Process of counting weakly luminescent particles

The present invention relates to a method of counting faintly luminescent particles, in particular for counting cells and micro-organisms marked with a colored or fluorescent marker selectively marking the micro-organisms and cells to be counted.

The methods according to the prior art consist of acquiring an image in a high-g rossissement and to conduct an analysis of the image after data processing of pattern recognition, for example of contour recognition. Due to the high power required to obtain detailed images of objects to be counted and the very low density of these objects, it is necessary to explore many fields on a sample, and then perform a statistical analysis to deduce the density of objects in the sample.

The development of automated equipment shows that the natural tendency of the art is to continuously improve the individual counting of events, passing the visual counting of colonies formed by culture on a nutrient medium in a Petri box, to the combination of membrane filtration and automatic counting by image processing and automation of moving the microscope XY position and depth of field to increase the number of analyzed fields. Another approach, also on a individual counting principle of events and known by the name of flow cytometry is to individually count the passage of an event in a carrier stream.

The unit counting of events is only possible with the use of advanced optical means ensuring high discrimination of objects. usually used a video camera associated with a high-power microscope to obtain a very fine image, so covering a small part of the sample. These powerful optical means - and therefore fragile - are associated with efficient mechanical means of providing micro-movements in three dimensions of the sample holder plate with respect to the objective to allow the scanning of a number of fields statistically significant. It is therefore undeniable that improved reliability translates into a sophisticated optical and mechanical performance and results in a high cost and difficult to use outside the laboratory.

The present invention will radically against this process of improving the technical unit counting events, proposing an acquisition of an image of microorganisms and cells fixed on a marked under low magnification support by a camera photon accumulation. By low magnification will be understood magnification substantially less than that of a microscope normally used for unit counting devices of the prior art, for example a lower magnification 100.

According to the invention is carried out using a photon-accumulating camera for a short interval relative to the rate of change of said luminescent particles, to acquire a plurality of images I_i corresponding to different acquisition times, to proceed to the images I_i a thresholding so as to obtain binary images Ib_i, To carry out at least one dimensional filtering and counting the number N_i of objects belonging to a class of dimensions given in each of the binary images Ibi, the number of luminescent particles corresponding to the maximum value of N_i.

The accumulation camera delivers a signal whose dynamic is very high and that, combined with digital processing provides an accurate result notwithstanding the low resolution. In particular, the acquisition by a photon accumulation camera eliminates focus, and work simultaneously and uniformly on the field of observation.

The accumulation of photons and the implementation of the method according to the invention allow to obtain usable information on poor resolution images, and therefore covers a wide field.

Low magnification, for example 4, is used to count the photons, but in no case recognize bacteria. The use of low-resolution image acquisition means, under low magnification, which is against the teaching of processes and plants according to the prior art, provides essential advantages.

First, focusing problems are considerably simplified or eliminated. The methods of the prior art were to implement very complex means to ensure a perfect focus on the plane passing through the objects to be counted, and often require scan several successive focusing planes for a given observation field. Counting the time was therefore increased considerably.

Second, the low resolution image acquisition allows us to explore a wide range of sample or all of a sample. The objects to count generally of very low occurrence, this advantage is decisive. Indeed, the methods of the prior art are generally content to explore a number of statistically representative fields, not the entire sample. For phenomena with a large standard deviation due to artifacts (particle agglomeration, uneven distribution of items in the sample), this approach is certainly less satisfactory than the method of the invention allows a global acquisition on all of the sample.

Third, the digital processing is focused on binary images, and consists essentially of a class of binary images of objects according to their size. It uses simple algorithms and requiring no power or long calculation, contrary to the methods of the prior art, based on heavy pattern recognition methods and complex and intensive processing computation time .

Fourth, it is possible to use an excitation source of low-power fluorescence, eliminating the "burn" the markers. This advantage is important as avoiding degradation markers during the listing operation, it can keep the samples and subsequently carry out checks of the measure.

In a first embodiment, the dimensional filtering concerns the objects of the binary image occupying an area greater than the apparent nominal section of particles to count.

In a second variation, the dimensional filtering concerns the objects of the binary image occupying a surface area substantially equal to the apparent nominal section of particles to count.

According to a particular mode of implementation is determined more object classes for the counting of different types of particles.

The invention also relates to an installation for counting faintly luminescent particles, in particular counting cells process and micro-organisms marked with a colored or fluorescent marker selectively marking the micro-organisms and cells to be counted, comprising a source of excitation luminescent particles and a photon-accumulating camera associated with an optical low magnification own to form a scanned image of the sample with sufficient resolution so that the apparent image of the particles to be counted corresponds to several pixels. The accumulation camera is controlled by a microcomputer so as to produce a plurality of images I_i corresponding to different acquisition time, the signal provided by said camera being analyzed so as to allow the enumeration of particles of one or several types of data.

Preferably, the signal delivered by the camera storage and corresponding to each of the images I_i is processed by a microcomputer performing a thresholding so as to obtain binary images Ib_i, And at least one dimensional filtering and counting the number N_i of objects belonging to a class of dimensions given in each of the binary images Ib_i, The number of luminescent particles corresponding to the maximum value of N_i.

According to a preferred embodiment, the accumulating camera is associated with a lower optical magnification to 40.

According to a particular mode of implementation, the samples are carried by a strip passing sequentially under the optics associated with the photon accumulation camera.

According to a particular embodiment, the installation according to the invention comprises a sample preparation device formed by a strip passing sequentially, said strip being formed by the combination of a filtering layer and a film protection, said band passing successively in a clean way to deposit a fluid containing the particles to be counted, in a clean way to deliver a nominal amount of a labeling reagent, as a suction means, then under the counting means.

The invention will be better understood in the following, on reading the description referring to the accompanying drawings in which:<ul><li>1 shows a sectional view of a first embodiment of an installation according to the invention;</li><li>2 shows the curve of the change in the apparent size of the object as a function of acquisition time;</li><li>3 shows a diagram of the sequence of enumeration process;</li><li>4 shows the curve of the "brightness fluorescence" function.</li></ul>

1 shows a schematic view, in section, of an embodiment of a cell counting installation according to the invention. Counting is performed from a micro-porous membrane (1) having a matrix of filter areas (2, 3, 4) whose configuration is determined by the filtration apparatus employed, for example a filtration module marketed by the company bioMérieux under the name COBRA (trademark).

The sample preparation is carried out in known manner by filtration and coloring.

The installation comprises a housing (1) and rigid opaque by opening a hatch for introducing a micro-porous membrane (2) in an indexed flexible support carrying filtrates (3 to 5). This membrane (2) is placed on a metal support forming a drawer sliding between two guide rails (6, 7).

A source of ultraviolet radiation (8) associated with a condenser (9) is disposed below the tray carrying the membrane (2). A bandpass filter (11) permeable to UV radiation in the marker excitation wavelength is interposed between the UV source beam (8) and the membrane 42). A metallic reflector (12) returns part of the radiation to the filter (11).

The side opposite to the UV ramp, the installation comprises a photon accumulation sensor (13) associated with an optical block (14) having a magnification of the order of 10, preferably between 4 and 40.

The photon accumulation sensor comprises for example 768 * 562 pixels of 256 bits.

2 shows the evolution curve of the apparent size of the object on the binarized image, depending on the acquisition time.

For an acquisition time of less than a time T₁, Photon acquisition camera does not detect a photon or a photon number less than the threshold value determined for binarizing the image.

For an acquisition time between T₁ and T₂, The dimension of the object believes substantially proportionally with the acquisition time T. In this time range there is an appearance of the object cone whose position on the binarized image of the tip corresponds approximately to the center of severity of the luminescent object.

For an acquisition time between T₂ and T₃, The dimension of the object remains substantially constant or slightly croisante. The apparent size of the "spot" of light substantially corresponds to the apparent size of the luminescent object. subsequently be designated the area as "nominal cone".

This acquisition time range is relatively wide, and is the range of time for which the counting of binarized objects entering a pre defined class size will be representative of the number of luminescent objects in the sample.

For a time interval between acquisition of T₃ and T₄, The size of the apparent object increases rapidly, due to the light scattering phenomena emitted by the sample is taken. Found in this range a diffusion cone steep.

Beyond T acquisition time₄, The dimension of the stain increases very rapidly until it reaches a sensor saturation. This zone corresponds to the brightness of the background, which is detectable for very large acquisition times, and the phenomena of intrinsic noise in the acquisition camera.

The optimum acquisition duration will vary from one object to another in function of the brightness intensity. However, the acquisition time range between T₂ and T₃ is large enough to hold on at least one of the images I_i both the beginning of the nominal cone objects of very low light than the end of the nominal cone highly luminescent objects.

For this purpose, the procedure is as shown diagrammatically in FIG 3 in a "burst" acquisition for different time durations, e.g. croisantes. The increase in acquisition times is not critical. It may be a geometric progression so as to optimize the time required for obtaining a satisfactory image. The total duration of the burst of acquisitions will be brief in view of the evolution of the observed phenomena, including the weakening of the luminescence, or the variation of the excitation light.

The filter consists of a digital processing of the binarized image. all the objects is counting the binarized image whose dimension, that is to say the number of adjacent pixels, verifies a specific function.

This function is for example the "greater than N pixels" function where N pixels corresponds to the mean value of the apparent section of an object to be counted, or "between N-ε and N + ε pixels".

The result is a function represented schematically in Figure 3, corresponding to a Gaussian.

The treatment of the image delivered after each acquisition can be performed in real time or offline, each image Ii is then stored in memory for processing each of the images stored at the end of the acquisition burst.

4 shows the graph of the acquisition time required to reach the threshold value Vs implementation for the binarization of the acquired image according to the brightness of objects.

There is a first bell-shaped region and a divergent portion corresponding to the background brightness. The bell-shaped part moves in the direction of the diverging portion when the fluorescence objects is smaller (gray curve).

The invention according to a particular variant of a continuous monitoring system.

The samples were deposited on a film set formed by a layer of a carrier material, eg a plastic such as polyethylene, on which is deposited a second layer of a film forming filter. An upper protective film is punctured periodically to present disc-shaped filter areas.

The film is wound on a reel. It takes place sequentially so that a filtering zone passes successively under a means for depositing a sample, under one or more means for depositing reagents, then on a clean suction area to ensure the concentration of objects to count on the filter and then in an observation area. The support film is separated from the filtration film before passing into the suction area.

The invention is described in the above by way of example. The skilled person will be able to implement different variants without leaving the scope of the invention.