moment when the growth of the bubbles makes it impossible to obtain reliable measurements, both to avoid taking into account any measurements occurring after, but also to be able to still use the measurements made before. Other types of alerts can be provided. Failure by the image mask to respect the activity threshold can also indicate that the masking threshold used is not suitable. It may therefore be considered to resume the generation of an image mask with modified masking thresholds.

[0045] The biomass parameter may be of any type that can account for the quantitative spatial distribution of biological agents in the field of view 11. In particular, the value of the biomass parameter may be derived from a number of biological agents appearing in the active pixels of the holographic image. figure 7 is an example of the result of applying the image mask of the figure 6 on the holographic image of the figure 4 . We find the black rounded shapes 25 corresponding to the air bubbles, without the white traces 24, as well as the black traces 26. These pixels are considered inactive and are not taken into account. A region 26 of active pixels is framed, and its enlargement is illustrated by the figure 8 .In this figure, we can see the presence of biological agents 30 in this region 26 of active pixels. These appear clearly and distinctly as small circles. It is therefore possible to detect the presence of these biological agents, and even to quantify them. The biomass parameter is for example a number of biological agents in the field of view, or for example a proportion of the extent of the holographic image occupied by biological agents. It is for example possible to count the number of biological agents in the holographic image. When the holographic image is a hologram, the interference patterns typically appear as rings around a biological agent.A ring is a particularly easy shape to identify using a shape recognition algorithm, and it is therefore possible to analyze the holographic image in order to identify all the rings appearing there, corresponding to as many biological agents.

[0046] Rather than counting the number of biological agents, which can be time-consuming and resource-intensive for an unreliable result, determining the value of the biomass parameter may comprise determining, for each of a plurality of active pixel areas of the holographic image (typically several thousand areas), the presence or absence of biological agents in said active pixel area. The biomass parameter may then correspond to the number or proportion of active pixel areas in which biological agents are present, or more easily correspond to a number of areas where biological agents are absent, which is easier to highlight. The size of the area is chosen to be small enough to allow biological agents to be isolated without necessarily cutting off their representation. For example, the area may be between 5 and 20 times larger than the typical size of the biological agents sought.In one example, each active pixel area is 20x20 pixels. However, other sizes can be chosen, for example depending on the size of the biological agents whose presence is to be detected, or depending on the degree of precision sought or the data processing resources available, and the speed of data processing.

[0047] Determining whether or not a biological agent is present in an area of ​​the holographic image can, for example, be determined by comparing the average gray level (or light intensity) in an area with a gray level threshold. It is also possible to compare the pattern of the area with a database of reference patterns corresponding to a plurality of appearances of biological agents, and to identify the reference pattern having the greatest similarity with the area pattern. The characteristics associated with this reference pattern are considered to be those of the area pattern, which makes it possible, in addition to detecting the presence of biological agents in the area, to deduce additional characteristics, such as the individual growth of the biological agents, based on the characteristics of the appearances entered in the database.

[0048] The acquisition of a holographic image and the determination of a value of the biomass parameter, as well as the determination of an image mask, form a measurement cycle. The method therefore comprises a plurality of measurement cycles consisting of steps (S02, S03, S04, S05) implemented repeatedly for a plurality of measurement times of a measurement duration in order to obtain analysis results. These measurement cycles are typically repeated over a period ranging from one minute to 30 minutes, depending on the speed of the analysis instrument, the number of biological samples processed in parallel, and for example depending on the number of analysis receptacles 2 in an analysis card 1 or depending on the speed of the interactions between the biological agents and the reagents. The measurement duration extends over several hours, and typically more than 10 hours, resulting in several tens or even several hundreds of measurement times.The method therefore typically comprises more than 10 cycles, and preferably more than 20 measurement cycles during this measurement period. Each measurement cycle therefore makes it possible to obtain a value of a biomass parameter representative of the quantitative spatial distribution of biological agents in the field of view.

[0049] The analysis method comprises a final step of providing at least one value of the biomass parameter among the analysis results. Typically, the analysis results comprise the chronological organization of values ​​of the biomass parameter, in order to allow the temporal evolution of the quantitative spatial distribution of biological agents in the field of view 11 to be followed. Preferably, the provision of the value of the biomass parameter takes the form of a graphical representation allowing its communication to an operator, and its interpretation by the latter, such as for example a curve, an image or a table. The graphical representation can for example be displayed by a display screen, put into a format allowing its display, or can be transmitted to a printer to be printed.

[0050] The invention is not limited to the embodiment described and shown in the attached figures. Modifications remain possible, in particular from the point of view of the constitution of the various technical characteristics or by substitution of technical equivalents, without departing from the scope of protection of the invention.

Claims

1. A method for analyzing a biological sample by means of an analysis instrument, the biological sample comprising biological agents and being arranged in an analysis receptacle (2) in a field of view (11) of a holographic imaging system (10), the method comprising: - acquiring (S02) a holographic image of the biological sample at an instant of measurement, the holographic image associating an intensity value with each pixel of said holographic image, the intensity values lying within a range of intensities extending between a minimum intensity value and a maximum intensity value, characterized in that - determining (S03), from the holographic image acquired, an image mask which associates an active or inactive state with each pixel of the holographic image in accordance with the intensity values of the pixels of the holographic image, so that an inactive state is associated with pixels of the holographic image which correspond to artefacts caused by elements present in the field of view other than biological agents, the image mask being determined so that the inactive state is associated with at least some pixels of a connected set of pixels having intensity values within a masking range forming a subset of the intensity range, said masking range being delimited by at least one masking threshold, the determination of the image mask comprising the comparison of intensity values of the pixels of the holographic image with the at least one masking threshold and the association with each pixel of a first mask value indicating the inactive state of said pixel or a second mask value indicating the active state of said pixel in accordance with the result of the comparison, - determining (S05) a value of at least one biomass parameter representative of the quantitative spatial distribution of biological agents in the field of view (11), on the basis of only those pixels of the holographic image that exhibit an active state defined by the image mask, and - supplying (S06) said value of the biomass parameter among the analysis results.

2. The method of analysis as claimed in the preceding claim, wherein the masking range is delimited by at least a lower masking threshold corresponding to a value of between 5% and 25% of the extent of the intensity range, and preferably between 10% and 20% of the extent of the intensity range.

3. The method of analysis as claimed in any of claims 1 and 2, wherein the masking range is delimited by at least an upper masking threshold corresponding to a value of between 75% and 95% of the extent of the intensity range, and preferably between 80% and 90% of the extent of the intensity range.

4. The method of analysis as claimed in any of claims 1 to 3, wherein the determination of the image mask takes into account criteria of the size or shape of connected sets of pixels having intensity values in the masking range, for the purpose of assigning an active or inactive state to the pixels of said sets of pixels, and / or takes into account criteria of size or shape of connected sets of pixels having similar active or inactive states, for the purpose of modifying an active or inactive state for the pixels of said sets of pixels.

5. The method of analysis as claimed in any of the preceding claims, wherein the determination of the image mask comprises the implementation of mathematical morphology.

6. The method of analysis as claimed in any of the preceding claims, wherein the determination of the image mask comprises the use of a neural network preconfigured for associating an inactive state with pixels of the holographic image corresponding to artefacts caused by elements present in the field of view (11) other than biological agents.

7. The method of analysis as claimed in any of the preceding claims, wherein the value of the biomass parameter is derived from a number of biological agents appearing in the active pixels of the holographic image.

8. The method of analysis as claimed in any of the preceding claims, wherein the determination of the value of the biomass parameter comprises the determination, for each of a plurality of areas of active pixels of the holographic image, of the presence or absence of biological agents in said area of active pixels.

9. The method of analysis as claimed in the preceding claim, wherein the presence or absence of biological agents in an area of active pixels is determined by comparing a gray level value of the area with a threshold, or by comparing a pattern of the area with reference patterns in a database.

10. The method of analysis as claimed in any of the preceding claims, wherein the image mask associates an active state with at least 25% of the pixels of the holographic image, and preferably with at least 35% of the pixels of the holographic image.

11. An analysis instrument comprising a holographic system (10), with a field of view (11) configured for acquiring a holographic image, and data processing means, the analysis instrument being configured for receiving a biological sample in an analysis receptacle (2) in the field of view (11) of the holographic system (10) and for implementing the steps of the method of analysis as claimed in any of the preceding claims.