HU231356B1 - Combined bright field microscope and phase contrast microscope with image processing unit - Google Patents

Description

COMBINED BRIGHT-FIELD AND PHASE-CONTRAST MICROSCOPY SYSTEM AND SUPPLIED THEREOF

TECHNICAL AREA

IMAGE PROCESSING EQUIPMENT

SZTNH-100115553

The invention relates to a combined bright-field and phase-contrast microscopy system, as well as image processing equipment using it.

THE PRIOR ART

Current microscopy systems, for example, biological samples, especially in the field of body fluid analysis, use a measurement technique that is called is based on bright-field microscopy. Such a solution is also described in publication document WO/2013/104937. In a system built accordingly, preferably 15 grayscale (bit depth: 8), 1280x960 pixel digital images are taken of the sample that has been mixed and filled into a cuvette and then centrifuged. 2.2 pl native urine samples are examined. Before taking the pictures, each position is separately focused in order to get a sharp picture. The urine sediment images are evaluated with a complex algorithm based on neural networks, the only input of which is the recorded image. The images are so-called They are HPF (High Power Field), i.e. they offer a view as if looking through a microscope.

In image processing based on neural networks, it is advantageous if not only the recorded image but also additional information is made available for processing. According to the mentioned publication document, for example, not only the image, but also one or more transformed images created from the image with different functions or one or more versions of the image with different resolutions can be used as input in addition to or instead of the image. However, they all represent information based on the originally recorded image, and do not provide additional information for image processing.

Solutions are also known in which biological samples are imaged using the principle of phase contrast microscopy. The phase contrast-

-2 equipment consists of a special condenser and a so-called phase contrast objective, which differs from ordinary microscopy. The image obtained by the phase-contrast method is highly contrasted, and is therefore particularly well-suited for examining thinned cells stuck to the substrate and very thin (0.1-1 pm), disembodied sections in tissue culture. A disadvantage of the method is that a halo (halo) is created around structures with a strong contrast, which interferes with imaging.

In phase-contrast microscopy, an opaque plate is placed in front of the condenser lens, on which the light-blocking layer contains a ring-shaped transmissive area ("ring-diaphragm"). The beam exiting the condenser towards the preparation therefore passes through the object along a cone and is then imaged at the focal point of the objective. The phase-contrast objective has a plate containing a ring-shaped layer, which shifts the (comparative) light beam coming along the cylinder shell in phase by a quarter of the wavelength ("phase plate"), in a positive or negative direction. The phase-delayed refracted rays coming from the optically denser parts of the preparation to be examined do not pass through the phase ring and merge with the comparison rays in the image plane and interfere with them. This leads to the fact that the rays cancel out (or in the case of thicker structures strengthen) each other, so that the components of the object with different refractive index and thickness appear dark (or light) to different degrees. The purpose of the phase plate is to make the phase difference and thus the contrast relatively large even in the case of rays with a relatively small phase delay.

Phase-contrast imaging in itself is less suitable for image processing as described above, because the phase transitions are primarily displayed in the image, representing additional information, but due to the darkening of areas without phase transitions (light extinction), much of the visual information that appears in bright-field microscopy is lost.

According to the state of the art, there are solutions that ensure mechanical switchability between bright field imaging and phase contrast imaging, and thus could in principle ensure the joint use of both image types for image processing purposes. Such solutions are included, for example, in the patent documents GB 866,437, US 6,479,807 B1, US 5,731,894 and US 4,407,569. However, the known solutions are not suitable

- 3a for the purpose of the invention, for the examination of biological samples, because the image elements (e.g. bacilli) move during the relatively long mechanical switching. When switching, two optical elements must be moved; the ring diaphragm must be switched to a bright-field diaphragm, and the phase plate must also be removed from the 5 objective.

DESCRIPTION OF THE INVENTION

According to the invention, we realized that if we take not only one clear-field image, but also a phase-contrast microscopic image in 10 identical positions, we will have additional information about the sample and the image evaluation algorithm will be able to work from two input images instead of one. More input information results in more accurate output information, which results in a significant improvement in the reliability of image recognition, and can also ensure an appropriate choice between decision branches in the further course of image processing.

According to a further realization of the invention, the same phase-contrast lens and camera are used to capture both types of images. Bright-field and phase-contrast images are taken with separate lighting systems, which are connected to the same light path. The nature of the recorded image depends on which light source, or more precisely which diaphragm light source, is switched on. By switching the light sources in 20 specific sequences, two types of images can be recorded in every position in a few tenths of a second.

According to the invention, it is therefore a combined microscope that has both phase contrast and a compromise bright-field microscope solution, where the two different illuminations converge in Y~, but the part of the light path that reaches the sample and the 25 microscopic part is the same, the objective is the same, so the In the case of bright-field image recording, the phase ring is also present in the lens. As a result, the bright-field image is not a bright-field in the classic sense, but a compromised clear-field image, but our experiments proved that such a clear-field image is also completely suitable for the application according to the invention. The discovery of the invention 30 proved to be extremely advantageous, that for the switching between the two types of microscopy, the switching between the diaphragms was solved by switching the illumination units, while the phase plate is not removed from the objective,

-4, but we leave it in as a compromise solution. According to our experiments, the resulting slight deterioration in the quality of the image with a clear field of view is more than compensated by the additional information resulting from the phase-contrast image.

For example, the image size can remain at the original 1280x960. The aim of the invention is therefore to create a solution in which, in addition to the clear-field image, a phase-contrast microscopic image is taken in the same position, and with the additional information provided in this way, the image evaluation algorithm works more efficiently. The purpose of the invention is also to implement image processing in which, with the above-mentioned additional information, we enable a choice between processing branches after image recognition, and thereby direct the processing process to targeted decision modules. A further aim of the invention is to reduce and eliminate the aforementioned disadvantages of the state-of-the-art systems as far as possible.

The goal according to the invention was achieved with the microscopy system according to claim 1 and the image processing device according to claim 11. Advantageous embodiments of the invention are defined in the subclaims.

BRIEF DESCRIPTION OF THE DRAWINGS

Exemplary preferred embodiments of the invention are described below with drawings, where

Figure 1 is the schematic structure of a microscopy system according to the invention, a

Figures 2A and 2B are sets of bright-field and phase-contrast images taken from the same samples in pairs, shown one below the other, the

Figure 3 is a clear image of a digital urine sampler processed with the method according to the invention, the

Figure 4 is an enlargement of the selected part of the image according to Figure 3, that is

Figure 5. Schematic illustrating the creation of probability maps from clear-field and phase-contrast digital images, the

IIIIIIIIIIIIHI

SZTNH-100164455

Figure 6. A diagram of the groups representing assumedly present elements located on a probability map and the test areas assigned to them, the

Figure 7 is an example of the decision-making module in the device according to the invention! a sketch illustrating its input information, and a

Figure 8 is a diagram of an image processing device according to the invention.

METHODS OF IMPLEMENTING THE INVENTION

The present invention is partially based on the solution according to WO/2013/104937: the entirety of these 10 documents is hereby referred to and its teaching is considered part of the present application, with particular regard to the image processing examples and application methods described therein,

As an example shown in Figure 1! combined bright-field and phase-contrast microscopy system includes a bright-field 10 illumination unit, 15 phase-contrast 20 illumination units and a control unit 40 controlling the switching of the bright-field 10 illumination unit and the switching of the phase-contrast 20 illumination unit. The illumination unit 10 with a clear field of view and the illumination unit 20 with phase contrast have different optical axes 11 and 21, respectively. The microscopy system also includes optical elements 30 providing a common forward illumination 31 optical axis for the illumination coming from the illumination unit 20 with a bright field of view 10 and the illumination coming from the illumination unit 20 with phase contrast, as well as sample holders 33 and phase contrast objectives arranged one after the other along the common illumination 31 optical axis. , and the image recorder 50 was controlled to record images before and after each 25 switching between the two types of illumination. The latter control is also preferably implemented with the control unit 40. The control unit 40 can be of hardware or software design, or any suitable combination thereof. 34 samples to be tested can be placed on the sample holder 33.

The phase contrast objective 35 contains lenses 36 and 38 in a known manner, as well as 30 phase plates 37 arranged between the two.

-6 According to the invention, it is advantageous to use a 50 image recorder, for example a camera, controlled for image recordings with an interval of no more than 200 ms, preferably no more than 100 ms, for each switching. According to our experience, 200 ms is an upper limit which, if exceeded, may deteriorate the quality of image processing based on 5 double images due to the displacement of some elements detected in the biological sample, compared to the case based on a clear field of view image. Our experiments showed that the quality of the image processing according to the invention jumps in the case of an interval of 100 ms or less; this time limit provides images with essentially the same element location in the case of biological samples.

It is therefore a combined microscope that has phase contrast and a compromise bright-field microscope solution, where the two different illuminations converge in Y, but the part of the light path that hits the sample and the microscopic part are the same, the objective is the same, so in the bright-field case there is also a phase ring. As a result, the brightfield image is not bright-field in the classic sense, but rather a compromise, 15 an image that proved to be suitable for the given application. According to the invention, no mechanical step is required to switch between the two measurement modes; in this it differs from the switching solutions according to the state of the art.

As shown in Figure 1, the optical element 30 providing the common forward illumination optical axis 31 is preferably a prism, but for this purpose, of course, 20 translucent mirrors or other optical elements with a similar function can also be used.

The illuminations preferably have the same light color, and a phase contrast objective 35 tuned to this light color is used. The light color of the illuminations is preferably green, which is realized with green LED(s) 13, 23. The advantage of this is that commercially available 35 phase contrast lenses are usually optimized/tuned for green light, which is the middle of the 25 visible spectrum.

The phase-contrast illumination unit 20 is preferably also suitable for emitting an additional light-colored illumination, which can be used to obtain even more information for neural network-based image processing. The additional light color is preferably blue, which is realized by blue LED(s) 23; with the smaller wavelength blue light 30 better lateral resolution images can be obtained even if the phase contrast lens 35 is optimized for green.

The recorded images are preferably grayscale, i.e. the pixels record only the intensity values regardless of the color of the illumination.

The illumination unit 10 with a clear field of vision preferably contains 12 light sources and 15 light concentrating lenses, as well as a diaphragm 14 arranged between the two and adapted to the clear field of vision 5 . The phase-contrast lighting unit 20 also preferably contains a light source 22 and a light concentrating lens 25, and the annular diaphragm 24 arranged between the two.

As shown in Figure 1, a condenser lens or a condenser lens system 32 10 is arranged between the optical element 30 and the sample holder 33 providing the common forward illumination optical axis 31. An imaging lens or imaging lens system 39 is arranged between the phase contrast lens 35 and the image pickup 50; the task of the latter is to match the image to the given chip size, for example in the case of changing the camera. The phase contrast objective 35 is preferably a positive phase contrast objective, but of course a negative phase contrast objective can also be used for the purpose of the invention.

The most important elements of the new measurement technique implemented with the above system are the following: the combined bright-fieid - phase contrast microscopy technology, provides two images per field of view, i.e. additional information, which provides more accurate results in neural network-based image processing systems; The two out of 20 recorded images can be so-called "generate a composite image that contains both phase and intensity information. The two types of information can be applied to an RGB color image containing 3*1 bytes of data per pixel, for example, by assigning the intensity (25 images with a clear field of view) to one color and the phase information (image with phase contrast) to the other color. Assignment is possible with any suitable mathematical formula; for example, the phase can even be inversely proportional to the intensity of the given color. This can provide the user with a vivid and technology-specific visual experience. According to the invention, instead of the two separate images, this composite image carrying additional information 30 can also be input to the input of the image processing device, and in the context of the present application, the two separate images and the composite image formed from them are understood to be mutually equivalent and usable.

Figures 2A and 8A and 2B each show a pair of images taken from the same samples, consisting of 100 images with a bright field of view and 101 images with phase contrast. Switching between measurement modes can be done quickly with manual or even automatic software control, simply by switching the illuminations on/off one after the other and taking a recording of each case (the order does not matter, but it must be constant). In this way, the same arrangement of the elements of the sample can be ensured for both image recordings.

Two types of images taken from the same field of view are very beneficial for the evaluation of the images, as they contain additional information compared to cases with only a clear field of view or only 10 phase contrast. We verified this by teaching neural network structure and evaluating the final results. During this, we trained a network with two image inputs such that both images have the same bright field of view, both images have the same phase contrast, and also such that one image has a bright field of view and the other has a phase contrast image. In the latter case, the teaching proved to be faster 15 and the end result was better (with a higher recognition rate and a lower error rate). The structure of the neural network was completely identical in both cases, so the network and the number of samples were also the same. All this proved the applicability and advantages of the basic idea of the invention.

Preferably, no more time should elapse between the two types of recordings taken from the same field of view, 20 to 200 ms, even more preferably 100 ms. This means that with this technology, it is important to quickly switch between the two types of lighting, because it is important that the field of view does not change significantly in the time between taking pictures. It is necessary to make an effort because, for example, urine sediment-forming substances move, especially bacteria, which are small, but trichomonas 25 and other particles can also change their location.

Figure 3 is an example! you can see 100 pictures that were taken of a urine sample in connection with urinalysis. A part of the 100 images marked in the figure is shown enlarged in figure 4. Figure 4 shows the visual representation of a number of 116 elements to be categorized, i.e. different particles and image elements.

According to the invention, the element shown in the image 100 means the visual appearance of all objects that can be recognized and categorized. The following advantageous embodiment of the invention is for urinalysis, i.e. a digital image of a urine sample

-9 is described. Objects and elements that can be categorized in this way in a photograph taken by a watershed surveyor can be, for example, the following:

- bacteria (BAC),

- squamous epithelial cell (EPI, Squamous Epithelial Cell),

- small round epithelial cell (NEC, NonSquamous Epithelial Cell),

- red blood cell (RBC, Red Blood Cell),

- white blood cell (WBC), • background (BKGND, background) and the edge of the particle.

Of course, in addition to the above, further categorizable elements and objects can also be discovered in the photographs of urine samples. Together with the background, a total of typically 10 - 30 element categories can be set up.

As can be seen from the above, it is particularly advantageous if the background, which usually makes up the largest part of the digital photograph, i.e. the areas where other elements do not appear, are included in the analysis as separately categorized images. With the help of the comprehensive analysis according to the invention, the background can be separated from the further elements to be categorized in a more efficient way than the known solutions.

In the clear field image 100 and phase contrast image 101 shown in Figure 5, the elements to be categorized 20 carry different visual information. Based on the visual information perceptible in the images 100, 101, we generate probability maps 111 belonging to each predefined element category. Images 100, 101 can be submitted separately; in this case, an input number corresponding to the number of pixels of two images is required. If the images 100, 101 are entered in the form of a combined composite 25 images, only an input number corresponding to the number of pixels of one image is required. The individual probability maps 111 show the distribution of the probability of the presence of the element of the given category.

The probability maps 111i n can be produced in the same pixel size as the images 100, 101, but preferably the probability maps 111 vn are smaller. In a preferred embodiment 30, we use images 100, 101 with a resolution of 1280x960 (or composite image) and correspondingly W-un probability maps with a resolution of 160x120. The 111un probability for producing maps is preferably neural

- We use 10 networks. The neural network can be trained in any suitable manner, for example as described in WO/2013/104938. In order to create probability maps 111 u, where appropriate, we use not only the images 100, 101 (or composite image), but also one or more transformed images created from the images with different functions or one or more versions of the images 100, 101 with different resolutions.

The module that creates probability maps 111 i is called a recognition module (RM) according to the invention, which recognition module therefore contains inputs that receive the pixels of the images 100, 101 (or the composite image), as well as outputs that 11Η<· provides probability values for probability maps.

As the next step, we examine by category whether the images 100, 101 contain elements that are supposed to be present. As shown in Figure 6, the examination of the assumedly present elements is preferably carried out as 111 probability maps. During the test, we look for contiguous groups 112 of probability values above a predetermined threshold color, and then, based on the extent of the 112 groups and/or the size of the probability values in the 112 groups, we determine whether the group 112 corresponds to an element assumed to be present.

After determining the assumedly present elements, 113 test areas are designated. which is ideally positioned in the middle of the 112 groups corresponding to the supposedly present element. The test area 113 ideally consists of 5x5 probability values on the probability map 111; as shown in the figure, we define such an examination area 113 at the center, center of gravity, of each group 112.

According to the invention, it is. in connection with some test areas 113, we take into account at least one additional probability map 111 to make the decision regarding the presence of the element belonging to them, preferably also the probability values in the test area 113 positioned identically to the test area 113 above. As shown in Figure 7, we preferably include every 111;.. _t examination area positioned identically on the probability map 113in in making the decision. On these test areas 113?, in the preferred embodiment shown, there are 5x5 114 probability values, which represent the probability of the presence of elements belonging to each category in the given area, projected onto the 0-100 value scale.

According to the invention, the analysis of the 113t-n test areas is carried out with a device called a decision module (DM). which contains a neural network with inputs assigned to probability values 114 for the pixels in the 113 examination areas. The decision! module outputs preferably provide the probability of the elements falling into each category for a given test area.

With the joint analysis of the 113nn examination areas, object recognition and categorization can be performed with greater certainty. In this way, the probability values of mutually exclusive categories can also be taken into account, and the related information in the images 100, 101 can also be used to make the right decisions.

In the extremely advantageous embodiment shown in Figure 7, for the analysis of the decision module, not only the probability values 114 of the test areas 113i are used, but the statistical data 115 related to the elements are also taken into account. These can be, for example, statistical data of the type 115 described in document WO/2013/104937.

Figure 8 shows a diagram of an image processing device according to the invention. The image processing device is used to automatically categorize the elements in the images 100, 101 of the biological sample, and includes

- a microscopy system according to the invention, where the images 100, 101 of the biological sample are bright-field and phase-contrast images 100, 101 taken by the microscopy system, or a composite image produced from them,

- based on the visual information displayed on the images 100, 101 (or the composite image), the RM recognition module that produces THi n probability maps belonging to certain categories, which probability map 111 shows the distribution of the probability of the presence of the 116 elements of the given category,

- a BM branching module that selects one of the decision branches based on the joint processing of the 114^-0 probability maps, which decision branches

- each of 12 is connected to a DM decision module, which DM decision module contains a neural network trained with patterns characteristic of the given decision branch,

- the CM calling module that searches for the 5 elements assumed to be present in the images 100, 101 (or the composite image) and invokes a decision process with respect to each element assumed to be present, and

- the DM decision module, which performs the decision process per branch and provides information on the presence of the 116 elements belonging to each category based on the analysis of the 111 i.. _n probability maps, which DM decision module, in addition to the 111 probability maps belonging to the 116 element category, at least is designed taking into account an additional 111 probability maps.

Preferably, a CM calling module is connected to each of the decision branches to call the DM decision module on the same decision branch, but a solution 15 is also conceivable in which the output of the BM branch module is connected to a selector input of a CM calling module common to several DM decision modules.

The joint processing of the 111 i.n probability maps may include, for example, forming the sum or average of the 111 i n probability maps, determining the number of local peaks on them, or evaluating the magnitude of local peaks,

According to the idea of the invention, there is a decision point after the RM recognition module, where the evaluation can continue in several directions, and it is also possible to make "flag", i.e. one-bit, yes/no type decisions based on the image recognition results. The corresponding decision branch is 111:. _n probability maps are selected through a further joint processing of 25 with any suitable criteria; the specialist can develop this combined processing for the given samples, application with expedient considerations or experimentally. Depending on the image recognition results, directing the evaluation in different directions opens up great opportunities in the development of the evaluation. It doesn't matter, for example, whether different particles, 116 elements have to be recognized on a nice, clean, homogeneous 30 background (and teach neural network-based DM decision modules with nice background patterns), or with bacteria, mucus, crystal debris, or something on a background full of others. Squamous cells can range in size from a few IQ pm to typically 100-150 pm, and their internal texture closely resembles a background densely packed with bacteria. Therefore, the state-of-the-art image processing resulted in a relatively larger number of defective squamous cell hits. However, if images with a beautiful background can be distinguished from images with a background full of bacteria at the image recognition level, and with a BM branching module we can direct the further image processing to the appropriate branches, the same quality or with better image recognition, we can reduce the number of incorrect squamous cell results by up to 50-60%. We can train more of the DM decision modules, on different teaching staffs, which for example differ in the background. Separation can also be decided for other reasons; for example, in case of detection of a sample with a characteristic profile, which contains particles with a characteristic composition, etc.

If we can prepare the evaluation separately for a special case and we don't have to handle all the cases with one global module that teaches everything, we can expect a better end result. Of course, this requires more training. The point is, therefore, that after the first, always run, general phase of the evaluation, there is a decision point where separate flag-type decisions can be made, and for the further evaluation phase, you can choose from among the alternative paths developed for different typical cases. Each path has a separate DM decision module and even different FDM final decision modules (Id. below) can belong to each decision branch. The FDM final decision module can also be common to different DM decision modules, there is no restriction on this, it is decided by expediency. DM decision modules belonging to different decision branches must of course be taught with different teaching and test files specific to the given branch.

An example of a flag is whether the image shows a reliable, automatic result or whether we recommend a visual check by the operator after the measurement (Review flag). You can also decide on a different type of flag (e.g. are there artifacts, bubbles, etc. in the image). The flags can be displayed to the user by the user software as a warning or error message, or they may just be logged in a log file (logging), depending on the nature of the information.

The RM recognition module also preferably contains a neural network, which is a neural network with inputs receiving the pixels of the images 100, 101 (or the composite image) and outputting the probability values of the probability maps 114

- It has 14 ~ outputs, which are 111m probability maps with a lower resolution than our 100, 101 images.

Preferably, with respect to each presumptively present 116 element, the associated probability map 111 contains the CM call module that calls the DM decision module for 5 positioned 113 test areas for the presumptively present 116 elements, and the DM decision module also includes the identically positioned test areas 113 of at least one additional probability map 111 is designed taking into account.

The DM decision module preferably contains a neural network that has inputs that receive values of the probability 114 in the test areas 113m according to the invitation, and outputs that output the probability of the presence of elements 116 in each category in the test areas 113m according to the invitation.

The individual DM decision modules are preferably connected to the FDM final decision-making modules that accept the 116 elements as being present only above a threshold 15 probability, which threshold probability is preferably defined separately for each category.

The images 100 and 101 (or the composite image) or, where appropriate, one or more of their transforms or one or more versions with different resolutions are fed into the RM recognition module. The RM recognition module creates the 1111 probability maps, on the basis of which the elements assumed to be present are determined per category, i.e. per 111 probability maps, and the 113 test areas are assigned to them. The calling module CM of the device according to the invention calls the decision-making module DM for each punching point 25 on each probability map 111. The input of the DM decision-making module is provided by the test areas 1131* on the probability maps 111;-* assigned to the puncture point. In order to determine the presence of the element belonging to the given 113 test areas, the DM decision module also takes into account the probability values in the other 113 test areas, and based on this, it provides information about the presence of the elements belonging to the 30 individual categories.

- 15 The neural network of the DM decision module preferably outputs the presence probability of the elements falling into each category. Based on these probabilities, an additional FDM final decision-making module can make the final decision.

The FDM final decision-making module works expediently by examining the probability of the presence of the elements 5 and accepting the elements as present only above a predetermined threshold probability. During our experiments, it was confirmed that this threshold probability should be calculated separately for each category! to be determined, since the elements to be recognized and categorized in the images 100, 101 typically have different probability values 10,

The essence of the reliability test, which can be introduced with the FDM final decision-making module, is that the number of hits decreases as the threshold probability value increases, but the number of incorrect hits also decreases monotonically along a different curve. For each category, in this case particle type, it is advisable to set a threshold of 15, for which the rate of correct recognition is sufficiently low with a sufficiently low error. In the case of particles that can be recognized with high confidence and for which it is important to avoid the occurrence of incorrect recognitions, it is worth setting the threshold high.

The threshold probabilities of the FDM final decision-making module can also be determined separately not only for each category, but also for each application, or for each sample within the application. It may happen that a sample, i.e. a series of images consisting of 100, 101 images, has similar visual characteristics, so that the probabilities of recognition and categorization 25 also develop in a similar way.

During image processing, the first step requiring the most calculations is running the RM recognition module. Then the DM decision module is run, and finally the F DM final decision module decides where to put a label and where not to put a label on the clear field image, from among the previously identified places. According to the invention, therefore, the RM recognition module and the DM decision module are expediently implemented as a neural network. According to the invention, all modules of the equipment

- 16 can be implemented expediently in the form of a computer program module, but parts or all of these modules can even be designed in hardware form.

Above, the invention was described for urinalysis, images 100, 101 (composite images) made for urine samples, but of course this does not limit the applicability of the invention to this technical field. The element recognition and categorization according to the invention can also be advantageously used in the additional applications mentioned in the introduction that require image recognition and categorization.

The invention is not limited to the advantageous embodiments described in detail above, but further variations and modifications are possible within the scope of protection defined by claims 10. For example, the invention is not only suitable for processing two-dimensional images, but also for the analysis of images produced by a three-dimensional imaging process. In this case, the probability maps are also three-dimensional maps, and the test areas are three-dimensional parts of these maps.

Claims (11)

1. Combined bright field and phase contrast microscopy system, which contains - a lighting unit with a clear field of view (10), - phase contrast lighting unit (20), as well as - a control unit (40) controlling the switching of the bright-field lighting unit (10) and the switching of the phase-contrast lighting unit (20), characterized by the fact that the bright-field lighting unit (10) and the phase-contrast lighting unit (20) have different optical axes ( 11, 21) and the microscopy system also contains - the optical element (30) providing a common forward illumination optical axis (31) for the illumination coming from the bright-field illumination unit (10) and the illumination coming from the phase-contrast illumination unit (20), and the optical element (30) arranged in sequence along the common illumination optical axis (31) - sample holder (33), - phase contrast lens (35) and - the image recorder (50) controlled to record images before and after each switching between the two types of illumination. 2. The system according to claim 1, characterized in that it has an image recorder (50) controlled for image recordings with an interval of no more than 200 ms, preferably no more than 100 ms for each switching. 3. The system according to claim 1 or 2, characterized in that the optical element (30) providing the common forward illumination optical axis (31) is a prism or semitransparent mirror. Ms N N N 4. The 1-3. A system according to any one of the claims, characterized in that the illuminations have the same light color, and there is a phase contrast objective (35) tuned to this light color. 5. The system according to claim 4, characterized in that the light color of the illuminations is green, which is implemented with green LED(s) (13, 23). 6. The system according to claim 4, characterized in that the phase-contrast illumination unit (20) is also capable of emitting an additional light-colored illumination. 7. The system according to claim 6, characterized in that the additional light color is blue, which is implemented with blue LED(s) (23). 8. The 1-7. A system according to any one of claims, characterized in that - the bright field lighting unit (10) contains a light source (12) and a light concentrating lens (15), as well as the diaphragm (14) arranged between the two, adapted to the bright field of view, and - the phase contrast lighting unit (20) contains a light source (22) and a light concentrating lens (25), as well as the annular diaphragm (24) arranged between the two. 9. The 1-8. A system according to any one of the claims, characterized in that a condenser lens or a condenser lens system (32) is arranged between the optical element (30) providing the common forward illumination optical axis (31) and the sample holder (33). 10. The 1-9. A system according to any one of the claims, characterized in that an imaging lens (39) or an imaging lens system is arranged between the phase contrast objective (35) and the image pickup (50). Ms N N N 11. Image processing equipment, characterized in that 1-10. Contains a microscopy system according to any one of the claims.